Dynamics of Crustal Magma Transfer, Storage and Differentiation 9781862392588, 1862392587

Citation preview

Magmas are subject to a series of processes that lead to their differentiation during transfer through, and storage within, the Earth’s crust. The depths and mechanisms of differentiation, the crustal contribution to magma generation through wall-rock assimilation, the rates and timescales of magma generation, transfer and storage, and how these link to the thermal state of the crust are subject to vivid debate and controversy. This volume presents a collection of research articles that provide a balanced overview of the diverse approaches available to elucidate these topics, and includes both theoretical models and case studies. By integrating petrological, geochemical and geophysical approaches, it offers new insights to the subject of magmatic processes operating within the Earth’s crust, and reveals important links between subsurface processes and volcanism.

Dynamics of Crustal Magma Transfer, Storage and Differentiation

The Geological Society of London Books Editorial Committee Chief Editor

BOB PANKHURST (UK) Society Books Editors

JOHN GREGORY (UK) JIM GRIFFITHS (UK) JOHN HOWE (UK) PHIL LEAT (UK) NICK ROBINS (UK) JONATHAN TURNER (UK) Society Books Advisors

MIKE BROWN (USA) ERIC BUFFETAUT (FRANCE ) JONATHAN CRAIG (ITALY )

RETO GIERE´ (GERMANY ) TOM MC CANN (GERMANY ) DOUG STEAD (CANADA ) RANDELL STEPHENSON (NETHERLANDS )

Geological Society books refereeing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society’s Books Editorial Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society Book Editors ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for Journal of the Geological Society. The referees’ forms and comments must be available to the Society’s Book Editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. More information about submitting a proposal and producing a book for the Society can be found on its web site: www.geolsoc.org.uk.

It is recommended that reference to all or part of this book should be made in one of the following ways: ANNEN , C. & ZELLMER , G. F. (eds) 2008. Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304. LEEMAN , W. P., ANNEN , C. & DUFEK , J. 2008. Snake River Plain – Yellowstone silicic volcanism: implications for magma genesis and magma fluxes. In: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 237 –261.

GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 304

Dynamics of Crustal Magma Transfer, Storage and Differentiation

EDITED BY

CATHERINE ANNEN University of Geneva, Switzerland and

GEORG F. ZELLMER Academia Sinica, Taipei, Taiwan

2008 Published by The Geological Society London

THE GEOLOGICAL SOCIETY The Geological Society of London (GSL) was founded in 1807. It is the oldest national geological society in the world and the largest in Europe. It was incorporated under Royal Charter in 1825 and is Registered Charity 210161. The Society is the UK national learned and professional society for geology with a worldwide Fellowship (FGS) of over 9000. The Society has the power to confer Chartered status on suitably qualified Fellows, and about 2000 of the Fellowship carry the title (CGeol). Chartered Geologists may also obtain the equivalent European title, European Geologist (EurGeol). One fifth of the Society’s fellowship resides outside the UK. To find out more about the Society, log on to www.geolsoc.org.uk. The Geological Society Publishing House (Bath, UK) produces the Society’s international journals and books, and acts as European distributor for selected publications of the American Association of Petroleum Geologists (AAPG), the Indonesian Petroleum Association (IPA), the Geological Society of America (GSA), the Society for Sedimentary Geology (SEPM) and the Geologists’ Association (GA). Joint marketing agreements ensure that GSL Fellows may purchase these societies’ publications at a discount. The Society’s online bookshop (accessible from www.geolsoc.org.uk) offers secure book purchasing with your credit or debit card. To find out about joining the Society and benefiting from substantial discounts on publications of GSL and other societies worldwide, consult www.geolsoc.org.uk, or contact the Fellowship Department at: The Geological Society, Burlington House, Piccadilly, London W1J 0BG: Tel. þ44 (0)20 7434 9944; Fax þ44 (0)20 7439 8975; E-mail: [email protected]. For information about the Society’s meetings, consult Events on www.geolsoc.org.uk. To find out more about the Society’s Corporate Affiliates Scheme, write to [email protected]. Published by The Geological Society from: The Geological Society Publishing House, Unit 7, Brassmill Enterprise Centre, Brassmill Lane, Bath BA1 3JN, UK (Orders: Tel þ44 (0)1225 445046, Fax þ44 (0)1225 442836) Online bookshop: www.geolsoc.org.uk/bookshop The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. # The Geological Society of London 2008. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0305-8719/08/$15.00. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 978-1-86239-258-8 Typeset by Techset Composition Ltd, Salisbury, UK Printed by Antony Rowe, Chippenham, UK Distributors North America For trade and institutional orders: The Geological Society, c/o AIDC, 82 Winter Sport Lane, Williston, VT 05495, USA Orders: Tel þ1 800-972-9892 Fax þ1 802-864-7626 E-mail: [email protected] For individual and corporate orders: AAPG Bookstore, PO Box 979, Tulsa, OK 74101-0979, USA Orders: Tel þ1 918-584-2555 Fax þ1 918-560-2652 E-mail: [email protected] Website: http://bookstore.aapg.org India Affiliated East-West Press Private Ltd, Marketing Division, G-1/16 Ansari Road, Darya Ganj, New Delhi 110 002, India Orders: Tel þ91 11 2327-9113/2326-4180 Fax þ91 11 2326-0538 E-mail: [email protected]

Preface The contributions in this book are based on a session convened by the editors at the 2006 AGU Fall Meeting in San Francisco, entitled ‘Dynamics of crustal magma transfer, storage and differentiation – integrating geochemical and geophysical constraints’. This session focused on magmatic processes within the Earth’s crust and therefore formed the bridge between mantle processes and volcanism. It integrated evidence from a variety of disciplines in order to make progress in resolving the following issues regarding magma dynamics and evolution: At what rates and through which mechanisms do magmas ascend through the crust? At what pressures and temperatures

are they stored on their way to the surface? Are magma reservoirs small or large, shallow or deep, ephemeral or long-lasting? Where and how does magmatic differentiation take place? What are the links between magmatic processes at depth and volcanic eruption at the surface? The session combined contributions from igneous and experimental petrology, geochronology, geochemistry, geophysics and a number of other disciplines that could shed light on these and related questions. CATHERINE ANNEN GEORG F. ZELLMER

Contents Preface

vii

Introduction ZELLMER , G. F. & ANNEN , C. An introduction to magma dynamics

1

Magma transfer: from mantle to surface ZELLMER , G. F. Some first-order observations on magma transfer from mantle wedge to upper crust at volcanic arcs

15

CIGOLINI , C., LAIOLO , M. & BERTOLINO , S. Probing Stromboli volcano from the mantle to paroxysmal eruptions

33

Dynamics of magma transport BUNGER , A. P. A rigorous tool for evaluating the importance of viscous dissipation in sill formation: it’s in the tip

71

WRIGHT , T. L. & KLEIN , F. W. Dynamics of magma supply to Kı¯lauea volcano, Hawai‘i: integrating seismic, geodetic and eruption data

83

MARTIN -DEL POZZO , A. L., CIFUENTES , G., GONZA´ LEZ , E., MARTINEZ , A. & MENDIOLA , F. Magnetic signatures associated with magma ascent and stagnation at Popocatepetl volcano, Mexico, during 2006

117

Magma reservoir dynamics JERRAM , D. A. & MARTIN , V. M. Understanding crystal populations and their significance through the magma plumbing system

133

BAN , M., SAGAWA , H., MIURA , K. & HIROTANI , S. Evidence for a short-lived stratified magma chamber: petrology of the Z-To5 tephra layer (c. 5.8 ka) at Zao volcano, NE Japan

149

Processes of silicic melt generation DOSSETO , A., TURNER , S. P., SANDIFORD , M. & DAVIDSON , J. Uranium-series isotope and thermal constraints on the rate and depth of silicic magma genesis

169

GRAY , W., GLAZNER , A. F., COLEMAN , D. S. & BARTLEY , J. M. Long-term geochemical variability of the Late Cretaceous Tuolumne Intrusive Suite, central Sierra Nevada, California

183

BURGESS , S. D. & MILLER , J. S. Construction, solidification and internal differentiation of a large felsic arc pluton: Cathedral Peak granodiorite, Sierra Nevada Batholith

203

LEEMAN , W. P., ANNEN , C. & DUFEK , J. Snake River Plain – Yellowstone silicic volcanism: implications for magma genesis and magma fluxes

235

STRAUB , S. M. Uniform processes of melt differentiation in the central Izu Bonin volcanic arc (NW Pacific)

261

Index

285

An introduction to magma dynamics GEORG F. ZELLMER1,2 & CATHERINE ANNEN3 1

Institute of Earth Sciences, Academia Sinica, 128 Academia Road Sec. 2, Nankang, Taipei 11529, Taiwan, ROC (e-mail: [email protected])

2

Lamont-Doherty Earth Observatory of Columbia University, 61 Route 9W, Palisades, NY 10964, USA 3

De´partement de Mine´ralogie, Universite´ de Gene`ve, Rue des Maraıˆchers 13, 1205 Gene`ve, Switzerland

Abstract: A variety of methods have been employed to decipher magmatic systems, including geophysical, petrological, textural and geochemical approaches, and these elucidate a large variety of characteristics of different plumbing systems and magmatic differentiation processes. A common theme to the papers presented in this book is the observation of transport of small volume magma batches with a relatively high frequency, as opposed to less frequent transport of larger magma volumes that would require storage in large crustal reservoirs for long periods of time. The implications of this observation are discussed in the context of a possible tectonic control on crustal magma dynamics.

This book addresses the rapidly developing fields of crustal magma transfer, storage and evolution. During both transfer trough and storage within the crust, magmas are subject to a series of processes that lead to their differentiation. Depths and mechanisms of differentiation, crustal contributions to magma generation through wall-rock assimilation, rates and timescales of magma generation, transfer and storage, and how these link to the thermal state of the crust, are subject to lively debate and controversy. This volume presents a collection of papers that provide a balanced overview of the diverse approaches available to elucidate these topics, and includes both theoretical models and case studies. By integrating petrological, geochemical and geophysical approaches, it provides the reader with new insights to the subject of magmatic processes operating within the Earth’s crust, and reveals important links between subsurface processes and volcanism. This volume is divided into four sections: ‘Magma transfer: from mantle to surface’ addresses the ascent and evolution of magmas from the zone of melt generation in the mantle to eruption at the surface, forming a backdrop for the detailed studies of distinct parts of magma plumbing systems addressed later. ‘Dynamics of magma transport’ focuses on theoretical and geophysical approaches to understanding magma movement through the crust. ‘Magma reservoir dynamics’ provides insights from petrographic and mineral chemical studies into the processes occurring in crustal magma chambers. Finally, ‘Processes of

silicic melt generation’ concludes the book with a dedicated section on the long-standing question of where and how magma differentiation may take place. In nature, these issues are of course intimately related, and some of the papers in this volume address more than one of these aspects. Therefore, the reader may obtain additional insights to a particular theme by referring to the other sections of the book. With the exception of two contributions (Leeman et al.; Wright & Klein), all case studies presented in this volume deal with subduction zone magmatism. The inferences made here on the dynamics of magma ascent, storage and differentiation are therefore biased towards this tectonic setting. It may be argued that subduction-related magmatic systems are likely to have very different petrogenetic characteristics than ocean ridge and intraplate volcanism. Firstly, the primary magmas are produced at different depths within the mantle, and have different temperatures and compositions, particularly with regard to their volatile contents. Secondly, as a result of these differences, their petrogenetic evolution within crustal magma systems will differ significantly. For example, crystallization of volatile-rich arc magmas may be triggered by rapid decompression-induced degassing during magma ascent through the crust, a process that is not readily applicable for ocean ridge and intraplate magmatic systems. Thirdly, tectonic controls on the geometry of the plumbing systems differ considerably. Ocean ridges are in extension, resulting in rapid magma ascent through dykes and movement

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 1–13. DOI: 10.1144/SP304.1 0305-8719/08/$15.00 # The Geological Society of London 2008.

2

G. F. ZELLMER & C. ANNEN

of newly created crust away from the heat source. In contrast, intraplate magmatism may favour magma storage and differentiation within the crust due to repeated sill intrusion and resulting progressive elevation of the geothermal gradient. Arcs, on the other hand, may be situated within extensional, transtensional or compressional regimes, potentially resulting in differences between the plumbing systems of different arcs (cf. Zellmer). Further work will be required to gain a balanced understanding of the dynamics of magma plumbing systems within different tectonic settings.

approach is used by Cigolini et al. to elucidate the plumbing system of Stromboli volcano from upper mantle to surface. The data suggest that phenocrysts nucleate within a few days in a magma reservoir that extends vertically from 11 to 5.4 km below the summit of Stromboli. A new model is proposed where the magma chamber takes the shape of a vertically elongated ellipsoid that is penetrated by a feeder dyke sourced from over 30 km depth, i.e. in the mantle. According to this model, the instantaneous elastic rebound of the walls of the depressurizing subvolcanic reservoir explains the occurrence of intermittent paroxysmal eruptions at this volcano.

Magma transfer: from mantle to surface Studies of igneous processes often focus on those parts of the crustal plumbing system that are best elucidated by the samples or methods available. It is rare that systems have been studied in sufficient detail to inform the entire process from magma generation in the mantle to eruption at the surface. This book starts with two studies where the amount of data is sufficient to provide such insights, on one hand using global volcanological, geophysical and geochemical datasets to present a broad overview of magma transfer (Zellmer), on the other hand focussing on the single edifice of Stromboli Volcano to a gain detailed understanding of the petrogenetic processes occurring along the entire mantle-crust section beneath this volcano (Cigolini et al.). These papers address the dynamics of igneous processes operating at sites of ongoing volcanic activity, and therefore form a backdrop for the detailed studies of the distinct parts of magmatic plumbing systems focused on in the later sections of this book. From global correlations between eruptive style, surface heat flux and convergence rates of different volcanic arcs, Zellmer infers that the rate of melt production in the mantle wedge ultimately controls the dynamics of magma transfer through the crust, and thereby the chemical and physical properties of magmas and eruption products. It is shown that a deep crustal hot zone (Annen et al. 2006) does not buffer the effects of subduction velocity on melt production, and that the rate of magma generated in and released from the hot zone is proportional to the magma advected to the hot zone from the mantle wedge. Crystal size distributions, bubble content and magma rheology, petrology and chemistry are a number of parameters that – when studied in combination – may offer a very detailed picture of the processes operating within magma plumbing systems, and can be used to quantify pressures, temperatures and the rates of preeruptive crystallization and gas exsolution. Such a multi-faceted

Dynamics of magma transport Background The second section of this book deals with insights that can be gained from theoretical and geophysical research. Direct observations of eruptions along fractures, ground deformation and the distribution of seisms associated with magma intrusion (e.g. Pollard et al. 1983; Rubin & Pollard 1987; Peltier et al. 2005; Yamaoka et al. 2005; Aloisi et al. 2006; Mattia et al. 2007) provide evidence for the role of dykes in mafic magma transport (cf. Figs 1a & 2). In the case of andesitic volcanism, effusive or explosive, eruptions are more focused and the conduits are more cylindrical (cf. Fig. 1b). This may be explained through melting of host rocks (Quareni et al. 2001), and may also be related to a sharp increase in magma viscosity close to the surface due to decompression, degassing and crystallization. Rhyolitic magma, when associated with caldera formation, is transported to the surface through ring dykes. It has also been suggested that interconnected sills could transport magma through the crust (Marsh 2004; Cartwright & Hansen 2006; e.g. Fig. 3a). Interpretation of geochemical and geophysical data suggests that the plumbing system of many volcanoes may be a complex plexus of interconnected sills and dykes (Hildreth 1981; Lahr et al. 1994; Donoghue et al. 1995; La Delfa et al. 2001; Preston 2001; Dawson et al. 2004; Sanchez et al. 2004; cf. Fig. 1c). For granitic magma, diapirism was thought to be a common mechanism of magma transport and emplacement within the crust (e.g. Whitehead & Luther 1975; White & Chappell 1977; Pitcher 1979; Hildreth 1981; Marsh 1982; cf. Figs 1f & 3b). However, Hot-Stokes diapirism, characterized by ductile flow of host rock around the rising magma mass, is regarded as thermally and mechanically unrealistic within the mid to upper crust, and

INTRODUCTION

3

Fig. 1. Summary of possible crustal magma plumbing systems as inferred through studies in this volume. Magmas may have been processed within a lower crustal hot zone (Annen et al. 2006), although some mantle melts may be transferred instantly into and through the crustal section (cf. Straub). (a) Magma transport through dykes (cf. Martin-Del Pozzo et al.; Wright & Klein), at rates of centimetres to tens of kilometres per day. (b) Development of small vertical chambers (cf. Cigolini et al.; Dosseto et al.; Wright & Klein), with storage times between days and a few thousand years. (c) Magma transport through interconnected dykes and sills (cf. Bunger; Leeman et al.), with sill solidification timescales of the order of 10–100 years. (d) Development of small horizontal chambers, e.g. through repeated sill intrusion (cf. Ban et al.; Leeman et al.), with storage times between tens of years and thousands of years. (e) Development of plutons through repeated addition of small magma batches over long timescales of up to 105 –107 years (cf. Burgess & Miller; Gray et al.). Large magma chambers may form through rapid (a few hundred years, e.g. Michaut & Jaupart 2006) large-scale crustal melting when enough thermal energy has accumulated (cf. Leeman et al.). (f) Diapirism as a mechanism to emplace felsic magma reservoirs. For emplacement into the mid to upper crust, visco-elastic diapirism as suggested by Miller and Paterson (1999) would be required. Note that none of the studies in this volume find evidence for diapirism. Although the stalling of magma is not only controlled by the stress field, but also by the rheological properties and densities of crustal lithologies, the studies presented in this volume suggest that (a) and (b) may be favoured in extensional or transtensional settings, while (c)–(e) are more likely to occur in compressional tectonic regimes (see also Zellmer).

dyking as a magma transfer mechanism is favoured by a number of studies (Clemens & Mawer 1992; Clemens et al. 1997; Petford et al. 2000). Hutton et al. (1990) observed granite intrusion along

ductile extensional shear zones and noted that it was solving the room problem posed by pluton emplacement, making diapirism unnecessary. Conversely, Miller & Paterson (1999) introduced

4

G. F. ZELLMER & C. ANNEN

Fig. 2. Geophysical observations indicate subvolcanic magma transport through dykes. This example shows modelled sources obtained from dynamic inversion of Mount Etna tilt data and seismicity recorded during intrusion propagation, as taken from Aloisi et al., ‘Imaging composite dike propagation (Etna, 2002 case)’, Journal of Geophysical Research, 111, paper B06404, DOI:10.1029/2005JB003908, 21 June 2006. Copyright 2006 American Geophysical Union. Reproduced by permission of American Geophysical Union.

the term ‘visco-elastic diapir’ and argued that, if diapirism is defined as the upwelling of mobile material through or into overlying rocks (Van den Eeckhout et al. 1986), then it remains an important emplacement mechanism of felsic plutons. The following field observations are used to support the model of visco-elastic diapirism: downward movement of host rocks through multiple processes, including brittle deformation and stoping; involvement of multiple magma batches; and controls of regional deformation on pluton emplacement. Visco-elastic diapirism is called upon in a number of recent case studies of felsic plutons (e.g. Cabello et al. 2006; Farris et al. 2006; Zak & Paterson 2006). In the case of magma transport through fractures, crustal magma transfer can be very fast on the order of days to years (Clemens & Mawer 1992; Clemens et al. 1997; Petford 2003; Annen et al. 2006). Large excesses in 226Ra in mafic volcanic products indicate ascent from source to surface in the order of 1 ka or less (Turner et al. 2000; Zellmer et al. 2005). In exceptional cases, magma ascent rates through the crust can be extremely fast: 26 km per day have been estimated for

some Mexican andesites that carry hornblende – peridotite xenoliths, which reached the surface so rapidly that they were not affected by dissolution in their host magma during ascent from the mantle (Blatter & Carmichael 1998). In the case of felsic magma transport through visco-elastic diapirs, crustal magma transfer rates may be significantly slower, of the order of 1022 to 1 m year21 (Miller & Paterson 1999).

Contribution of this volume Independent of storage depth and composition of the magma, many magma bodies are thought to be established through repeated sill intrusions (Bridgwater et al. 1974; Benn et al. 1999; Cruden & McCaffrey 2001; de Saint-Blanquat et al. 2006; Pasquare & Tibaldi 2007). The work of Bunger provides insights into the parameters that govern sill propagation. It is shown that, during sill growth, fracture behaviour is strongly influenced by viscous flow within the near-tip region, and that the physics of viscous dissipation must therefore be taken into account when modelling sill growth.

INTRODUCTION

5

Fig. 3. (a) The cartoon depicts a ‘magmatic mush column’, where colour hotness (see HTML version) portrays magma temperature and small black squares depict large crystals. The mush column is characterized by a complex variety of local crystallization environments with different cooling rates, in which crystal inheritance from earlier crystallization episodes is common. Taken from Marsh, ‘A magmatic mush column Rosetta Stone: the McMurdo Dry Valleys of Antarctica’, EOS, 85, 497–508, 23 November 2004. Copyright 2004 American Geophysical Union. Modified by permission of American Geophysical Union. (b) In early models of lithospheric magmatism, intermediate composition magmas were thought to rise through the crust by diapiric mobilization before coalescing in shallow subvolcanic reservoirs. Idealized cartoon taken from Hildreth, ‘Gradients in silicic magma chambers: implications for lithospheric magmatism’, Journal of Geophysical Research, 86, 10153– 10192, 10 November 1981. Published 1981 American Geophysical Union. Modified by permission of American Geophysical Union.

Using seismic, geodetic and eruption data from Kilauea volcano, Wright & Klein evaluate the interplay between magma supply and spreading of associated rift zones that provide room for intrusions into the subvolcanic reservoirs. It is shown that the dynamics of magma supply affects spreading rate, intrusion frequency and degree of summit inflation, resulting in varying characteristics of intrusive and eruptive activity at this volcano. In a case study of the 2006 eruptions of Popocatepetl volcano, Martin-Del Pozzo et al. use varying magnetic anomalies to gain insights into magma ascent and lava dome extrusion. The magnetic signatures of these and associated processes are superimposed, but they can be distinguished on the basis of signal morphology, and correlated with data from other monitoring techniques. It is deduced from magnetic, seismic and petrological data that the plumbing system beneath Popocatepetl is essentially formed by dykes. The stagnation level is constituted from a series of interconnected dykes and does not involve a large magma chamber.

Magma reservoir dynamics Background The third section of this book focuses on petrographic, petrological and mineral chemical approaches to elucidate crustal magma reservoirs and plumbing systems (e.g. Fig. 1d). Magma reservoirs have been inferred to occur at a range of depths, from the base of the lower crust right up to near the surface beneath volcanic edifices. The temperatures and melt fractions within a magma reservoir depend on its volume and depth. If the melt fraction is below a critical value (40 –60%), the crystals form a rigid network (Vandermolen & Paterson 1979; Marsh 1981; Lejeune & Richet 1995). This mush is a porous medium where melt and gases may move independently through the crystal framework. In contrast, if the melt fraction is above the critical value, the crystals are suspended and the magma is a multiphase fluid able to convect. In a mush, the separation of the residual

6

G. F. ZELLMER & C. ANNEN

melt from the crystals is mostly by compaction and deformation (McKenzie 1984; Sparks & Huppert 1984; Petford 2003; Bachmann & Bergantz 2004). In a liquid magma this separation can be related to crystal sedimentation or floating, although thermal convection in the magma chamber can prevent crystal settling. In a convecting magma chamber, crystals and liquids can be separated by crystallization on the magma chamber walls and convection of the adjacent fluid (Sparks & Huppert 1984; Sparks et al. 1984). The shapes and sizes of magma reservoirs are not well known. Many plutons are roughly circular in outcrop with steep sides (Pitcher 1979), and magma chambers and plutons are often represented as spherical bodies. However, geophysical and structural studies show that some plutons are silllike, low aspect-ratio, tabular bodies (Bridgwater et al. 1974; Lefort 1981; Cruden 1998; Petford et al. 2000), while others are elongate intrusive bodies of a few kilometres vertical extent (e.g. Farris et al. 2006). The Socorro magma body, which can be seen on seismic profiles, is sill-like (Rinehart et al. 1979; Brocher 1981). Further, most igneous bodies may have been emplaced by the incremental assembly of discrete pulses (Coleman et al. 2004; Vigneresse 2004; de SaintBlanquat et al. 2006). Geochronological data indicate that the assembly of some plutons and batholiths lasted several millions years. For example, according to U – Pb geochronological data, the Tuolumne Intrusive Suite of the Sierra Nevada, California, was emplaced over 10 millions years (Coleman et al. 2004), and the Mount Stuart batholith and Tenpeak intrusion in the Northern Cascades were emplaced over 5.5 and 2.6 Ma, respectively (Matzel et al. 2006). Long emplacement timescales led Glazner et al. (2004) to question the relationship between plutons and large magma chambers: if plutons are emplaced slowly by amalgamation of discrete pulses of magma, the volume of molten rock during the construction of the pluton might not greatly exceed the volume of a single pulse. However, Matzel et al. (2006) argue that the construction of the Mount Stuart batholith was discontinuous and punctuated by high magma flux intervals that may have led to the formation of large magma reservoirs (cf. Fig. 1e). The existence of calderas, the distribution of vents around these calderas, the large volumes of ignimbrites and the common occurrence of zoning within these ignimbrites support the existence of large and shallow felsic magma chambers (Chapin & Elston 1979; Lipman 1984). In a recent paper, Lipman (2007) provides a series of arguments that support the link between plutons and large-volume ignimbrite volcanism associated with caldera formation, but stresses that large upper crustal magma reservoirs

may be short-lived, and that in some arcs the volcanic products may come directly from the mid or lower crust without involving shallow crustal magma chambers. The view that magma reservoirs shallower than 20 km are rare is supported by the absence of detected ground deformation associated with several eruptions in the Andes (Pritchard & Simons 2004), and the view that shallow reservoirs are short-lived is supported by the short residence time of crystals at magmatic temperatures recorded by trace element profiles in zones crystals (e.g. Zellmer et al. 1999, 2003; Costa et al. 2003; Morgan et al. 2004; Morgan & Blake 2005; Zellmer & Clavero 2006).

Contribution of this volume The variety of crystal populations that are found in volcanic products indicate that crustal magma transfer processes may in detail be complicated through uptake of xenocrysts, crystal recycling within the magmatic system of individual volcanoes, and crystal growth and resorption triggered by processes such as magma cooling, recharge, decompression and degassing. Jerram & Martin provide a review of how these processes may be deciphered and quantified, and outline avenues for future research, where a combination of textural and microgeochemical techniques may provide an ever more detailed picture of the magma plumbing system beneath individual volcanoes. Santorini, one of the decade volcanoes, is used as a well-studied example. The evolution of a shallow stratified magma chamber beneath Zao volcano in NE Japan is studied by Ban et al., using the petrology and geochemistry of eruption products within a c. 5.8 ka-old tephra layer. The authors deduce high rates of crystallization and infer that this chamber was short-lived. They also provide evidence for magma mixing due to intrusion of a new pulse of basaltic magma, which ultimately triggered the eruption.

Processes of silicic melt generation Background Since the seminal work of Bowen (1915), it is known that, when magmas crystallize, the separation of crystals from the residual melt leads to a chemical evolution of this melt and to the genesis of evolved magmas. Basalts generated by partial melting of the mantle, when cooling down and crystallizing at the crust-mantle boundary or within the crust, can differentiate and produce intermediate and silicic magmas. However, intermediate and

INTRODUCTION

silicic magmas can also be produced by partial melting of the crust. Experimental petrology shows that partial melting of amphibolites in the lower crust can produce calc-alkaline melts (Rapp & Watson 1995; Sisson et al. 2005). I-type granites are thought to be generated by this mechanism (Chappell & White 2001). The partial melting of crust of granodioritic, pelitic or greywacke composition produces more aluminous melts that are thought to be the origin of S- and some A-type melts (Clemens & Wall 1981; Patin˜o Douce 1997). Assimilation of crust and fractional crystallization may act concomitantly (AFC, DePaolo 1981), and these processes are variably important in the evolution of some silicic magmas. The generation of evolved magma by differentiation of mafic magma in shallow reservoirs is favoured by many authors (e.g. Sisson & Grove 1993; Grove et al. 1997; Pichavant et al. 2002). However, on the basis of geochemical data, Hildreth & Moorbath (1988) proposed that magma diversity is acquired at the mantle–crust interface in MASH zones, i.e. produced by processes of Mixing, Assimilation, Storage and Homogenization. Recently, more arguments based on petrological, geochemical and thermal data were presented in favour of differentiation in long-lived deep hot zones located in or below the lower crust where multiple sills of basalt are intruded and crystallise (Petford & Gallagher 2001; Bryan et al. 2002; Annen & Sparks 2002; Annen et al. 2006). In these studies, basalt differentiation and partial melting of the crust can happen simultaneously, although this is not necessarily the case. Thermal evolution and differentiation of each successive sill depend on its position on the geotherm, which changes with time as sills transfer heat to their surroundings. The geochemical and petrological characteristics of the melt depend on the time between sill emplacement and melt extraction. If melt segregation and extraction are by compaction, the melt can evolve further during segregation (Jackson et al. 2003). In recent years, isotopic data have elucidated the timescales of magmatic differentiation processes (cf. Turner & Costa 2007): decrease of 226Ra excesses during small degrees of differentiation within individual volcanic centres indicate that magmatic evolution may occur in the order of 103 years at some sites (George et al. 2003, and references therein). However, the general decrease of 238 U excesses from mafic towards andesitic compositions in arc lavas suggests that, in order to produce andesitic compositions from less evolved magmas, timescales of the order of 105 years are required, unless open system processes such as magma mixing are involved (Zellmer et al. 2005). Open system processes that involve mixing of young

7

mafic with older more felsic composition do appear to operate in many cases: they are evidenced through macroscopic and microscopic mixing and disequilibrium textures within intermediate volcanic products, and are suggested by the occasional persistence of 226Ra excesses to dacitic compositions (Zellmer et al. 2005, and references therein). Finally, the generation of some rhyolites has been shown to be associated with large scale rejuvenation of previously intruded felsic magmas from mid to upper crustal levels (Charlier et al. 2005; Lipman 2007). The rates of generation of such felsic magmas are more difficult to constrain, but evidence of assimilation of ancient crustal rocks indicates that thermal incubation times of the order of several hundred ka may be required (Zellmer et al. 2005). However, once the right thermal conditions are attained, rhyolite generation from basaltic magmas may proceed on timescales of the order of 104 years in some systems (Lowenstern et al. 2006).

Contribution of this volume Using uranium-series isotope constraints, Dosseto et al. show that differentiation from mafic to intermediate and felsic magmas is a rapid process at some arc volcanoes (e.g. ,2500 years for Mount St Helens). In contrast to suggestions of a longlived, deep-crustal hot zone as a key site for magmatic differentiation (Annen et al. 2006), the authors suggest that in the case of Mount St Helens, the magmatic evolution is constrained to the cooler environment of the mid crust. In their study of the Tuolumne Intrusive Suite of the Sierra Nevada Batholith, California, Gray et al. provide thermobarometric evidence for subsolidus exsolution of crystal phases at depths near 6 km, and argue that the scatter of trace element and isotopic signatures precludes fractional crystallization of a large magma chamber as the dominant process in the generation of the intrusive suite. Instead, the authors suggest a petrogenetic model based on mixing of mantle-derived mafic with granitic melts. Because the Tuolumne Intrusive Suite was emplaced slowly over at least 10 million years (Glazner et al. 2004), successive magma batches that formed the batholith were not molten simultaneously and could not mix after emplacement, which suggests that the magmas mixed at source level or during transport. In contrast, Burgess & Miller focus their study on the Cathedral Peak granodiorite, which is the largest unit of the Tuolumne Intrusive Suite. It is found that the Cathedral Peak was emplaced rapidly, suggesting that it may have represented a large mushy magma reservoir, where fractional crystallization and magma mixing did operate.

8

G. F. ZELLMER & C. ANNEN

The last two papers of the volume focus on bimodal volcanism. In the Snake River Plain– Yellowstone bimodal magmatic system, Leeman et al. argue that isotopic data on the initially erupted voluminous rhyolites indicate a dominant crustal component. Rhyolite petrogenesis is modelled through massive input of basalts into the mid to upper crust over several million years, leading to partial melting of several kilometres of the upper crust. Finally, Straub studied the composition of tephras of the central Izu–Bonin volcanic arc (NW Pacific) using mineral chemical data from an ODP drill core. Her work provides evidence for a remarkable constancy in tephra compositions over 42 Ma of eruptive history at this arc and, by inference, constancy in petrogenetic and eruptive processes over this timescale.

Discussion Integrating the evidence The processes of crustal magma transfer, storage and differentiation are inherently complex (Marsh 2004): ‘The local size, shape, and age of the system coupled with magma crystallinity, integrated flux, flushing frequency, and nature of wall rock involvement determines the local and systemwide products.’ The papers collected in this volume may serve as a means to move on from that acknowledgment towards an understanding of the dominating characteristics of magmatic systems. Despite the variety of systems addressed (Fig. 1), the overall theme of the papers in this volume is the importance of small volume magma batches that are transported through plumbing systems with a relatively high frequency, as opposed to less frequent transport of larger magma volumes that would have to be stored in larger crustal reservoirs for longer periods of time. This corroborates results from previous work on small volume systems (Bacon et al. 1981; Hildreth 1981; Detrick et al. 1987), and is consistent with studies of magma mixing (e.g. Sparks et al. 1977), and chamber replenishment and inflation (Bjo¨rnsson et al. 1977; Blake 1981). Frequent transport of small magma batches is directly evident at Popocatepetl, where magma transport to the surface occurs through repeated dyke injection (Martin-Del Pozzo et al.); at Santorini, where crystal size distribution and textural constraints indicate the presence of one or more small reservoirs that repeatedly experience recharge (Jerram & Martin); at Stromboli, where upper crustal magma residence times of a few days are inferred (Cigolini et al.); at Zao volcano and Mount St Helens, where rapid

crystallization indicates that the subvolcanic chambers are small and short-lived (Ban et al.; Dosseto et al.); and at Kilauea, where repeated injections into the volcanic edifice can be geophysically observed and correlated with the eruptive behaviour of the volcano (Wright & Klein). Highfrequency transport of small magma batches is also evident within the structurally less mature intra-oceanic Izu –Bonin volcanic arc, where Straub argues that the large number of tephra deposits with distinct compositions implies magma transfer processes that do not involve longterm storage within large crustal reservoirs. Further, it is consistent with the global study of Zellmer, which suggests that magma throughput through a lower crustal hot zone is close to steady state, with magma input from the mantle proportional to magma output into the overlying crust. Even the generation of large volumes of evolved melts forming the Snake-River Plain rhyolites (Leeman et al.) is seen as a result of repeated injection of relatively small volume magma batches into mid to upper crustal levels over millions of years. Finally, according to Gray et al., the Tuolumne Intrusive suite was generated by mixing of a number of successive magma batches. If movement of small volume melt batches is a common phenomenon in the generation and evolution of crustal magmatic systems, critical parameters responsible for variations in magma dynamics are the intrusion frequency and the melt segregation threshold. These will ultimately control the thermal structure of the crust, and therefore the amount and composition of magmas transferred through the plumbing system. Intrusion frequency is dependent on the rate of melt generation within the mantle. Further studies are required to determine the controls on melt segregation, although there is some indication that the local and regional stress regimes are important. These in turn will differ considerably between different tectonic settings.

Considerations regarding the tectonic setting The tectonic controls on magma ascent and storage, and links to differentiation mechanisms, have not yet been investigated in sufficient detail, and promise to yield many insights into the operation of crustal magma plumbing systems. Future research should therefore include studies that specifically focus on melt differentiation at ocean ridges with variable spreading rates, including onand off-axis sites of volcanic activity; at intra-plate volcanoes sampled at variable distance from the plume centre; and at arcs situated within contrasting tectonic regimes, e.g. extensional v. compressional. Traditionally, research in ocean ridge and intraplate

INTRODUCTION

settings has focused on the origin of the melts, the melting process, and insights that may be gained into the dynamics and composition of the underlying mantle. Studies of crustal magma evolution may, however, yield many important additional insights. For example, detailed results on oceanic rhyolitic volcanism in Iceland show that differentiation mechanisms range from near-liquidus processes (crystal fractionation + assimilation, e.g. Carmichael 1964; Macdonald et al. 1990; Nicholson et al. 1991; Lacasse et al. 2007) to nearsolidus scenarios (crustal melting, e.g. O’Nions & Gro¨nvold 1973; Sigvaldason 1974; He`mond et al. 1988; Sigmarsson et al. 1991; Jo´nasson 1994; Lacasse et al. 2007; Martin & Sigmarsson 2007; Zellmer et al. 2008). Future studies of melt evolution at ocean ridge and intraplate settings, as well as above subduction zones, will further refine our understanding of the processes and timescales operating during crustal magma transfer, storage and differentiation, and how they are affected by tectonic controls.

Conclusions 1.

2.

3.

4.

Crustal magma plumbing systems are characterized by a variety of processes that include dyke and sill propagation; magma accumulation, segregation, and mixing; crystallization, crystal resorption and recycling; magmatic volatile exsolution and degassing; and various assimilation processes that involve both old and juvenile crust. These can be linked to eruptive behaviour and the chemistry and petrology of erupted or intruded magmatic products. Despite the complexity and variety of crustal magmatic systems, a common theme in the dynamics of magma transfer and storage mechanisms appears to be the high-frequency processing of small melt batches, as opposed to long-term storage of large volumes of melt accumulated in crustal reservoirs. This evidence is mainly based on studies of subduction zone magmatism. Tectonic setting has the potential to exert a strong control on the geometry and evolution of crustal magma plumbing systems, and on the dynamics of magma transfer and storage within these systems. Future studies should address the links between magmatism and tectonism to improve our understanding of the crustal magmatic processes that operate in different tectonic regimes. The length and timescales governing the development of crustal magma reservoirs remain key to the understanding of the petrogenetic

9

processes operating in crustal magmatic systems. The authors would like to acknowledge all colleagues who have contributed to the many advances in our understanding of igneous processes through constructive discussions led throughout the last few years. We are grateful for the opportunity to present some of this progress within a dedicated volume. Our initial book proposal was improved through constructive comments by N. Petford, M. Wilson and an anonymous reviewer. This introduction benefited from reviews by P. Leat and P. Kokelaar. A. Hills provided administrative support throughout the editorial process. G. F. Z. thanks J. Erzinger at the GeoForschungsZentrum Potsdam for his hospitality during a productive sabbatical visit in summer 2007, and acknowledges support by the Institute of Earth Sciences, Academia Sinica and the National Science Council of Taiwan (NSC 96-2116-M001-006).

References A LOISI , M., B ONACCORSO , A. & G AMBINO , S. 2006. Imaging composite dike propagation (Etna, 2002 case). Journal of Geophysical Research–Solid Earth, 111, DOI: 10.1029/2005JB003908. A NNEN , C. & S PARKS , R. S. J. 2002. Effects of repetitive emplacement of basaltic intrusions on thermal evolution and melt generation in the crust. Earth and Planetary Science Letters, 203, 937– 955. A NNEN , C., B LUNDY , J. D. & S PARKS , R. S. J. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology, 47, 505– 539. B ACHMANN , O. & B ERGANTZ , G. W. 2004. On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes. Journal of Petrology, 45, 1565– 1582. B ACON , C. R., M ACDONALD , R., S MITH , R. L. & B AEDECKER , P. A. 1981. Pleistocene high-silica rhyolites of the Coso volcanic field, Inyo County, California. Journal of Geophysical Research, B86, 10223–10241. B ENN , K., R OEST , W. R., R OCHETTE , P., E VANS , N. G. & P IGNOTTA , G. S. 1999. Geophysical and structural signatures of syntectonic batholith construction: the South Mountain Batholith, Meguma Terrane, Nova Scotia. Geophysical Journal International, 136, 144– 158. B JO¨ RNSSON , A., S AEMUNDSSON , K., E INARSSON , P., T RYGGVASON , E. & G RO¨ NVOLD , K. 1977. Current rifting episode in north Iceland. Nature, 266, 318– 323. B LAKE , S. 1981. Volcanism and the dynamics of open magma chambers. Nature, 289, 783–785. B LATTER , D. L. & C ARMICHAEL , I. S. E. 1998. Hornblende peridotite xenoliths from central Mexico reveal the highly oxidized nature of subarc upper mantle. Geology, 26, 1035–1038. B OWEN , N. L. 1915. Crystallization – differentiation in silicate liquids. American Journal of Science, 39, 175– 191.

10

G. F. ZELLMER & C. ANNEN

B RIDGWATER , D., S UTTON , J. & W ATTERSON , J. 1974. Crustal downfolding associated with igneous activity. Tectonophysics, 21, 57–77. B ROCHER , T. M. 1981. Geometry and physical properties of the Socorro, New Mexico, Magma bodies. Journal of Geophysical Research, 86, 9420– 9432. B RYAN , S. E., R ILEY , T. R., J ERRAM , D. A., S TEPHENS , C. J. & L EAT , P. 2002. Silicic volcanism; an undervalued component of large igneous provinces and volcanic rifted margins. In: M ENZIES , M. A., K LEMPERER , S. L., E BINGER , C. J. & B AKER , J. (eds) Volcanic Rift Margins. Geological Society of America, Special Papers, 362, 97–118. C ABELLO , G. C., G ARZA , R. M. ET AL . 2006. Geology and paleomagnetism of El Potrero pluton, Baja California: Understanding criteria for timing of deformation and evidence of pluton tilt during batholith growth. Tectonophysics, 424, 1 –17. C ARMICHAEL , I. S. E. 1964. The petrology of Thingmuli, a Tertiary volcano in eastern Iceland. Journal of Petrology, 5, 435 –460. C ARTWRIGHT , J. & H ANSEN , D. M. 2006. Magma transport through the crust via interconnected sill complexes. Geology, 34, 929–932. C HAPIN , C. E. & E LSTON , W. E. (eds). 1979. Ash Flow Tuffs. Geological Society of America, Boulder, CO, Special Papers. C HAPPELL , B. W. & W HITE , J. R. 2001. Two contrasting granite types: 25 years later. Australian Journal of Earth Sciences, 48, 489–499. C HARLIER , B. L. A., W ILSON , C. J. N., L OWENSTERN , J. B., B LAKE , S., VAN C ALSTEREN , P. W. & D AVIDSON , J. P. 2005. Magma generation at a large, hyperactive silicic volcano (Taupo, New Zealand) revealed by U–Th and U–Pb systematics in zircons. Journal of Petrology, 46, 3 –32. C LEMENS , J. D. & M AWER , C. K. 1992. Granitic magma transport by fracture propagation. Tectonophysics, 204, 339–360. C LEMENS , J. D. & W ALL , W. J. 1981. Origin and crystallization of some peraluminous (S-type) granitic magmas. Canadian Mineralogist, 19, 111–131. C LEMENS , J. D., P ETFORD , N. & M AWER , C. K. 1997. Ascent mechanisms of granitic magmas: causes and consequences. In: H OLNESS , M. B. (ed.) Deformationenhanced Fluid Transport in the Earth’s Crust and Mantle. Chapman & Hall, London, 145–172. C OLEMAN , D. S., G RAY , W. & G LAZNER , A. F. 2004. Rethinking the emplacement and evolution of zoned plutons: Geochronologic evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology, 32, 433–436. C OSTA , F., C HAKRABORTY , S. & D OHMEN , R. 2003. Diffusion coupling between trace and major elements and a model for calculation of magma residence times using plagioclase. Geochimica et Cosmochimica Acta, 67, 2189–2200. C RUDEN , A. R. 1998. On the emplacement of tabular granites. Journal of the Geological Society, 155, 853– 862. C RUDEN , A. R. & M C C AFFREY , K. J. W. 2001. Growth of plutons by floor subsidence: Implications for rates of emplacement, intrusion spacing and melt-extraction mechanisms. Physics and Chemistry of the Earth Part a – Solid Earth and Geodesy, 26, 303–315.

D AWSON , P., W HILLDIN , D. & C HOUET , B. 2004. Application of near real-time radial semblance to locate the shallow magmatic conduit at Kilauea Volcano, Hawaii. Geophysical Research Letters, 31. DE S AINT -B LANQUAT , M., H ABERT , G., H ORSMAN , E., M ORGAN , S. S., T IKOFF , B., L AUNEAU , P. & G LEIZES , G. 2006. Mechanisms and duration of nontectonically assisted magma emplacement in the upper crust: The Black Mesa pluton, Henry Mountains, Utah. Tectonophysics, 428, 1– 31. D E P AOLO , D. 1981. Trace element and isotopic effects of combined wallrock assimilation and fractional crystallisation. Earth and Planetary Science Letters, 53, 189–202. D ETRICK , R. S., B UHL , P., V ERA , E., M UTTER , J., O RCUTT , J., M ADSON , J. & B ROCKER , T. 1987. Multi-channel seismic imaging of a crustal magma chamber along the East Pacific Rise. Nature, 236, 35–41. D ONOGHUE , S. L., G AMBLE , J. A., P ALMER , A. S. & S TEWART , R. B. 1995. Magma mingling in an andesite pyroclastic flow of the Pourahu member, Ruapehu volcano, New Zealand. Journal of Volcanology and Geothermal Research, 68, 177– 191. F ARRIS , D. W., H AEUSSLER , P., F RIEDMAN , R., P ATERSON , S. R., S ALTUS , R. W. & A YUSO , R. 2006. Emplacement of the Kodiak batholith and slab-window migration. Geological Society of America Bulletin, 118, 1360–1376. G EORGE , R. M. M., T URNER , S. P., H AWKESWORTH , C. J., B ACON , C. R., N YE , C., S TELLING , P. & D REHER , S. 2003. Chemical versus temporal controls on the evolution of tholeiitic and calc-alkaline magmas at two volcanoes in the Alaska– Aleutian arc. Journal of Petrology, 45, 203 –219. G LAZNER , A. F., B ARTLEY , J. M., C OLEMAN , D. S., G RAY , W. & T AYLOR , Z. T. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today, 14, 4– 11. G ROVE , T. L., D ONNELLY -N OLAN , J. M. & H OUSH , T. 1997. Magmatic processes that generated the rhyolite of Glass Mountain, Medicine Lake volcano, N. California. Contributions to Mineralogy and Petrology, 127, 205–223. H E` MOND , C., C ONDOMINES , M., F OURCADE , S., A LLE` GRE , C. J., O SKARSSON , N. & J AVOY , M. 1988. Thorium, strontium and oxygen isotopic geochemistry in recent tholeiites from Iceland: crustal influence on mantle-derived magmas. Earth and Planetary Science Letters, 87, 273–285. H ILDRETH , W. 1981. Gradients in silicic magma chambers: implications for lithospheric magmatism. Journal of Geophysical Research, 86, 10153–10192. H ILDRETH , W. & M OORBATH , S. 1988. Crustal contribution to arc magmatism in the Andes of Central Chile. Contributions to Mineralogy and Petrology, 98, 455–489. H UTTON , D. H. W., D EMPSTER , T. J., B ROWN , P. E. & B ECKER , S. D. 1990. A new mechanism of granite emplacement – intrusion in active extensional shear zones. Nature, 343, 452– 455. J ACKSON , M. D., C HEADLE , M. J. & A THERTON , M. P. 2003. Quantitative modeling of granitic melt generation and segregation in the continental crust. Journal of Geophysical Research, B7, DOI: 10.1029/2001JB001050.

INTRODUCTION J O´ NASSON , K. 1994. Rhyolite volcanism in the Krafla central volcano, north-east Iceland. Bulletin of Volcanology, 56, 516– 528. L ACASSE , C., S IGURDSSON , H., C AREY , S. N., J O´ HANNESSON , H., T HOMAS , L. E. & R OGERS , N. W. 2007. Bimodal volcanism at the Katla subglacial caldera, Iceland: insight into the geochemistry and petrogenesis of rhyolitic magmas. Bulletin of Volcanology, 69, 373– 399. L A D ELFA , S., P ATANE , G., C LOCCHIATTI , R., J ORON , J. L. & T ANGUY , J. C. 2001. Activity of Mount Etna preceding the February 1999 fissure eruption: inferred mechanism from seismological and geochemical data. Journal of Volcanology and Geothermal Research, 105, 121– 139. L AHR , J. C., C HOUET , B. A., S TEPHENS , C. D., P OWER , J. A. & P AGE , R. A. 1994. Earthquake classification, location and error analysis in a volcanic environment – implications for the magmatic system of the 1989–1990 eruptions at Redoubt volcano, Alaska. Journal of Volcanology and Geothermal Research, 62, 137–151. L EFORT , P. 1981. Manaslu leucoganite – a collision signature of the Himalya – a model for its genesis and emplacement. Journal of Geophysical Research, 86, 545–568. L EJEUNE , A. & R ICHET , P. 1995. Rheology of crystalbearing silicate melts: an experimental study at high viscosity. Journal of Geophysical Research, 100, 4215–4229. L IPMAN , P. 1984. The roots of Ash flow calderas in Western North America: windows into the tops of granitic batholiths. Journal of Geophysical Research, 89, 8801– 8841. L IPMAN , P. W. 2007. Incremental assembly and prolonged consolidation of Cordilleran magma chambers: evidence from the Southern Rocky Mountain volcanic field. Geosphere, 3, 42– 70. L OWENSTERN , J. B., C HARLIER , B. L. A., C LYNNE , M. A. & W OODEN , J. L. 2006. Extreme U–Th disequilibrium in rift-related basalts, rhyolites and granophyric granite and the time scale of rhyolite generation, intrusion and crystallization at Alid Volcanic Center, Eritrea. Journal of Petrology, 47, 2105–2122, DOI: 10.1093/petrology/egl038. M ACDONALD , R., M C G ARVIE , D. W., P INKERTON , H., S MITH , R. L. & P ALACZ , Z. A. 1990. Petrogenetic evolution of the Torfajo¨kull Volcanic Complex, Iceland I. Relationship between the magma types. Journal of Petrology, 31, 429–459. M ARSH , B. 2004. A magmatic mush column Rosetta Stone: the McMurdo dry valleys of Antarctica. EOS Transactions, American Geophysical Union, 85, 497–508. M ARSH , B. D. 1981. On the crystallinity, probability of occurrence, and rheology of lava and magma. Contributions to Mineralogy and Petrology, 78, 85–98. M ARSH , B. D. 1982. On the mechanics of igneous diapirism, stoping, and zone-melting. American Journal of Science, 282, 808–855. M ARTIN , E. & S IGMARSSON , O. 2007. Crustal thermal state and origin of silicic magma in Iceland: the case of Torfajo¨kull, Ljo´sufjo¨ll and Snæfellsjo¨kull volcanoes. Contributions to Mineralogy and Petrology, 153, 593– 605.

11

M ATTIA , M., P ATANE , D., A LOISI , M. & A MORE , M. 2007. Faulting on the western flank of Mt Etna and magma intrusions in the shallow crust. Terra Nova, 19, 89–94. M ATZEL , J. E. P., B OWRING , S. A. & M ILLER , R. B. 2006. Time scales of pluton construction at differing crustal levels: Examples from the Mount Stuart and Tenpeak intrusions, North Cascades, Washington. Geological Society of America Bulletin, 118, 1412– 1430. M C K ENZIE , D. 1984. The generation and compaction of partially molten rock. Journal of Petrology, 25, 713– 765. M ICHAUT , C. & J AUPART , C. 2006. Ultra-rapid formation of large volumes of evolved magma. Earth and Planetary Science Letters, 250, 38– 52. M ILLER , R. B. & P ATERSON , S. R. 1999. In defense of magmatic diapirs. Journal of Structural Geology, 21, 1161– 1173. M ORGAN , D. J. & B LAKE , S. 2005. Magmatic residence times of zoned phenocrysts: introduction and application of the binary element diffusion modelling (BEDM) technique. Contributions to Mineralogy and Petrology, 151, 58–70. M ORGAN , D. J., B LAKE , S., R OGERS , N. W., D E V IVO , B., R OLANDI , G., M ACDONALD , R. & H AWKESWORTH , C. J. 2004. Timescales of crystal residence and magma chamber volume from modelling of diffusion profiles in phenocrysts: Vesuvius 1944. Earth and Planetary Science Letters, 222, 933– 946. N ICHOLSON , H., C ONDOMINES , M., F ITTON , J. G., F ALLICK , A. E., G RO¨ NVOLD , K. & R ODGERS , G. 1991. Geochemical and isotopic evidence for crustal assimilation beneath Krafla, Iceland. Journal of Petrology, 32, 1005–1020. O’N IONS , R. K. & G RO¨ NVOLD , K. 1973. Petrogenetic relationships of acid and basic rocks in Iceland: Sr-isotopes and rare-earth elements in late and postglacial volcanics. Earth and Planetary Science Letters, 19, 397– 409. P ASQUARE , F. & T IBALDI , A. 2007. Structure of a sheet-laccolith system revealing the interplay between tectonic and magma stresses at Stardalur Volcano, Iceland. Journal of Volcanology and Geothermal Research, 161, 131 –150. P ATIN˜ O D OUCE , A. E. 1997. Generation of metaluminous A-type granites by low-pressure melting of calc-alkaline granitoids. Geology, 25, 743– 746. P ELTIER , A., F ERRAZZINI , V., S TAUDACHER , T. & B ACHELERY , P. 2005. Imaging the dynamics of dyke propagation prior to the 2000–2003 flank eruptions at Piton de La Fournaise, Reunion Island. Geophysical Research Letters, 32, DOI: 10.1029/ 2005GL023720. P ETFORD , N. 2003. Rheology of granitic magmas during ascent and emplacement. Annual Review of Earth and Planetary Sciences, 31, 399–427. P ETFORD , N. & G ALLAGHER , K. 2001. Partial melting of mafic (amphibolitic) lower crust by periodic influx of basaltic magma. Earth and Planetary Science Letters, 193, 483– 499. P ETFORD , N., C RUDEN , A. R., M C C AFFREY , K. J. W. & V IGNERESSE , J. L. 2000. Granite magma formation, transport and emplacement in the Earth’s crust. Nature, 408, 669– 673.

12

G. F. ZELLMER & C. ANNEN

P ICHAVANT , M., M ARTEL , C., B OURDIER , J.-L. & S CAILLET , B. 2002. Physical conditions, structure, and dynamics of a zoned magma chamber: Mount Pele´e (Martinique, Lesser Antilles Arc). Journal of Geophysical Research, 107, DOI: 10.1029/ 2001JB000315. P ITCHER , W. S. 1979. The nature, ascent and emplacement of granitic magmas. Journal of the Geological Society, 136, 627– 662. P OLLARD , D., D ELANEY , P., D UFFIELD , W., E NDO , E. & O KAMURA , A. 1983. Surface deformation in volcanic rift zones. Tectonophysics, 94, 541 –584. P RESTON , R. J. 2001. Composite minor intrusions as windows into subvolcanic magma reservoir processes: mineralogical and geochemical evidence for complex magmatic plumbing systems in the British Tertiary Igneous Province. Journal of the Geological Society, 158, 47–58. P RITCHARD , M. E. & S IMONS , M. 2004. An InSARbased survey of volcanic deformation in the central Andes. Geochemistry Geophysics Geosystems, 5, DOI: 10.1029/2003GL000610. Q UARENI , F., V ENTURA , G. & M ULARGIA , F. 2001. Numerical modelling of the transition from fissureto central-type activity on volcanoes: a case study from Salina Island, Italy. Physics of the Earth and Planetary Interior, 124, 213–221. R APP , R. P. & W ATSON , E. B. 1995. Dehydration melting of metabasalt at 8– 32 kbar; implications for continental growth and crust– mantle recycling. Journal of Petrology, 36, 891–931. R INEHART , E. J., S ANFORD , A. R. & W ARD , R. M. 1979. Geographic extent and shape of an extensive magma body at mid-crustal depths in the Rio Grande rift near Socorro, New Mexico, in Rio Grande Rift. In: R IECKER , R. E. (ed.) Tectonics and Magmatism. AGU, Washington, DC, 237– 251. R UBIN , A. M. & P OLLARD , D. D. 1987. Origins of bladelike dikes in volcanic rift zones. In: D ECKER , R. W., W IGHT , T. L. & S TUFFER , P. H. (eds) Volcanism in Hawaii. US Geoligical Survey Professional Papers, 1350, 1449–1470. S ANCHEZ , J. J., W YSS , M. & M C N UTT , S. R. 2004. Temporal –spatial variations of stress at Redoubt volcano, Alaska, inferred from inversion of fault plane solutions. Journal of Volcanology and Geothermal Research, 130, 1– 30. S IGMARSSON , O., H EMOND , C., C ONDOMINES , M., F OURCADE , S. & O SKARSSON , N. 1991. Origin of silicic magma in Iceland revealed by Th isotopes. Geology, 19, 621–624. S IGVALDASON , G. E. 1974. The petrology of Hekla and origin of silicic rocks in Iceland. Societas Scientarium Islandica, 5, 1 –44. S ISSON , T. W. & G ROVE , T. L. 1993. Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology, 113, 143– 166. S ISSON , T. W., R ATAJESKI , K., H ANKINS , W. B. & G LAZNER , A. F. 2005. Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology, 148, 635–661.

S PARKS , R. S. J. & H UPPERT , H. E. 1984. Density changes during the fractional crystallization of basaltic magmas – fluid dynamic implications. Contributions to Mineralogy and Petrology, 85, 300– 309. S PARKS , R. S. J., S IGURDSSON , H. & W ILSON , L. 1977. Magma mixing: a mechanism for triggering acid explosive eruptions. Nature, 267, 315– 318. S PARKS , R. S. J., H UPPERT , H. E. & T URNER , J. S. 1984. The fluid-dynamics of evolving magma chambers. Philosophical Transactions of the Royal Society of London Series a – Mathematical Physical and Engineering Sciences, 310, 511. T URNER , S., B OURDON , B., H AWKESWORTH , C. & E VANS , P. 2000. 226Ra– 230Th evidence for multiple dehydration events, rapid melt ascent and the time scales of differentiation beneath the Tonga– Kermadec island arc. Earth and Planetary Science Letters, 179, 581–593. T URNER , S. P. & C OSTA , F. 2007. Measuring time scales of magmatic evolution. Elements, 3, 267– 272. V AN DEN E ECKHOUT , B., G ROCOTT , J. & V ISSERS , R. 1986. On the role of diapirism in the segregation, ascent and final emplacement of granitoid magmas – discussion. Tectonophysics, 127, 161– 166. V ANDERMOLEN , I. & P ATERSON , M. S. 1979. Experimental deformation of partially-melted granite. Contributions to Mineralogy and Petrology, 70, 299–318. V IGNERESSE , J. L. 2004. A new paradigm for granite generation. Transactions of the Royal Society of Edinburgh – Earth Sciences, 95, 11– 22. W HITE , A. J. R. & C HAPPELL , B. W. 1977. Ultrametamorphism and granitoid genesis. Tectonophysics, 43, 7– 22. W HITEHEAD , J. A. & L UTHER , D. S. 1975. Dynamics of laboratory diapir and plume models. Journal of Geophysical Research, 80, 705–717. Y AMAOKA , K., K AWAMURA , M., K IMATA , F., F UJII , N. & K UDO , T. 2005. Dike intrusion associated with the 2000 eruption of Miyakejima Volcano, Japan. Bulletin of Volcanology, 67, 231– 242. Z AK , J. & P ATERSON , S. R. 2006. Roof and walls of the Red Mountain Creek pluton, eastern Sierra Nevada, California (USA): implications for process zones during pluton emplacement. Journal of Structural Geology, 28, 575– 587. Z ELLMER , G. F. & C LAVERO , J. 2006. Using trace element correlation patterns to decipher a sanidine crystal growth chronology: an example from Taapaca volcano, Central Andes. Journal of Volcanology and Geothermal Research, 156, 291–301, DOI: 10.1016/j.jvolgeores.2006.03.004. Z ELLMER , G. F., A NNEN , C., C HARLIER , B. L. A., G EORGE , R. M. M., T URNER , S. P. & H AWKESWORTH , C. J. 2005. Magma evolution and ascent at volcanic arcs: constraining petrogenetic processes through rates and chronologies. Journal of Volcanology and Geothermal Research, 140, 171–191. Z ELLMER , G. F., B LAKE , S., V ANCE , D., H AWKESWORTH , C. & T URNER , S. 1999. Plagioclase residence times at two island arc volcanoes (Kameni islands, Santorini, and Soufriere, St. Vincent) determined by Sr diffusion systematics. Contributions to Mineralogy and Petrology, 136, 345– 357.

INTRODUCTION Z ELLMER , G. F., R UBIN , K. H., G RO¨ NVOLD , K. & J URADO -C HICHAY , Z. 2008. On the recent bimodal magmatic processes and their rates in the Torfajo¨kull-Veidivo¨tn area, Iceland. Earth and Planetary Science Letters, 269, 387– 397.

13

Z ELLMER , G. F., S PARKS , R. S. J., H AWKESWORTH , C. J. & W IEDENBECK , M. 2003. Magma emplacement and remobilization time scales beneath Montserrat: insights from Sr and Ba zonation in plagioclase phenocrysts. Journal of Petrology, 44, 1413–1431.

Some first-order observations on magma transfer from mantle wedge to upper crust at volcanic arcs GEORG F. ZELLMER1,2 1

Institute of Earth Sciences, Academia Sinica, 128 Academia Road Sec. 2, Nankang, Taipei 11529, Taiwan, ROC (e-mail: [email protected]) 2

Lamont-Doherty Earth Observatory of Columbia University, 61 Route 9W, Palisades, New York 10964, USA Abstract: The viscosity of lavas erupted at volcanic arcs varies over orders of magnitude. A comparison of the relative abundance of viscous lava dome eruptions indicates that the average viscosity of arc lavas also varies considerably between arcs. It is shown that, for continental or transitional arcs with little within-arc crustal deformation and without underlying slab windows or tears, average lava viscosity is anticorrelated with average surface heat flux. The latter may be influenced by crustal thickness and crustal magma throughput. To constrain the relative contributions of these parameters, variations of average lava viscosity with average crustal thickness and plate convergence rate are assessed. While crustal thickness appears to have little effect on average lava viscosity, a good anticorrelation exists between average lava viscosity and plate convergence rate, with the exception of two arcs that show significant intra-arc crustal deformation. If plate convergence rate is a good proxy of the rate of melt generation within the mantle wedge, these first-order observations indicate that, where the rate of mantle melting is high, crustal magma throughput is rapid and efficient, resulting in low-viscosity melts migrating through a hot overriding crust; in contrast, where the rate of mantle melting is low, crustal magma transfer is slow and inefficient, resulting in high-viscosity melts that may frequently stall within a cool overriding crust prior to eruption. Uranium series geochemical evidence from dome lavas is presented and lends support to this interpretation. Finally, some explanations are offered for the observed average viscosity variations of arcs with underlying slab windows or tears and/or significant intra-arc crustal deformation.

Volcanic activity above subduction zones is characterized by a variety of eruption styles: in the explosive regime, these include plinian eruptions (e.g. Santorini c. 1645 BC , Vesuvius AD 79, Taupo AD 186, Krakatau 1883, St Helens 1980), vulcanian eruptions (e.g. Vulcano 1888– 1890, Stromboli 1930, Irazu 1965, Rabaul 1998, Sakura-jima ongoing) and strombolian eruptions (e.g. Stromboli ongoing, Paricutin 1943–1952, Izalco 1770–1966). In the effusive regime, they include the formation of lava lakes (e.g. Batur 1926, Ambrym 1935 –1996, Sakura-jima, 1955, Tolbachik, 1964, Masaya and Villarica ongoing), eruption of lava flows (e.g. Bagana, Bezymianny, Ceboruco, Colima, Lonquimay, Mayon, Newberry, and many others), extrusion of lava domes (e.g. Bezymianny, El Chichon, Merapi, Soufrie`re Hills, St Helens, Unzen, and many others) and extrusion of spines (e.g. Pele´e 1902 and 1929, Fuego 1955, St Helens 1980, Pinatubo 1992, Soufrie`re Hills 1995; see Simkin & Siebert 2002, for references). Disregarding external factors such as water– magma interaction, eruptive style appears to be dominated by two factors: firstly, the content and evolution of pre- and syn-eruptive volatile

content of the magma, which strongly influences the explosivity of volcanism (e.g. Burnham 1975; Sparks et al. 1977), and secondly the viscosity of the magma, which is principally a function of temperature and crystal content (Pinkerton & Stevenson 1992; Lejeune & Richet 1995; Costa 2005), and to some degree of melt composition. The latter, however, appears to play a subordinate role: an increase in total crystallinity from 40 to 60 vol% has a three orders of magnitude greater effect on viscosity than a change from basaltic to dacitic liquid compositions at a given temperature (Lejeune & Richet 1995; Giordano & Dingwell 2003). The variety of eruption styles observed at any individual volcanic arc, and even through time at individual volcanic centres, indicates that volatile content and viscosity are subject to small-scale spatial and temporal variations. Determining firstorder differences in magma transfer dynamics between different arcs requires the establishment of an average measure of the character of the erupted products at each arc. Average lava viscosity, as a function of average composition (including volatile content), temperature and crystallinity, combines the effects of all of these

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 15– 31. DOI: 10.1144/SP304.2 0305-8719/08/$15.00 # The Geological Society of London 2008.

16

G. F. ZELLMER

parameters, and may therefore be a suitable variable for distinguishing first-order differences between volcanic arcs. In this study, average lava viscosities of a variety of volcanic arcs are determined using information from the Holocene eruption database of the Global Volcanism Program (GVP), (Simkin & Siebert 2002). Variations in average lava viscosities between arcs are then investigated in the context of variations in average surface heat flux, average crustal thickness, and plate convergence rate. This leads to constraints on the dynamics of magma transfer from mantle wedge to upper crust, which will be discussed with reference to recent uranium series geochronological data. As this contribution deals with averaged data and considers arcs within a global comparison, its inferences are not intended to constrain the details of processes that operate on a more local scale. The insights from this work merely provide a framework within which results from local studies may be discussed, particularly where exceptions may be evident. Nevertheless, it will be shown that correlations exist between averaged parameters, indicating that magma generation and transfer at volcanic arcs follow common principles globally. Improvements of the resolution of existing datasets on surface heat flux, crustal thickness and convergence rate, in combination with additional volcanological and geochemical studies, are likely to provide further insights into these principles.

Methodology Determining average viscosity: definition and significance of the lava dome proportion In order to constrain the average viscosity of lavas erupted at volcanic arcs, the GVP Holocene eruption database (Simkin & Siebert 2002) is used to characterize the style of effusive volcanism. The total number of Holocene effusive eruptions at any individual arc segment is given by the number of entries in the GVP eruption database that record lava flows, lakes, domes or spines within that segment. To compare the relative proportions of effusive arc lavas of different viscosities between different arcs or arc segments, the lava dome proportion is defined as the proportion of Holocene effusive eruptions that produce a lava dome or spine. Table 1 details the total number of effusive eruptions and the number and proportion of domes/spines for each of the 29 arc segments included in this study. The lava dome proportion is thus a dimensionless proxy of the average viscosity of the erupted lavas within an arc segment, ranging from 0 (no domes or spines are formed by

effusive activity) to 100% (all effusive activity involves dome or spine formation). Explosive eruptions may also be a marker for high magma viscosity or water content. However, without detailed geochemical and volcanological information on the eruption products, it is impossible to determine the relative role that viscosity and water content played in producing explosive activity. To circumvent this issue, this study exclusively considers effusive eruptions to evaluate average magma viscosities.

Characterization of volcanic arcs and definition of irregular arcs In this study, a total of 29 arcs or arc segments with at least 10 Holocene effusive eruptions, as recorded in the GVP eruption database, are considered (cf. Table 1). Excluded are Adaman, Sumatra, Banda, Tonga, Kermadec, South Central Alaska, Peru and Tierra del Fuego, as these have less than 10 Holocene effusive eruptions and therefore do not provide good constraints on average magma viscosity. (The Tongean island of Niuafo’ou, on which eight lava flows have been recorded, has not been included in the analysis of the Tonga arc, because it lies over 200 km behind the volcanic front.) A detailed list of arc segments and their volcanoes with Holocene effusive eruptions is given in the Appendix. For the purposes of this analysis, arcs are divided into oceanic arcs (ALU, ANT, BIS, CVA, HAL, MAR, NIZ, SAN, SCO, and SOL; see Table 1 for acronyms) and transitional to continental arcs (all other arcs listed in Table 1). Some arcs show striking irregularities either in the features of the overriding crust or the subducting plate, and such arcs are here defined as irregular arcs as opposed to regular arcs with no such features: irregularities in terms of ongoing crustal deformation may affect the processes by which magmas migrate to the surface. Crustal deformation occurs in CAS, where intra-arc shear is resulting in northward movement of the Cascadia margin relative to North America at a rate of up to 9 mm year21 (Miller et al. 2001), NEJ, where intra-arc thrusting has been identified to occur on basis of geological and geodetic constraints (Seno 1999; Townend & Zoback 2006), and NZL, which is the only arc that lies on a continental rift boundary as defined by Bird (2003). Further, some arc segments have underlying slab windows, tears or cracks, which are known to affect the characteristics of the mantle source region, e.g. through upwelling of hotter asthenospheric mantle, frequently associated with partial melting of the slab edges and resulting adakite-type volcanism at the surface (Yogodzinski et al. 2001). Thus, the following

Table 1. Arcs with 10 or more Holocene effusive eruptions Code

AEG AEO AKP ALU ANT BIS CAM CAS1,2 COL CVA ECU EJV HAL KAM LSU MAR MEX1 NCH1 NEJ1,3 NIZ NZL4 PHL SAN SCH SCO SKU SOL SWJ1 WJV 1

Arc

Aegean Volcanic Arc Aeolian Volcanic Arc Alaska Peninsula Aleutian Arc Lesser Antilles Volcanic Arc Bismarck/New Britain Arc Central American Arc Cascades Volcanic Arc Columbia Central Vanuatu/New Hebrides Ecuador E. Java Halmahera Arc Kamchatka (incl. N. Kurile) Lesser Sunda Arc Mariana Arc Mexican Volcanic Belt N. Chile Northeast Japan N. Izu Arc New Zealand Philippine Arc Sangihe (incl. N. Sulawesi) S. Chile Scotia/S. Sandwich Arc S. Kurile Solomon Arc Southwest Japan and N. Ryukyu W. Java

Effusive eruptions Total

Domes and spines

11 16 24 58 28 41 124 92 16 41 34 40 10 248 40 15 59 10 37 80 27 37 41 77 10 21 23 42 60

10/90.9% 3/18.8% 8/33.3% 9/15.5% 26/92.9% 2/4.9% 12/9.7% 36/39.1% 12/75.0% 0/0.0% 17/50.0% 6/15.0% 3/30.0% 79/31.9% 12/30.0% 2/13.3% 31/52.5% 4/40.0% 20/54.1% 7/8.8% 16/59.3% 11/29.7% 29/70.7% 4/5.2% 0/0.0% 7/33.3% 5/21.7% 15/35.7% 42/70.0%

Surface heat flux

Crustal thickness

Standard Average Standard Average Standard Average (km) deviation (%) (mm year21) deviation (%) (mW m22) deviation (%) 58.8 86.2 76.9

1.4 1.5 6.7

82.3 79.9 58.9

5.1 10.8 2.2

67.6 79.6

3.8 2.0

84.1 71.3

3.0 6.3

110.5 97.6 81.8

18.8 12.6 17.7

79.9 74.1

15.7 3.5

78.8

6.0

70.4

5.0

87.2 66.2

3.4 2.2

28.2 24.9 27.4 18.9 24.7 22.5 28.0 38.8 43.2 15.6 38.4 28.9 27.8 24.6 32.0 14.5 30.3 65.1 29.2 20.5 28.6 27.8 27.4 39.9 11.8 18.3 19.7 24.5 26.7

2 4 15 23 3 29 25 5 4 1 12 7 9 22 5 7 18 1 9 13 11 16 8 3 1 15 3 14 6

Confirmed slab discontinuity. Volcanic arc lies within crustal active shear zone. Volcanic arc lies within active thrust zone. 4 Volcanic arc lies within zone of active crustal extension. 5 Pasquale et al. (2005). 6 Based on 103.3 mm year21 Molucca Sea convergence, split 1:3 to 2:3 between SAN and HAL, based on relative volcano spacing. 7 Conservative estimate. 2 3

Convergence rate

34.6 60.05 58.7 61.8 16.8 88.8 71.6 34.2 43.0 92.8 51.1 63.2 68.96 80.4 66.6 60.0 52.1 70.4 88.5 87.1 44.2 73.7 34.46 73.5 72.7 81.8 75.8 58.4 56.5

8 207 3 7 17 26 8 18 18 5 9 1 207 4 2 10 11 10 5 7 9 17 207 7 18 4 19 14 8

Trench segment for convergence rate Longitude, initial (8E)

Longitude, initial (8N)

Longitude, final (8E)

Longitude, final (8N)

20.760 16.824 2160.910 179.688 257.424 148.587 296.219 2124.742 280.292 166.262 281.599 110.159 126.872 155.742 113.756 147.462 297.691 271.847 145.077 142.067 178.566 124.891 126.872 276.006 224.456 145.077 153.93 131.309 104.576

37.098 37.398 53.621 50.506 11.661 27.395 15.361 40.313 1.549 214.889 21.814 210.440 2.346 47.725 210.964 14.966 15.508 227.248 41.319 35.164 240.424 14.704 2.346 245.659 259.386 41.319 26.265 28.451 28.167

24.593 17.845 2150.547 2161.664 259.819 152.653 285.044 2126.486 278.008 167.155 280.292 113.756 125.969 164.066 120.886 147.019 2105.247 272.266 142.067 141.858 179.366 127.219 125.969 272.678 224.130 146.198 159.503 135.026 110.159

34.340 38.461 56.499 53.496 17.548 25.640 8.937 47.996 6.005 217.307 1.549 210.964 0.190 55.209 211.493 20.805 18.762 218.722 35.164 33.704 238.538 8.136 0.190 231.531 257.425 42.076 210.021 32.326 210.440

18

G. F. ZELLMER

arcs are classified as irregular on basis of slab discontinuities: CAS, where adakite-type volcanism and high surface heat flux are observed in the south at Lassen and Mount Shasta, east of the Mendocino triple junction and above the southern edge of the subducting Juan de Fuca plate (Baker et al. 1994; Borg et al. 1997); MEX, where slab detachment is evidenced by a lack of seismicity directly beneath the arc, and where eastward propagating late Miocene OIB-type and adakitic volcanism occurred due to asthenospheric mantle upwelling (Ferrari 2004), consistent with the present-day surface heat flux increase towards the west; NCH, where ongoing work indicates slab tearing as far south at 218S (Rietbrock et al. 2006); NEJ, where a slab window in the Philippine Sea plate widens toward the backarc north of Mount Fuji (Ishida 1992; Mazzotti et al. 1999), and where an extensive slab crack beneath the Hokkaido corner (Katsumata et al. 2003) widens towards the backarc (Lundgren & Giardini 1990), consistent with elevated surface heat flux in both the central Honshu backarc and northern Honshu; and SWJ, where the slab ruptures between Honshu and Kyoshu (Zhao et al. 2002) with slab melting proposed to result in adakitetype volcanism in SW Honshu (Morris 1995; Kimura et al. 2005) and in NE Kyushu (Sugimoto et al. 2006). Slab windows and tears have also been identified in the westernmost Aleutians (Yogodzinski et al. 2001), south of the Central American volcanic front in southeastern Costa Rica and Panama (Johnston & Thorkelson 1997), beneath Tierra del Fuego north and east of the Austral Volcanic Zone (Gorring & Kay 2001), and in the southern Ryukyu arc (Lin et al. 2004). However, due to the low number of Holocene effusive eruptions, these arc segments are not part of this study; although Shiveluch in northernmost KAM erupts adakitetype lavas related to the western Aleutian slab window (Yogodzinski et al. 2001), this does not affect the average character of this arc with about 50 Holocene volcanoes, and KAM is therefore defined here as a regular arc.

Calculation of the weighted average surface heat flux Reliable surface heat flux measurements are generally difficult to make at arcs due to considerable short-wavelength variations in heat flux close to areas of active volcanism. However, a good approximation of average surface heat flux is available for continental and transitional arcs using inferred heat flux distributions guided by a global seismic model (Shapiro & Ritzwoller 2004), thereby filtering out any short-wavelength

variations. Shapiro & Ritzwoller (2004) inferred probability distributions of global surface heat flux on a 1  1 degree global grid, guided by a three-dimensional shear velocity model of the crust and uppermost mantle. Here, weighted average surface heat fluxes are calculated for volcanic arc segments as follows: firstly, a surface heat flux value is attributed to each Holocene effusive eruption based on the surface heat flux at the 1  1 degree grid square on which the eruption occurred. Then the average and standard deviation of all attributed surface heat flux values of the arc segment are obtained. Standard deviations are typically around 5% or lower (see Table 1), indicating that variations in average surface heat flux along individual arc segments are small compared with the range of average surface heat flux values that exists between arcs. This suggests that the relative differences are real and reliable, despite the low precision of the average surface heat flux of individual grid squares, which typically carry standard deviations of 35 to 65 mW m22. Exceptions are five irregular arcs (CAS, MEX, NCH, NEJ, NZL) that display significant variations in along- and/or across-arc surface heat flux. The average surface heat flux of oceanic arcs cannot be calculated through this approach because model resolution is too coarse to resolve these narrow arcs (Shapiro, pers. comm., 2006).

Calculation of the weighted average crustal thickness There are significant variations in crustal thickness estimates at any one arc, depending on the crustal velocity model employed and the time of data collection, with early data usually being quite imprecise. Taking the Alaska –Aleutian arc as an example, we find that early estimates range from 15 –25 km in the western Aleutians to 35–40 km in the eastern Aleutians and Alaska (cf. compilation by Leeman 1983). Later seismic studies indicated maximum crustal thicknesses of 25– 30 km in the central and eastern Aleutians (Fliedner & Klemperer 1999; Holbrook et al. 1999; Lizarralde et al. 2002). However, the most recent wide-angle seismic studies argue for a more mafic, seismically faster middle and lower crust, resulting in a significantly greater crustal thickness of 35–37 km (Shillington et al. 2004; Van Avendonk et al. 2004). Similar discrepancies in crustal thickness estimates exist in many other arcs. While estimates of the absolute crustal thickness in any particular area may be highly dependent on seismic data quality and the crustal velocity structure used, for the purpose of this study it is critical to obtain reliable relative estimates of

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS

crustal thicknesses between different arc segments. Instead of using constraints from different regional studies, this contribution therefore uses a recent global crustal model at 2  2 degrees, CRUST 2.0, administered by the US Geological Survey and the Institute for Geophysics and Planetary Physics at the University of California (Bassin et al. 2000). CRUST 2.0 is an updated version of CRUST 5.1, a global crustal model at 5  5 degrees (Mooney et al. 1998). Both models are based on seismic refraction data published up to 1995 and a detailed compilation of sediment thickness. Here, weighted average crustal thicknesses are calculated for volcanic arc segments as follows: first, the total crustal thickness of CRUST 2.0 is transposed to a 1  1 degrees grid by linear interpolation. While interpolation may not produce accurate results for individual grid squares, it is unlikely to lead to significant errors in average crustal thickness, because in most arc segments a large number of effusive eruptions from different grid squares are considered. Second, a crustal thickness value is attributed to each Holocene effusive eruption based on the crustal thickness of the 1  1 degree grid square on which the eruption occurred. Finally, the average and standard deviation of all attributed crustal thickness values of the arc segment are obtained. Standard deviations range from 1 to about 30% (Table 1), but are typically lower than 15%. Given the relatively old (pre-1996) seismic data, it is likely that the crustal thickness estimates are not very accurate for any given arc, and probably are in many cases underestimates in the view of more recent work that advocates seismically fast lower arc crust. For example, a crustal thickness of 18.9 + 4.3 km (1s) is estimated for the Aleutian arc using CRUST 2.0 (Table 1), which is very thin compared with recent estimates, although it includes the western Aleutians that are situated on thinner crust. Nevertheless, using a global crustal model is the most coherent approach when comparing crustal thicknesses between arcs, providing some confidence in the relative precision of differences in crustal thicknesses between arcs.

Calculation of the plate convergence rate Early estimates of plate convergence rates were based on work by Chase (1978) and Minister & Jordan (1978), and have previously been compiled in a study on relations among subduction parameters (Jarrard 1986). Since then, a number of plate motion models have been published and refined by increasingly precise geodetic measurements (e.g. Argus & Gordon 1991; DeMets et al. 1994; Altamimi et al. 2002; Kreemer et al. 2003). These models constrain the relative motions of the larger plates, but are generally not sufficiently

19

detailed to compare the convergence across many of the arc segments of interest, particularly in the complex neotectonic areas of the western Pacific and Southeast Asia. However, the recent publication of global digital data sets on topography, seismicity, seafloor age, and geodetic velocity allowed Bird (2003) to present a global digital model of plate boundaries and motions for a total of 52 plates. This model is used here to calculate the average plate convergence rates at individual arc segments as follows: firstly, the width of the volcanic arc segment is defined by the position of the outermost volcanic edifices that produce Holocene effusive eruptions in each segment. Initial and final coordinates of the trench (Table 1) are then chosen accordingly, taking into account the relative direction of motion between the subducting and the overriding plates. Finally, the average convergence rates and standard deviations are calculated by averaging the convergence rates across all plate boundary steps that make up the trench segment under consideration, weighted by the relative lengths of these steps. Strike –slip motions along individual plate boundary steps are not considered. The obtained standard deviations range from 1 to about 25% (Table 1), but are typically lower than 15%.

Results Figure 1 shows the variation of lava dome proportion with weighted average surface heat flux at 19 continental and transitional volcanic arcs. Two principle observations can be made: 1.

2.

For the regular arcs, a good anticorrelation exists between the lava dome proportion and the weighted average surface heat flux (R 2 ¼ 0.77, MSWD ¼ 3.6). Irregular arcs all plot towards high average surface heat flux, and most show significant along- and/or across-arc variations in surface heat flux, as indicated by their large standard deviations.

Figure 2 shows the variation of lava dome proportion with weighted average crustal thickness at all 29 volcanic arcs or arc segments under consideration. NCH with a crustal thickness of around 65 km plots off scale. Evidently, there is no correlation of lava dome proportion with crustal thickness, although it may be noted that, in arcs with crustal thickness of less than about 25 km, the lava dome proportion is typically less than 40%. Figure 3 shows the variation of lava dome proportion with average convergence rate at all 29 arcs included in this study. For consistency with Figure 1, a correlation coefficient is calculated on basis of the regular arcs only. With the exception

20

G. F. ZELLMER

Fig. 1. Variation of lava dome proportion (a dimensionless proxy of average lava viscosity) with weighted average surface heat flux for regular (open diamonds) and irregular (grey diamonds) continental to transitional volcanic arcs. See Methodology section for definitions. Standard deviations (+1s) in average surface heat flux are indicated. Uncertainties in lava dome proportion are based on the movement of a data point with the next effusive eruption (up in case of a lava dome eruption) and are an indication of the number of eruptions used to constrain average lava viscosity. Refer to Table 1 or the Appendix for arc acronyms. A good anticorrelation is evident for regular continental and transitional arcs.

of two irregular arcs (CAS and NEJ), there is a good anticorrelation between lava dome proportion and average convergence rate (R 2 ¼ 0.71), although the relatively high MSWD of 9.5 indicates that there may be some additional scatter introduced by other parameters (Ludwig 2003).

Discussion As outlined in the methodology section, the lava dome proportion may be understood as a dimensionless proxy of average lava viscosity. The following discussion therefore focuses on the variation of average lava viscosity with parameters such as heat flux, crustal thickness and convergence rate. Evidence will also be presented from recent uranium series geochronological data. However,

before these variations are discussed, the reliability of the GVP eruption database with respect to the style of eruptive activity is briefly considered.

Reliability and accuracy of the GVP eruption style data In the literature, no quantitative definition of the difference between lava flows and lava domes can be found. The GVP states that ‘Lava domes are formed when viscous magma slowly extrudes from a vent and piles up around it. . . . Domes are steep-sided structures typically a few tens of meters to a few hundred meters high’, and ‘Lava flows . . . are distinguished from lava domes by their elongated extent downslope’. While this distinction may not always be easy to make in the

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS

21

Fig. 2. Variation of lava dome proportion (a dimensionless proxy of average lava viscosity) with weighted average crustal thickness for regular continental and transitional arcs (open diamonds), oceanic arcs (black diamonds) and irregular arcs (grey diamonds). See Methodology section for definitions. Standard deviations (+1s) in average crustal thickness are indicated. Uncertainties of lava dome proportions as in Figure 1. Refer to Table 1 or the Appendix for arc acronyms. No coherent variation is observed, although viscous dome eruptions are uncommon in arcs situated on thin crust, as indicated by the dashed line.

field, the lack of a more quantitative definition may point to the fact that the difference between flows and domes is usually visually very obvious, and that the transition between flow-forming and domebuilding eruptions may be quite abrupt. This would certainly be expected for crystal bearing lavas, where a small change in crystallinity may lead to a large change in viscosity (Lejeune & Richet 1995). It is therefore concluded here that the GVP record of effusive eruption style may be quite reliable for any individual eruption. Another issue is the quantity of effusive eruptions recorded in the GVP database for individual volcanic arcs. The GVP does not claim to provide complete coverage of all Holocene eruptions, and some arcs are better studied than others. For example, very few effusive eruptions are recorded for Sumatra: according to the database, only five

of a total of 177 Holocene eruptions occurred effusively. Other arcs for which fewer than 10 Holocene effusive eruptions are recorded are South Central Alaska, Tonga, Banda, Adaman, Kermadec, Peru and Tierra del Fuego. These arcs are all situated in regions that are sparsely populated, lack infrastructure, or are densely vegetated. As many effusive deposits may have been overlooked in these arcs, they are not considered in this study. The coverage of the GVP database is probably much better for accessible and well studied arcs, and there the relative proportion of lava domes and lava flows can probably be regarded as reasonably accurate. Finally, the good correlations of lava dome proportion with some geophysical parameters suggest that the dataset presented here yields important systematics and is not subject to large random errors that would be expected if the GVP data was

22

G. F. ZELLMER

Fig. 3. Variation of lava dome proportion (a dimensionless proxy of average lava viscosity) with average convergence rate for regular continental and transitional arcs (open diamonds), oceanic arcs (black diamonds) and irregular arcs (grey diamonds). See Methodology section for definitions. Standard deviations (+1s) in average convergence rate are indicated. Uncertainties of lava dome proportions as in Figure 1. Refer to Table 1 or the Appendix for arc acronyms. For consistency with Figure 1, the anticorrelation is calculated on basis of the regular (including oceanic) arcs only, although including MEX, NCH, NZL and SWJ does not significantly change the correlation. Note the higher MSWD compared with Figure 1.

unreliable. These systematics will now be discussed in detail.

Correlation with surface heat flux The striking anticorrelation of average lava viscosity with average surface heat flux in regular continental and transitional arcs (Fig. 1) implies a strong thermal control on the viscosity of arc magmas and the resulting style of effusive arc volcanism. The parameters that control surface heat flux are the key to the interpretation of this data. Using Fourier’s law, qH ¼ krT, and given there is no a priori information about potential differences in average thermal conductivity k between different arcs, surface heat flux qH is proportional to the near-surface geothermal gradient, which in turn is dependent on (1) the background

geothermal gradient, (2) upper crustal radiogenic heat production and (3) any thermal perturbations introduced through shallow level magma reservoirs. These will be considered in turn below. 1.

The background geothermal gradient at arcs can be inferred by constraining Moho temperature and crustal thickness. Kelemen et al. (2003) summarized the evidence for arc Moho temperatures of about 1150 + 150 8C. Lower temperatures yielded by studies of exposed arc sections (e.g. DeBari & Coleman 1989) and metaplutonic xenoliths (DeBari et al. 1987) are closure temperatures and therefore underestimates. There is seismic evidence of extensive melt lenses in the uppermost mantle of some arcs (e.g. Zhao & Hasegawa 1994; Zhao et al. 1997), and petrological

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS

2.

3.

studies constrain magma fractionation temperatures to 1100–1300 8C at the depth of the Moho in the Cascades (Elkins Tanton et al. 2001). These considerations suggest a variation of less than +15% in Moho temperature, significantly lower than the crustal thickness variations of a factor of 3–5 that exist between different arc lavas (cf. Table 1 and Fig. 2). It may therefore be concluded that the background geothermal gradient is largely a function of crustal thickness. On average, about 25 mW m22 of the total continental heat flux arises through radiogenic heat production within the crust (e.g. Pollack 1982). This is about one-quarter to one-third of the average heat flux above most continental and transitional arcs. Further, there is a general tendency for surface heat flux to increase with increasing near-surface radiogenic heat generation, although in most areas the depth distribution of radiogenic heat generation is not well constrained (cf. Drury 1987). While globally there is significant lateral heterogeneity in the composition of the continental crust, there is no a priori evidence for a systematic variation in radiogenically produced heat between different volcanic arcs. Given these observations, it is unlikely that differences in radiogenically produced heat may account for the observed range of c. 60 to .100 mW m22 in average surface heat flux of continental and transitional volcanic arcs. In contrast, there is evidence from surface heat flux measurements in volcanically active areas that thermal perturbations introduced by shallow level magma reservoirs have a very strong effect on the thermal gradient within the upper crust. In the Taupo Volcanic Zone, for example, an average heat flux of 700 mW m22 is an order of magnitude greater than the arc averages, and implies local temperature gradients of up to 300 8C km21 (Bibby et al. 1995). The rate at which magmas migrate to upper crustal levels depends on the rate of magma generation within the mantle wedge, the geometry of the crustal plumbing system and the dynamics of magma ascent through the arc crust. A priori, these parameters are difficult to constrain. However, mantle melting occurs in the mantle wedge in response to fluid release from the subducted slab (e.g. Gill 1981; Arculus 1994). Hence, to a first order the rate of magma generation is a function of the rate of fluid release into the wedge, which in turn is controlled by the degree of hydration of the subducting slab and the rate at which it descends into the mantle, i.e. the plate

23

convergence rate. While the fertility of the mantle wedge, its thermal properties as for example controlled by the thickness of the overriding plate, and the fluid content of the subducting slab may also influence the rate of melt generation to some degree, melt production rates have been shown to increase with increasing plate convergence rates in twodimensional models that include solid mantle flow and associated temperature distributions along with buoyant fluid migration and melting (Cagnioncle et al. 2007). In the following, the relative contributions of variations in crustal thickness and plate convergence rate to the observed anticorrelation of average lava viscosity with average surface heat flux in regular arcs are evaluated, providing insights into magma dynamics at arcs.

Effects of crustal thickness and convergence rate Although it should be noted from Figure 2 that the lava dome proportion is generally less than 40% at arcs situated on crust with a thickness of less than c. 25 km, the observed general scatter suggests that there is no coherent relationship between average lava viscosity and average crustal thickness in the 29 arc segments studied here. Thus, a thick overriding crust may be a necessary but is not a sufficient condition of abundant lava dome extrusion. This indicates that crustal thickness, and therefore the background geothermal gradient, does not exert a strong control on average lava viscosity at arcs. In contrast, Figure 3 shows a good anticorrelation (R 2 ¼ 0.71) between average lava viscosity and average plate convergence rate for all regular and some irregular arcs. Parameters other than plate convergence rate, such as the degree of hydration of the downgoing slab and the geometry of the crustal plumbing system, may introduce additional uncertainties not accounted for in this analysis, which may be the reason for the relatively high MSWD of 9.5. However, the good anticorrelation does indicate that the rate of magma production in the mantle wedge exerts a first-order control on average lava viscosity at arcs. Taken together, these observations indicate the following for regular arcs: 1.

The average surface heat flux is dominated by heat released from shallow level magma reservoirs. The background geothermal gradients as determined by arc crustal thicknesses, or potential differences in radiogenically produced heat, do not appear to exert a strong control.

24

G. F. ZELLMER

Fig. 4. (a) Global 238U– 230 Th disequilibria in young arc volcanics at the time of eruption. Samples in secular equilibrium plot at (238U– 230 Th)0 ¼ 1. Disequilibria are produced in the mantle (Gill & Williams 1990; Turner et al. 2001) and decay back towards equilibrium within c. 5 half-lives of 230 Th (i.e. within c. 5  75 ka). Dashed lines indicate the amount of vertical movement with time. To ensure reliable eruption age correction, volcanics that erupted

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS

2.

3.

Magma production in the mantle wedge and magma intrusion into shallow magma reservoirs are closely linked. Potential differences between arcs in the degrees of hydration of the subducting slab, the fertility of the mantle wedge, the geometry of the crustal magma plumbing system and the dynamics of magma ascent are second-order phenomena. In arcs with high magma production rates, average lava viscosities are low, resulting in a dominance of lava flows. Where magma production rates are low, average viscosities are higher, resulting in a dominance of lava domes. In the latter case, it may be hypothesized that crustal magma transfer timescales are prolonged, resulting in extrusion of cooler and more viscous lavas. In the following, geochemical evidence in support of this hypothesis is presented.

Evidence from geochronological data Recent U-series data of arc volcanic products have provided insights into the rates of magma ascent and evolution at arcs (Turner et al. 2001; Zellmer et al. 2005), but have not addressed links between eruptive style and extent of observed U-series disequilibria. In Figure 4, all available 238U – 230Th data of 75 ka old arc volcanics have been compiled (Reagan et al. 2003; Zellmer et al. 2005, and references therein; Zellmer & Turner 2007), although most of the data represent volcanic products of Holocene age or from historic eruptions. To ensure accuracy, samples older than 75 ka (¼one half-life of 230Th) are not considered here because of the large age corrections that have to be applied to calculate their eruptive (238U – 230Th) activity ratios. Eruptive style was constrained from the literature, through pers. comm. (2006), or in the case of known eruption ages through the GVP Holocene eruption database (Simkin & Siebert 2002). It is evident from Figure 4a that there is a general decrease in 238U– 230Th disequilibria from basaltic andesites towards andesites and dacites, which has previously been attributed to aging (Reagan

25

et al. 2003) and differentiation processes (Garrison et al. 2006) of the magmas in the crust. However, it is also apparent that 90% of the dome lavas, including those with less evolved compositions, are within 10% of 238U – 230Th equilibrium. In comparison, about one-third of other similarly differentiated (SiO2 . 58 wt%) eruptive products display greater than 10% 238U– 230Th disequilibrium (Figure 4b). There are three possible explanations in principle: firstly, magmas that erupt as lava domes may never have acquired significant 238 U – 230Th disequilibria; secondly, processes that generate 238U excesses have been balanced by processes that generate 230Th excesses during the petrogenesis of these rocks; or thirdly, initial disequilibria of lava domes have decayed during preeruptive magma storage in the crust. However, given that arc magmatism is characterized by fluid-induced melting of the mantle wedge, which generates 238U excesses (e.g. Gill & Williams 1990; Turner et al. 2001), and given that there is no evidence for principle geochemical differences between dome lavas and other arc eruptive products, it is difficult to argue that the magmas erupting as lava domes have always been close to 238 U – 230Th equilibrium. This leaves processes that generate 230Th excesses and pre-eruptive aging as possible explanations for the low disequilibria displayed by dome lavas. For very evolved samples, fractionation of accessory phases such as zircon may produce 230Th excess, either due to fractional crystallization (e.g. Condomines 1997) or due to small degree partial melting, e.g. within a lower crustal hot zone (Annen et al. 2006; Zellmer & Turner 2007), although the latter requires long thermal incubation times of the order of 104 –106 years. However, in andesitic samples, accessory phases that may fractionate U and Th are generally not stable. Therefore, the simplest interpretation of the 238U – 230Th data is that lava dome materials have on average significantly (of the order of 104 –105 years) longer crustal residence times than other intermediate arc samples. These observations argue for a link between lava viscosity and crustal residence time, suggesting that viscosity increases as a function of cooling (and concomitant crystallization) in crustal

Fig. 4. (Continued) more than 75 ka ago have been excluded, and most erupted during the Holocene. Samples from lava dome eruptions are plotted as black triangles (shaded field, unlabelled data from Montserrat, Lesser Antilles); all other eruptive products are given as grey circles. Their eruptive style is often unknown, and although most are probably lavas, it is not possible to preclude that some may be tephras. (b) Histogram of 238U– 230Th disequilibria at the time of eruption. Dashed line indicates secular equilibrium. Ninety per cent of the dome samples display less than 10% 238 U– 230Th disequilibrium. In comparison, about one-third of similarly evolved arc eruptive products show greater than 10% 238U – 230Th disequilibrium.

26

G. F. ZELLMER

magma reservoirs. The hypothesis of longer crustal magma transfer times at arcs with lower magma production rates is therefore supported by the geochemical data.

Inferences on magma migration processes from irregular arcs Irregular arcs do not show a simple relationship between average lava viscosity and average surface heat flux, but instead display elevated heat flux values (Fig. 1). With exception of CAS and NEJ they do, however, follow the anticorrelation of average lava viscosity and convergence rate (Fig. 3). These observations may provide additional constraints on magma transfer processes and are therefore discussed below for (1) arcs within zones of crustal extension, (2) arcs within crustal shear zones, (3) arcs within thrust zones, and (4) arcs with discontinuities in the subducting slab. 1.

2.

NZL lies on the boundary between the Australian and the Kermadec plate and is therefore the only arc on an active continental rift boundary as defined by Bird (2003). Rifting leads to crustal thinning and mantle upwelling, thereby increasing the background geothermal gradient and the surface heat flux, consistent with Figure 1. However, no concomitant lowering of average lava viscosity is observed, and average convergence rate appears to remain a controlling factor on lava viscosity (Fig. 3) and therefore the timescale of magma transfer from wedge to upper crust. This suggests that magma production rates are not significantly changed by the rifting process. It may be speculated that at NZL mantle melting is limited by the amount of fluids available; that enhanced mantle upwelling due to rifting in the overriding plate may only have a second-order effect on melt generation; and that crustal magma transfer rates are not increased by the existence of extensional faults, which are typically not vertical and therefore may not provide as suitable magma ascent pathways as, for example, shear zones. There is both geological and geodetic evidence that CAS lies within a right-lateral shear zone along which the Cascadia Margin is displaced northward with respect to North America at a rate of up to 9 mm year21 (Miller et al. 2001). The average surface heat flux at CAS is only slightly elevated (Fig. 1), which is attributed to a slab discontinuity in the south (i.e. the edge of the subducting Juan de Fuca plate), where heat flux values are highest at Lassen and Mount Shasta. However, the

3.

4.

average lava viscosity at CAS is very low given the low plate convergence rate (Fig. 3), suggesting that the rate of crustal magma transfer is significantly elevated compared with regular arcs. It is therefore hypothesized that near-vertical faulting within the crustal shear zone provides efficient magma transport pathways that allow rapid transfer of melts from mantle wedge to upper crust. This interpretation is consistent with the north – south alignment of volcanic vents seen in many parts of the High Cascades (Hildreth 2008). While thrusting is known to occur within mobile mountain belts at back-arcs (e.g. Hyndman et al. 2005), within-arc convergence appears to be more uncommon. NEJ is an example of within-arc thrusting as identified through geological and geodetic studies (Seno 1999; Townend & Zoback 2006). The elevation in average surface heat flux of NEJ (Fig. 1) may be attributed to discontinuities in the subducting slab at the northern and southern ends of this arc segment (see below). However, considering the high average convergence rate at NEJ, one would expect much lower average lava viscosities than observed (Fig. 3). The relatively high proportion of lava domes may indicate that crustal magma transport rates are decreased due to a compressive stress field that impedes the opening of magmatic conduits. Slab contortion, cracking or tearing, slab window formation and slab detachment will have variably profound effects on the temperature distribution of the mantle wedge. In most cases, the wedge will experience a lower degree of conductive cooling from the subducting slab relative to the regular subduction scenario, and in the case of large slab discontinuities, upwelling of asthenospheric mantle may significantly elevate upper mantle temperatures locally or regionally. This may explain an elevated surface heat flux as observed in Figure 1 for CAS, MEX, NCH, NEJ and SWJ. However, with the exception of CAS and NEJ as discussed above, average convergence rate remains a controlling factor in determining average lava viscosity and therefore the timescale of magma transfer from wedge to upper crust (Fig. 3). It appears that slab discontinuities do not significantly influence the rate of arc magma production. It may again be speculated that mantle melting is limited by the amount of fluids available, and that a possibly elevated wedge temperature may only have second order effects on the rate of melt generation.

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS

Concluding remarks Focusing on effusive eruptions at volcanic arcs, the data presented here indicate that the average viscosity of arc lavas is highly variable between different arcs. Average viscosity anticorrelates well with weighted average surface heat flux at continental and transitional arcs formed on plates with convex margins and little intra-arc crustal deformation. Arcs situated within zones of crustal deformation or above discontinuities in the subducting slab show elevated surface heat flux values compared with most other arcs. Crustal thickness appears to have little control over average lava viscosity. In contrast, average plate convergence rate anticorrelates well with average lava viscosity for most arcs. Further, viscous dome lavas are on average significantly closer to 238U– 230Th equilibrium than other arc eruptive products. It has been shown here that these first-order observations imply the following: 1.

2.

3.

4.

The rate of magma generation at arcs is to a first-order controlled by the rate of fluid release into the mantle wedge. Asthenospheric upwelling or wedge temperature variations may have second order effects. The rate of magma transfer from mantle wedge to upper crust is to a first-order controlled by the rate of melt generation. Where the rate of melt generation is high, magma transfer is fast, resulting in low average lava viscosity. Where the rate of melt generation is low, magma transfer is slow, with on average 104 –105 years longer transfer times required for the generation of viscous lava domes through cooling and concomitant crystallization of magmas in crustal reservoirs prior to eruption. This also implies that, if a lower crustal hot zone is invoked (Annen et al. 2006), it must be close to steady state, i.e. magmas are released from the zone into the upper crust at a rate approximately proportional to the rate they enter the zone from the mantle wedge. By comparison, crustal thickness does not show a coherent variation with average lava viscosity, suggesting that variations in crustal thickness may only have a second-order influence on average magma transfer timescales. Average lava viscosity is comparatively low in the Cascades, which lie within a regional shear zone, and comparatively high in NE Japan, which is an area of thrusting and active mountain building. Interestingly, intra-arc rifting such as observed in New Zealand does not appear to have first-order effects on magma transfer times, possibly due to a lack of near-vertical faults that would facilitate magma ascent.

27

Numerous researchers provided information on the eruption style of arc samples. J. Blundy, C. Wright and Y. Iizuka are thanked for fruitful discussions, and O. Bachmann and A. Dosseto for constructive reviews. Editorial comments by C. Annen improved the manuscript. As Lamont Adjunct Associate Research Scientist, the author acknowledges the electronic resources made available through Columbia University. Funding was provided by Academia Sinica and the National Science Council of Taiwan (NSC 95-2116-M-001-006).

Appendix: Holocene effusive arc volcanism at global subduction zones AEG, Aegean Volcanic Arc, Greece: formed due to subduction of the Africa plate beneath the Aegean Sea plate. Volcanoes with Holocene effusive eruptions: Methana, Santorini. AEO, Aeolian Volcanic Arc, Italy: formed due to subduction of the Africa plate beneath the Eurasia plate. Bird (2003) models this area as part of the Alps orogen. Thus, the convergence rate of 60 mm year21 adopted here is instead taken from a regional study of the Ionian slab (Pasquale et al. 2005). Volcanoes with Holocene effusive eruptions: Stromboli, Lipari, Vulcano. AKP, Alaska Peninsula: formed due to subduction of the Pacific plate beneath the North America plate. Volcanoes with Holocene effusive eruptions: Pavlof, Veniaminof, Aniakchak, Yantarni, Ukinrek Maars, Trident. ALU, Aleutian Arc: formed due to subduction of the Pacific plate beneath the North America plate. The average convergence rate of about 62 mm year21 excludes the westernmost part of the arc where right-lateral slip dominates. Volcanoes with Holocene effusive eruptions: Kiska, Little Sitkin, Gareloi, Tanaga, Kanaga, Great Sitkin, Atka, Seguam, Amukta, Cleveland, Okmok, Makushin, Akutan, Westdahl, Shishaldin, Isanotski; and Bogoslov and Amak behind the volcanic front. ANT, Lesser Antilles Volcanic Arc: formed due to subduction of the South America plate beneath the Caribbean plate. Volcanoes with Holocene effusive eruptions: Soufrie`re Hills, Soufrie`re Guadeloupe, Morne Trois Pitons, Morne Plat Pays, Pele´e, Soufriere St Vincent, Kick ’em Jenny. BIS, Bismarck/New Britain Arc: formed due to subduction of the Solomon Sea plate beneath the South Bismarck plate. BIS extends with its western part to the area north of the continental convergent boundary between the Woodlark and South Bismarck plates. The average convergence rate of about 89 mm year21 is based on the eastern part of BIS, where active subduction is occuring. Volcanoes with Holocene effusive eruptions: Manam, Karkar, Long Island, Langila, Pago, Hargy, Ulawun, Rabaul; and Dakataua and Lolobau behind the volcanic front. CAM, Central American Arc, including Southern Mexico: formed due to subduction of the Cocos plate beneath the North America, Caribbean and Panama

28

G. F. ZELLMER

plates. Volcanoes with Holocene effusive eruptions: San Martı´n, Tacana´, Santa Marı´a, Almolonga, Atitla´n, Fuego, Pacaya, Santa Ana, Izalco, San Salvador, Ilopango, San Miguel, Cosigu¨ina, Telica, Cerro Negro, Momotombo, Masaya, Concepcio´n, Arenal, Poa´s, Irazu´. A single lava flow has also been recorded about 370 km from Irazu´ at La Yeguada in Panama, but has not been included here. CAS, Cascades Volcanic Arc: formed due to subduction of the Juan de Fuca plate beneath the North America plate. Volcanoes with Holocene effusive eruptions: Baker, Glacier Peak, Adams, St Helens, West Crater, Indian Heaven, Hood, Jefferson, Sand Mountain Field, Belknap, North Sister, South Sister, Bachelor, Davis Lake, Crater Lake, Shasta, Lassen; and Newberry and Medicine Lake behind the volcanic front. Surface heat flux varies by a factor of 1.4 along the volcanic arc. COL, Colombia: formed due to subduction of the Nazca plate beneath the North Andes plate. Volcanoes with Holocene effusive eruptions: Cerro Bravo, Santa Isabel, Purace´, Don˜a Juana, Galeras, Azufral. CVA, Central Vanuatu/New Hebrides: formed due to subduction of the Australia plate beneath the New Hebrides plate. Volcanoes with Holocene effusive eruptions: Aoba, Ambrym, Lopevi. Note that few recent effusive eruptions also occurred at Tinakula in the Santa Cruz Islands, but these have not been included due to the much lower surface heat flux in this area north of the New Hebrides Plate. Also excluded is the single lava lake eruption recorded 350 km further south at Yasur, where convergence rates are significantly higher than in the central section of the arc. ECU, Ecuador: formed due to subduction of the Nazca plate beneath the North Andes plate. Volcanoes with Holocene effusive eruptions: Soche, Cuichoca, Cayambe, Reventador, Pululagua, Guagua Pichincha, Atacazo, Chacana, Antisana, Cotopaxi, Quilotoa, Tungurahua, Sangay. EJV, Eastern Java: from Sumatra to the Lesser Sunda islands, the Australia Plate subducts beneath the Sunda plate. EJV forms the western part of the Deformed Sundaland of Rangin et al. (1999). To the West, it is divided by a regional fault zone from Western Java (WJV), which displays lower average surface heat flux and is part of the Undeformed Sundaland of Rangin et al. (1999). To the East, it is divided by the Bali Strait from the Lesser Sunda Arc (LSU), which also displays lower average surface heat flux. Volcanoes with Holocene effusive eruptions: Kelut, Semeru, Tengger Caldera, Lamongan, Raung. HAL, Halmahera Arc: formed due to subduction of the Molucca Sea plate beneath the Birds Head plate. The average convergence rate of about 69 mm year21 is based on a total average convergence of 103.3 mm year21 across the Molucca Sea, which is modelled to be taken up by the Sangihe and Halmahera arcs in 1:3 to 2:3 proportion, as suggested by an average volcano spacing of about 32 km in the Sangihe/North Sulawesi arc compared with about

16 km in the Halmahera arc. Volcanoes with Holocene effusive eruptions: Dukono, Ibu, Gankonora, Gamalama, Makian. KAM, Kamchatka and Northern Kurile: from Northern Japan through Kurile to Kamchatka, the Pacific plate subducts beneath the Okhotsk plate. KAM is the northern section of the arc, divided from the southern section with lower average surface heat flux (SKU) by a .75 km gap in the volcanic front. Volcanoes with Holocene effusive eruptions: in the Central Kamchatka Depression, Shiveluch, Kliuchevskoi, Bezymianny, Tolbachik, Kizimen; in the Eastern Volcanic Front, Vysoky, Komarov, Gamchen, Krasheninnikov, Kikhpinych, Taunshits, Bolshoi Semiachik, Maly Semiachik, Karymsky, Akademia Nauk, Zavaritsky, Bakening, Kostakan, Veer, Avachinsky, Koryaksky, Vilyuchik, Tolmachev Dol, Opala, Gorely, Mutnovsky, Khodutka, Ksudach, Zheltovsky, Ilyinsky, Kurile Lake, Diky Greben; in the northern Kurile arc, Alaid, Chikurachki, Nemo Peak, Tao-Rusyr Caldera, Kharimkotan, Sinarka, Ekarma, Chirinkotan. LSU, Lesser Sunda Arc: from Sumatra to the Lesser Sunda islands, the Australia Plate subducts beneath the Sunda plate. LSU forms the eastern part of that arc. To the west, it is devided by the Bali Strait from Eastern Java (EJV), which displays higher average surface heat flux. LSU extends with its eastern part to the area north of the continental convergent boundary between the Australia and Timor plates. The average convergence of about 67 mm year21 is based on the western part of LSU, where active subduction is occuring. Volcanoes with Holocene effusive eruptions: Batur, Agung, Rinjani, Tambora, Sangeang Api, Ranakah, Ebulobo, Lewotobi, Iliboleng, Iliwerung; and Paluweh and Batu Tara behind the volcanic front. MAR, Mariana Arc: formed due to subduction of the Pacific plate beneath the Mariana plate. Volcanoes with Holocene effusive eruptions: Farallon de Pajaros, Asuncion, Pagan, Guguan, Anatahan. MEX, Mexican Volcanic Belt: formed due to subduction of the Cocos plate beneath the North America plate. Volcanoes with Holocene effusive eruptions: Ceboruco, Colima, Michoaca´n– Guanajuato, Zita´cuaro– Valle de Bravo, Nevado de Toluca, Chichinautzin, Popocate´petl, La Malinche, Pico de Orizaba, Las Cumbres, Cofre de Perote, Naolinco Volcanic Field. Surface heat flux increases by a factor of 1.7 from east to west along the volcanic arc. NCH, Northern Chile: formed due to subduction of the Nazca plate beneath the Altiplano and North America plates. Volcanoes with Holocene effusive eruptions: Parinacota, Isluga, Lascar. Surface heat flux increases by a factor of 1.3 from north to south along the volcanic arc. NEJ, NE Japan: formed due to subduction of the Pacific plate beneath the Okhotsk plate. Volcanoes with Holocene effusive eruptions: on the Amur plate, Haku-san, Yake-dake; on the Okhotsk plate, NiigataYake-yama, Asama, Kusatsu-Shirane, Haruna, Hiuchi,

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS Takahara, Nasu, Azuma, Chokai, Akita-Komaga-take, Iwate, Towada, Oshima-Oshima, Komaga-take, Usu, Kuttara, Shikotsu. Surface heat flux varies by a factor of 1.7 along/across the volcanic arc, some sections of which extend 100 km or more from arc front to back. NIZ, Northern Izu Arc: formed due to subduction of the Pacific plate beneath the Philippine Sea and Okhotsk plates. Volcanoes with Holocene effusive eruptions: Izu-Tobu, Hakone, Fuji, Oshima, Nii-jima, Kozu-shima, Miyake-jima. Note that few recent effusive eruptions also occur 180–400 km further south, at Aoga-shima, Bayonnaise Rocks, and Tori-shima, but these have not been included due to the much lower plate convergence rate in this area compared with the northernmost section of the Izu arc. NZL, New Zealand: formed due to subduction of the Pacific plate beneath the Kermadec plate. The volcanic arc lies within the continental rift boundary, between the Kermadec plate and the Australia Plate (Bird 2003). Volcanoes with Holocene effusive eruptions: Okataina, Taupo, Tongariro, Ruapehu; and Major Island and Taranaki behind the arc front. PHL, Philippine Arc: formed due to subduction of the Philippine Sea plate beneath the Sunda plate. Volcanoes with Holocene effusive eruptions: Pinatubo, Taal, Mayon, Bulusan, Canlaon, Camiguin. SAN, Sangihe and North Sulawesi: formed due to subduction of the Molucca Sea plate beneath the Sunda plate. The average convergence rate of about 34 mm year21 is based on a total average convergence of 103.3 mm year21 across the Molucca Sea, which is modelled to be taken up by the Sangihe and Halmahera arcs in 1:3 to 2:3 proportion, as suggested by an average volcano spacing of about 32 km in the Sangihe/North Sulawesi arc compared with about 16 km in the Halmahera arc. Volcanoes with Holocene effusive eruptions: Awu, Banua Wuhu, Karangetang, Ruang, Tongkoko, Lokon-Empung, Soputan. SCH, Southern Chile: formed due to subduction of the Nazca plate beneath the South America plate. Volcanoes with Holocene effusive eruptions: Tupungatito, Maipo, Plancho´n-Peteroa, Cerro Azul, Nevados de Chilla´n, Antuco, Lonquimay, Llaima, Sollipulli, Villarrica, Lanı´n, Huanquihue Group, Carra´n-Los Venados, Cordo´n Caulle, Osorno, Calbuco, Huequi, Minchinma´vida, Mentolat, Cerro Hudson. SCO, Scotia/South Sandwich Arc: formed due to subduction of the South America plate beneath the Sandwich plate. Volcanoes with Holocene effusive eruptions: Bristol Island, Montagu Island, Michael. SKU, Southern Kurile: from Northern Japan through Kurile to Kamchatka, the Pacific plate subducts beneath the Okhotsk plate. SKU is the southern section of the arc, divided from the northern section with higher average surface heat flux (KAM) by a .75 km gap in the volcanic front. Volcanoes with Holocene effusive eruptions: in Hokkaido, Tokachi, Akan, Kutcharo; in the southern Kurile arc, Mendeleev, Tiatia, Medvezhia,

29

Kolokol Group, Chirpoi, Goriaschaia Sopka, Zavaritzki Caldera, Ketoi, Ushishur, Sarychev. SOL, Solomon Arc: formed due to subduction of the Solomon Sea plate beneath the North Bismarck plate in the northern section, and of the Solomon Sea, Woodlark and Australia plates beneath the Pacific plate in the southern section of the arc. Volcanoes with Holocene effusive eruptions: Bagana, Kavachi, Savo. Surface heat flux increases by a factor of 1.25 from north to south along the volcanic arc. SWJ, Southwest Japan and Northern Ryukyu: formed due to subduction of the Philippine Sea plate beneath the Amur and Okinawa plates. Volcanoes with Holocene effusive eruptions: in Kyushu, Tsurumi, Kuju, Aso, Kirishima, Sakura-jima, Ibuzuki and Unzen behind the volcanic front; in the northern Ryukyu arc, Kikai, Suwanose-jima. WJV, Western Java: from Sumatra to the Lesser Sunda islands, the Australia Plate subducts beneath the Sunda plate. WJV lies east of the Ninety East-Sumatra orogen and is part of the Undeformed Sundaland of Rangin et al. (1999). To the east, it is divided by a regional fault zone from Eastern Java (EJV), which displays higher average surface heat flux and is part of the Deformed Sundaland of Rangin et al. (1999). Volcanoes with Holocene effusive eruptions: Krakatau, Guntur, Galunggung, Slamet, Dieng, Sundoro, Merapi.

References A LTAMIMI , Z., S ILLARD , P. & B OUCHER , C. 2002. ITRF2000: A new release of the International Terrestrial Reference Frame for earth science applications. Journal of Geophysical Research, 107, 2214, DOI: 10.1029/2001JB000561. A NNEN , C., B LUNDY , J. D. & S PARKS , R. S. J. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology, 47, 505– 539. A RCULUS , R. J. 1994. Aspects of magma genesis in arcs. Lithos, 33, 189– 208. A RGUS , D. F. & G ORDON , R. G. 1991. No-net-rotation model of current plate velocities incorporating plate motion model NUVEL-1. Geophysical Research Letters, 18, 2039– 2042. B AKER , M. B., G ROVE , T. L. & P RICE , R. C. 1994. Primitive basalts and andesites from the Mt. Shasta region, N. California; products of varying melt fraction and water content. Contributions to Mineralogy and Petrology, 118, 111–129. B ASSIN , C., L ASKE , G. & M ASTERS , T. G. 2000. The current limits of resolution for surface wave tomography in North America. EOS Transactions of the AGU, 81, F897; available at http://mahi.ucsd.edu/Gabi/ rem.html. B IBBY , H. M., C ALDWELL , T. G., D AVEY , F. J. & W EBB , T. H. 1995. Geophysical evidence on the structure of the Taupo Volcanic Zone and its hydrothermal circulation. Journal of Volcanology and Geothermal Research, 68, 29–58.

30

G. F. ZELLMER

B IRD , P. 2003. An updated digital model of plate boundaries. Geochemistry Geophysics Geosystems, 4, 1– 52, DOI: 10.1029/2001GC000252. B ORG , L. E., C LYNNE , M. A. & B ULLEN , T. D. 1997. The variable role of slab-derived fluids in the generation of a suite of primitive calc-alkaline lavas from the southernmost Cascades, California. Canadian Mineralogist, 35, 425 –452. B URNHAM , C. W. 1975. Water and magmas; a mixing model. Geochimica et Cosmochimica Acta, 39, 1077–1084. C AGNIONCLE , A.-M., P ARMENTIER , E. M. & E LKINS -T ANTON , L. T. 2007. Effect of solid flow above a subducting slab on water distribution and melting at convergent plate boundaries. Journal of Geophysical Research, 112, DOI: 10.1029/ 2007JB004934. C HASE , C. G. 1978. Plate kinematics: the Americas, East Africa, and the rest of the world. Earth and Planetary Science Letters, 37, 355–368. C ONDOMINES , M. 1997. Dating recent volcanic rocks through 230Th– 238U disequilibrium in accessory minerals: Example of the Puy de Dome (French Massif Central). Geology, 25, 375– 378. C OSTA , A. 2005. Viscosity of high crystal content melts: dependence on solid fraction. Geophysical Research Letters, 32, DOI: 10.1029/2005GL024303. D E B ARI , S. M. & C OLEMAN , R. G. 1989. Examination of the deep levels of an island arc: Evidence from the Tonsina ultramafic-mafic assemblage, Tonsina, Alaska. Journal of Geophysical Research, 94, 4373–4391. D E B ARI , S. M., K AY , S. M. & K AY , R. W. 1987. Ultramafic xenoliths from Adagdak volcano, Adak, Aleutian Islands, Alaska: deformed igneous cumulates from the Moho of an island arc. Journal of Geology, 95, 329 –341. D E M ETS , C., G ORDON , R. G., A RGUS , D. F. & S TEIN , S. 1994. Effect of recent revisions to the geomagnetic reversal time scale on estimates of current plate motions. Geophysical Research Letters, 21, 2191–2194. D RURY , M. 1987. Heat flow provinces reconsidered. Physics of the Earth and Planetary Interiors, 49, 78–96. E LKINS T ANTON , L. T., G ROVE , T. L. & D ONNELLY N OLAN , J. 2001. Hot, shallow mantle melting under the Cascades volcanic arc. Geology, 29, 631–634. F ERRARI , L. 2004. Slab detachment control on mafic volcanic pulse and mantle heterogeneity in central Mexico. Geology, 32, 77–80, DOI: 10.1130/ G19887.1. F LIEDNER , M. M. & K LEMPERER , S. L. 1999. Structure of an island-arc: Wide-angle seismic studies in the eastern Aleutian Islands, Alaska. Journal of Geophysical Research, 104, 10667–10694. G ARRISON , J., D AVIDSON , J, R EID , M. & T URNER , S. 2006. Source versus differentiation controls on U-series disequilibria: Insights from Cotopaxi Volcano, Ecuador. Earth and Planetary Science Letters, 244, 548 –565. G ILL , J. B. 1981. Orogenic Andesites and Plate Tectonics, Springer, Heidelberg. G ILL , J. B. & W ILLIAMS , R. W. 1990. Th isotope and U-series studies of subduction-related volcanic

rocks. Geochimica et Cosmochimica Acta, 54, 1427– 1442. G IORDANO , D. & D INGWELL , D. W. 2003. NonArrhenian multicomponent melt viscosity: a model. Earth and Planetary Science Letters, 208, 337– 349. G ORRING , M. L. & K AY , S. M. 2001. Mantle processes and sources of Neogene slab window magmas from southern Patagonia, Argentina. Journal of Petrology, 42, 1067– 1094. H ILDRETH , W. 2008. Quaternary magmatism in the Cascades - Geologic perspectives. U. S. Geological Survey Professional Paper, 1744, 125 pp. H OLBROOK , W. S., L IZARRALDE , D., M C G EARY , S., B ANGS , N. & D IEBOLD , J. 1999. Structure and composition of the Aleutian island arc and implications for continental crustal growth. Geology, 27, 31–34. H YNDMAN , R. D., C URRIE , C. A. & M AZZOTTI , S. P. 2005. Subduction zone backarcs, mobile belts, and orogenic heat. GSA Today, 15, 4–10, DOI: 10:1130/ 1052-5173(2005)015. I SHIDA , M. 1992. Geometry and relative motion of the Philippine Sea plate and Pacific plate beneath the Kanto– Tokai district, Japan. Journal of Geophysical Research, 97, 489– 513. J ARRARD , R. 1986. Relations among subduction parameters. Reviews of Geophysics, 24, 217– 284. J OHNSTON , S. T. & T HORKELSON , D. J. 1997. CocosNazca slab window beneath Central America. Earth and Planetary Science Letters, 146, 465– 474. K ATSUMATA , K., W ADA , N. & K ASAHARA , M. 2003. Newly imaged shape of the deep seismic zone within the subducting Pacific plate beneath the Hokkaido corner, Japan– Kurile arc-arc junction. Journal of Geophysical Research, 108, 2565, DOI: 10.1029/ 2002JB002175. K ELEMEN , P. B., P ARMENTIER , E. M., R ILLING , J., M EHL , L. & H ACKER , B. R. 2003. Thermal convection in the mantle wedge beneath subduction-related magmatic arcs. American Geophysical Union Monograph, 138, 293– 311. K IMURA , J.-I., T ATENO , M. & O SAKA , I. 2005. Geology and geochemistry of Karasugasen lava dome, Daisen– Hiruzen volcano group, southwest Japan. The Island Arc, 14, 115– 136. K REEMER , C., H OLT , W. E. & H AINES , A. J. 2003. An integrated global model of present-day plate motions and plate boundary deformation. Geophysical Journal International, 154, 8–34. L EEMAN , W. P. 1983. The influence of crustal structure on compositions of subduction-related magmas. Journal of Volcanology and Geothermal Research, 18, 561–588. L EJEUNE , A. & R ICHET , P. 1995. Rheology of crystalbearing silicate melts: an experimental study at high viscosity. Journal of Geophysical Research, 100, 4215– 4229. L IN , J.-Y., H SU , S.-K. & S IBUET , J.-C. 2004. Melting features along the Ryukyu slab tear, beneath the southwestern Okinawa Trough. Journal of Geophysical Research, 31, L19607, DOI: 10.1029/2004GL020862. L IZARRALDE , D., H OLBROOK , W. S., M C G EARY , S., B ANGS , N. L. & D IEBOLD , J. B. 2002. Crustal construction of a volcanic arc, wide-angle seismic results from the western Alaska Peninsula. Journal

VOLCANIC ARCS: FIRST-ORDER OBSERVATIONS of Geophysical Research, 107, 2164, DOI: 10.1029/ 2001JB000230. L UDWIG , K. R. 2003. Isoplot/Ex ver. 3, A Geochronological Toolkit for Microsoft Excel. Berkeley Geochronology Center, Berkeley, CA. L UNDGREN , P. R. & G IARDINI , D. 1990. Lateral structure of the subducting Pacific plate beneath the Hokkaido corner from intermediate and deep earthquakes. Pure and Applied Geophysics, 134, 385–404. M AZZOTTI , S., H ENRYA , P., L E P ICHON , X. & S AGIYAC , T. 1999. Strain partitioning in the zone of transition from Nankai subduction to Izu–Bonin collision (Central Japan): implications for an extensional tear within the subducting slab. Earth and Planetary Science Letters, 172, 1 –10. M ILLER , M. M., J OHNSON , D. J., R UBIN , C. M., D RAGERT , H., W ANG , K., Q AMAR , A. & G OLDFINGER , C. 2001. GPS-determination of alongstrike variation in Cascadia margin kinematics; implications for relative plate motion, subduction zone coupling, and permanent deformation. Tectonics, 20, 161–176. M INISTER , J. B. & J ORDAN , T. H. 1978. Present-day plate motions. Journal of Geophysical Research, 83, 5331–5354. M OONEY , W. D., L ASKE , G. & M ASTERS , T. G. 1998. Crust 5.1: a global crustal model at 5  5 degrees. Journal of Geophysical Research, 103, 727–747. M ORRIS , P. A. 1995. Slab melting as an explanation of Quaternary volcanism and aseismicity in southwest Japan. Geology, 23, 395– 398. P ASQUALE , V., V ERDOYA , M. & C HIOZZI , P. 2005. Thermal structure of the Ionian slab. Pure and Applied Geophysics, 162, 962–986. P INKERTON , H. & S TEVENSON , R. J. 1992. Methods of determining the rheological properties of magmas at sub-liquidus temperatures. Journal of Volcanology and Geothermal Research, 53, 47–66. P OLLACK , H. N. 1982. The heat flow from the continents. Annual Reviews in Earth and Planetary Sciences, 10, 459–481. R ANGIN , C., L E P ICHON , X., M AZZOTTI , S., P UBELLIER , M., C HAMOT -R OOKE , N., A URELIO , M., W ALPERSDORF , A. & Q UEBRAL , R. 1999. Plate convergence measured by GPS across the Sundaland/Philippine Sea Plate deformed boundary: the Philippines and eastern Indonesia. Geophysical Journal International, 139, 296– 316. R EAGAN , M. K., S IMS , K. W. W., E RICH , J., T HOMAS , R. B., C HENG , H., E DWARDS , R. L., L AYNE , G. & B ALL , L. 2003. Timescales of differentiation from mafic parents to rhyolite in North American continental arcs. Journal of Petrology, 44, 1703–1726. R IETBROCK , A., H ABERLAND , C. & N IPPRESS , S. 2006. A tear in the subducting Nazca slab at 21 S revealed from accurate locations of intermediate depth seismicity. Eos Transactions of the AGU, 87, Fall Meeting Supplement, Abstract S43D-08. S ENO , T. 1999. Syntheses of the regional stress fields of the Japanese islands. The Island Arc, 8, 66–97. S HAPIRO , N. M. & R ITZWOLLER , M. H. 2004. Inferring surface heat flux distributions guided by a global

31

seismic model: particular application to Antarctica. Earth and Planetary Science Letters, 223, 213–224. S HILLINGTON , D. J., V AN A VENDONK , H. J. A., H OLBROOK , W. S., K ELEMEN , P. B. & H ORNBACH , M. J. 2004. Composition and structure of the central Aleutian island arc from arc-parallel wide-angle seismic data. Geochemistry, Geophysics, Geosystems, 5, Q10006, DOI: 10.1029/2004GC000715. S IMKIN , T & S IEBERT , L. 2002. Volcanoes of the World: an Illustrated Catalog of Holocene Volcanoes and their Eruptions. Smithsonian Institution, Global Volcanism Program, Digital Information Series, GVP-3, http://www.volcano.si.edu/world. S PARKS , R. S. J., S IGURDSSON , H. & W ILSON , L. 1977. Magma mixing: a mechanism for triggering acid explosive eruptions. Nature, 267, 315– 318. S UGIMOTO , T., S HIBATA , T., Y OSHIKAWA , M. & T AKEMURA , K. 2006. Sr– Nd– Pb isotopic and major and trace element compositions of the Yufu– Tsurumi volcanic rocks: implications for the magma genesis of the Yufu– Tsurumi volcanoes, northeast Kyushu, Japan. Journal of Mineralogical and Petrological Sciences, 101, 270–275. T OWNEND , J. & Z OBACK , M. D. 2006. Stress, strain, and mountain building in central Japan. Journal of Geophysical Research, 111, B03411, DOI: 10.1029/ 2005JB003759. T URNER , S., E VANS , P. & H AWKESWORTH , C. 2001. Ultrafast source-to-surface movement of melt at island arcs from 226Ra– 230Th systematics. Science, 292, 1363–1366. V AN A VENDONK , H. J. A., S HILLINGTON , D. J., H OLBROOK , W. S. & H ORNBACH , M. J. 2004. Inferring crustal structure in the Aleutian island arc from a sparse wide-angle seismic data set. Geochemistry, Geophysics, Geosystems, 5, Q08008, DOI: 10.1029/ 2003GC000664. Y OGODZINSKI , G. M., L EES , J. M., C HURIKOVA , T. G., D ORENDORF , F., W OERNER , G. & V OLYNETS , O. N. 2001. Geochemical evidence for the melting of subduction oceanic lithosphere at plate edges. Nature, 409, 500–504. Z ELLMER , G. F. & T URNER , S. P. 2007. Arc dacite genesis pathways: evidence from mafic enclaves and their hosts in Aegean lavas, Lithos, 95, 346 –362, DOI: 10.1016/j.lithos.2006.08.002. Z ELLMER , G. F., A NNEN , C., C HARLIER , B. L. A., G EORGE , R. M. M., T URNER , S. P. & H AWKESWORTH , C. J. 2005. Magma evolution and ascent at volcanic arcs: constraining petrogenetic processes through rates and chronologies. Journal of Volcanology and Geothermal Research, 140, 171 –191. Z HAO , D. & H ASEGAWA , A. 1994. Teleseismic evidence for lateral heterogeneities in the northeastern Japan arc. Tectonophysics, 237, 189– 199. Z HAO , D., M ISHRA , O. P. & S ANDA , R. 2002. Influence of fluids and magma on earthquakes: seismological evidence. Physics of the Earth and Planetary Interiors, 132, 249–267. Z HAO , D., Y INGBIAO , X., W EINS , D. A., D ORMAN , L., H ILDEBRAND , J. & W EBB , S. 1997. Depth extent of the Lau backarc spreading center and its relation to subduction. Science, 278, 254– 257.

Probing Stromboli volcano from the mantle to paroxysmal eruptions CORRADO CIGOLINI, MARCO LAIOLO & SARA BERTOLINO Dipartimento di Scienze Mineralogiche e Petrologiche, Universita` di Torino, Via Valperga Caluso 35, Torino, Italy (e-mail: [email protected]) Abstract: We investigated the plumbing system of Stromboli volcano from the upper mantle to the surface. Thermobarometric estimates indicate that the deeper detected part of the plumbing system is located in the upper mantle, at approximately 34– 24 km depth where, during their ascent, primitive Stromboli basalts (HKCA to shoshonitic) interact with peridotitic materials. In this region magma flow is probably channelled along fracture zones that may converge into a feeder dyke that crosscuts the Moho at about 17 km depth. During their ascent, basaltic magmas will interact with lower crust materials represented by cumulates of earlier Strombolitype basalts at 13–10 km depth. This zone is also the section of the plumbing system where the feeder dyke is entering the chamber. Thermobarometric estimates, obtained by constructing a grid of selected reactions, indicate that current primitive Stromboli basalts equilibrate at 0.3–0.15 GPa and temperatures approaching 1200 8C, and progressively crystallize and degas before being erupted. Crystal size distributions on lavas and juvenile tephra erupted in 2002– 2003 give very variable residence times. Based on average bubble distances, the estimated times for the exsolution of the gaseous phases range from 2 –7 days to 45 min for the lavas and scorias, down to about 15 h to 12 min for the pumices erupted during paroxysmal explosions. Estimated syneruptive viscosities range from 102 Pa s for the anhydrous basaltic pumices at 1200 8C, to 103 – 104 Pa s for lavas approaching their effusion temperatures (1100–1150 8C). In turn, viscosities for the hydrous basaltic melt that led to the formation of the basaltic pumices may be around 10 Pa s or lower. In the light of the above, we discuss the possible shapes and volumes of Stromboli magma chamber by considering a sphere, an ellipsoid (geometrically concordant with the regional stress distribution) and a feeder dyke, the last two being more likely. In the light of volcanological, structural and geophysical data on conduit thickness, we propose an alternative model that takes into account the volumes of recently erupted lavas. This model consists of a convective ellipsoidal magma chamber ‘injected’ by an active feeder dike of undegassed magma of higher temperature, lower density and lower viscosity. This dyke will evolve into a magma column inside the chamber and will separate the reservoir into two lateral, nearly symmetric convective regions. Crystallization would occur preferentially in the proximity of the wallrocks, particularly where the chamber is entering the conduit. The onset of paroxysmal explosions during major effusive cycles may be explained by a drastic increase in the intrusion rates at the base of the chamber that will produce a progressive inflation of the magma column dynamically transferred to the chamber walls. The ceasing of ‘anomalous’ intrusion rates at the base of the chamber, coupled with higher discharge rates, will progressively depressurize the chamber to a critical threshold, until the stress transferred to the walls is dynamically released: at this point the walls themselves will undergo a nearly instantaneous elastic rebound and contract in the attempt to recover their original pre-eruptive geometry. These dynamics will squeeze up portions of the undegassed magma column, triggering a paroxysmal explosion with the ejection of ‘golden pumices’.

Ascent and storage of magmas from the mantle to the earth surface has been a debated issue in recent years. Since the pioneer work of Shaw (1980), essentially focused on the mechanisms of magma transport in relation to the regional structural setting, efforts have been concentrated on evaluating the role of the ‘neutral buoyancy’ that rules the storage of basaltic magmas before they are erupted (e.g. Ryan 1993, 1994). Marsh’s innovative ideas on magma chambers (Marsh 1989) were essentially coeval with the modelling

of these complex magma bodies (Dragoni & Magnanesi 1989). However, recent geophysical research has been concentrated on investigating, by means of seismic tomography, the depth and extension of magma reservoirs as well as their possible geometry (Zollo et al. 1998; Auger et al. 2001). In addition, the segregation of melts from their mantle source has been extensively analysed in the light of U –Th disequilibria to constrain their time of origin and their ascent rates (e.g. Condomines et al. 1982; Sigmarsson 1996, among

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 33– 70. DOI: 10.1144/SP304.3 0305-8719/08/$15.00 # The Geological Society of London 2008.

34

C. CIGOLINI ET AL.

others). Similarly, short-lived isotopes have been used to infer the time associated with the storage of magma batches responsible for a given eruptive cycle (Gauthier & Condomines 1999; Gauthier et al. 2000). Alternatively, studies on crystal size distribution (CSD; e.g. Cashman 1990; Cashman & McConnell 2005) provided estimates of the residence time of phenocrysts and/or microlites prior to the onset of eruptive events, as well as the tools to identify their environment of formation: reservoir, conduit and/or subaerial. However, the above approaches need to be integrated into a compatible model to successfully decipher the timescales of complex magmatic processes (Hawkesworth et al. 2004). The study of individual volcanoes also plays a crucial role in the understanding of the dynamic behaviour of magmas on their way to the surface. One of the best known natural laboratories is Stromboli volcano: an open and dynamic magmatic system. In this contribution we estimate the P –T regimes of Stromboli magmas during their ascent from the upper mantle to the crust, where they are

finally stored within an active magma chamber. Part of the work is also focused on the CSD and bubble contents for lavas and ejected juvenile tephra that have been used to characterize the rheological properties of Stromboli basalts. We finally discuss the plumbing system of this volcano from the upper mantle to the surface, focusing on the possible geometry of the magma chamber within the regional geotectonic setting.

Geological setting and volcanological outline Stromboli is the northeasternmost island of the Aeolian Arc (Fig. 1) and is located on a NE –SW strike –slip fault: the Stromboli–Panarea alignment, a branch connected to the Tindari– Letojanni fault that, in turn, propagates though eastern Sicily, underlying Etna volcano. Stromboli volcano has been built on rather thin continental crust (15 –18 km thick according to Finetti & Del Ben 1986; Pontevivo & Panza 2006). Fragments

Fig. 1. Sketch of the Aeolian Islands with the major tectonic units of the region. Stromboli is located on the Stromboli –Panarea alignment, a normal fault with a left-lateral strike–slip component (Caccamo et al. 1996).

STROMBOLI: FROM MANTLE TO ERUPTION

of quarzite together with rocks of the tonalite – diorite suite have been recently found as xenoliths in the older lavas outcropping at Stromboli (Renzulli et al. 2001; Vaggelli et al. 2003). From the geodynamic point of view, the Aeolian Islands were built in the last 1.3 Ma (Gillot & Keller 1993). Erupted lavas and tephra are subduction-related calcalkaline, high K-calcalkaline (HKCA), shoshonitic and potassic suites (Barberi et al. 1974; Beccaluva et al. 1985). Subduction of the Ionian plate beneath the Calabrian Arc ceased about 1 Ma ago and was followed by a regional uplift (0.5–0.7 Ma.), associated with crustal extension (Westaway 1993; Hippolyte et al. 1994). Uplift occurred within the forearc region, and was ascribed to the rebound of the upper plate (Calabrian arc and part of the Ionian lithosphere) eventually decoupled from the ‘main’ Ionian plate, as postulated by Gvirtzman & Nur (1999, 2001). This process followed the 1–0.7 Ma rollback of the slab. Plate decoupling was accompanied by mantle upwelling, first controlled by a system of faults trending WNW– ESE. Since Pleistocene times, the southern propagation of the Tyrrhenian rifting and the western margin of the ‘roll-backing’ crust generated the NNW– SSE striking ‘Tindari – Letojanni (TL)’ fault (Ventura et al. 1999; De Astis et al. 2003). Mount Etna and the central cluster of the Aeolian Islands are located on this major structure (Fig. 1). According to the latter authors, the Moho is located at c. 25 km depth below the Aeolian Islands. Conversely, Morelli et al. (1975) locate this discontinuity at about 18– 20 km below Stromboli. Recent systematic analyses of teleseismic data by Pontevivo & Panza (2006) and Barberi et al. (2007) support the idea that the Moho below this volcano could be even shallower, and located at about 15 –17 km depth.

Stromboli volcano Stromboli is a composite stratovolcano with a unique open-system activity, taken as a reference case in volcanology to identify minor to intermediate volcanic eruptions (e.g. Newhall & Self 1982). The so-called ‘mild’ and persistent strombolian activity may be interrupted by lava effusions, major explosions and paroxysms often coeval with the generation of tsunamis (Barberi et al. 1993). The volcano is approximately 3 km high with the top reaching 924 m a.s.l. and most of the edifice extending below sea-level. The cone of the volcano rising above sea level was formed during the last 100 ka (Gillot & Keller 1993). Volcanic activity is essentially strombolian, with continuous explosions and eruptions of scoriae, lapilli, ash and bombs (Rosi et al. 2000) at

35

three summit vents located in the upper part of the Sciara del Fuoco. This structural and morphological feature is a collapsed sector delimited by a horseshoe-shaped scarp opening northwestward (Tibaldi 2001). During paroxysmal explosions ‘golden pumices’, generated by fragmentation of the deep undegassed basaltic magma, are ejected together with black scorias and accidental blocks (Francalanci et al. 1999; Me`trich et al. 2001, 2005; Bertagnini et al. 2003). Petrochemically, current lavas and tephra are high-K calc-alkaline basalts straddling the shoshonite field. The volcanic cone of Stromboli has been subdivided into five units (Hornig-Kjarsgaard et al. 1993; Francalanci et al. 1993; Fig. 2): three of them refer to the older part of the volcano, known as Paleostromboli. The fourth is named Neostromboli and is located along the northern and western sectors of the island. K– Ar dating by Gillot & Keller (1993) show that these units are relatively young. Older lavas and tephra (Paleostromboli I, with ages of 110–85 ka) are essentially high-K calc-alkaline (HKCA) basaltic andesites and andesites that are overlain by similar volcanic rocks with intermediate compositions, ranging from calc-alkaline (CA) to high-K calc-alkaline, named Paleostromboli II, aging 64–55 ka. The lavas and tephra of Paleostromboli III (35+6 ka) are on top of the latter sequence and are essentially represented by HKCA andesites and shoshonites (SHO) with different degrees of evolution. The Vancori SHO sequence (ranging from basalts to trachytes, with ages ranging from 26 to 13 ka) lies on top of these rocks and is, in turn, overlain by the rocks of the so-called Neostromboli (13–6 ka). These are essentially shoshonitic with some lavas showing a potassic affinity (KS), as summarized by Francalanci et al. (1989, 1993). The more recent San Bartolo lavas were erupted onto the NE flank and are HKCA basalts which carry abundant mafic and ultramafic xenoliths ascribed to the roots-zone of the deep feeding system (Laiolo & Cigolini 2006).

Structural framework Several papers have been published on the structural evolution of Stromboli. The bulk architecture of the island is the result of summit collapses of caldera type that, according to Pasquare` et al. (1993) and Tibaldi (1996), occurred from 100 to 24 ka BP . More recently, Tibaldi (2001) has shown the existence of four lateral collapses that affected the NW flank within the last 13 ka. The last one produced the so-called Sciara del Fuoco. The main structural features of the island are sketched in Figure 2. The sequence of lateral collapses has been attributed, according to the cited

36

C. CIGOLINI ET AL.

Fig. 2. Simplified geologic map of Stromboli (modified after Keller et al. 1993). Major units are represented together with the recent lava fields of ‘Nel Cannestra`’ and ‘San Bartolo’. Major structural features are represented as well (see text for details).

authors, to the combined affects of dyke intrusion, basement morphology and slope erosion coupled with changes in sea-level. Currently, summit vents are located along a N408E fracture. This trend is consistent with the main direction of dyking (cf. Pasquare` et al. 1993; Tibaldi 2001). However, recent work on CO2 degassing revealed the existence of a second system of fractures essentially oriented N608E (Finizola et al. 2002). The sector of the cone comprised between these two fracture zones seems to be the zone where current diffuse degassing occurs most efficiently (cf. Cigolini et al. 2005, 2007). In a recent paper, Tibaldi et al. (2003) have shown that the stress distribution, in agreement with focal mechanisms solutions (e.g. Caccamo et al. 1996), is consistent with a N408E normal fault with a minor strike–slip component. Therefore, smax ¼ s1 will be vertical, s2 is horizontal along the above direction and s3 would be the extensional component onto the horizontal plane (trending N508W, i.e. normal to the s1 – s2 plane). In Figure 3, we report two recent images of the summit of Stromboli collected after the last major eruption. It is clearly visible that summit vents are located along a fracture which exhibits a grabenlike structure, consistent with an extensional regime. However, the topography of the cone as well as the morphology of Sciara del Fuoco may produce local anisotropies of stress distributions which may affect dyke intrusions and the opening

of subsidiary vents onto Sciara del Fuoco (e.g. Acocella & Tibaldi 2005; Acocella et al. 2006).

Historic and current major eruptions As previously mentioned, the relatively ‘mild’ and persistent Strombolian activity can be interrupted by lava effusions, major explosions and paroxysms eventually coupled with the generation of tsunamis (Barberi et al. 1993). San Bartolo basalts effused from a parasitic vent, located NE of the present crater area, during Roman times (200 –400 AD according to Arrighi et al. 2004). These lavas were erupted as a composite lava field (extending about 0.65 km2) onto the Vancori lavas and the Cannestra` lava field (Fig. 2), both exposed in the northern sector of the island. Unfortunately, the chronology of eruptions during the following centuries is not well known. However, Nappi (1976) filed a compilation with a description of the major eruptive events, from 1850 to 1975 (Nappi 1976, and references therein). In addition, the eruption of 1985 was reported by De Fino et al. (1988). According to the accurate reconstruction of Barberi et al. (1993), 27 paroxysms have occurred at Stromboli since 1558, but only a few took place in connection with effusive eruptions. Seven tsunamis have been recorded in the last 120 years: in 1879, 1916, 1919, 1930, 1944, 1954 and 2002. All these events were coeval with paroxysmal

STROMBOLI: FROM MANTLE TO ERUPTION

37

Fig. 3. Selected images of Stromboli volcano following the most recent major eruption. (a) The northeastern side of Stromboli with the crater area and the N408E fracture zone where the craters are aligned, and have generated a graben-like structure. The dashed white lines define the Sciara del Fuoco scarp; the full line is the N408E summit fracture zone which subdivides the Upper and Lower Terrace. This picture was taken during a helicopter survey on 28 June 2007 looking south. (b) The graben-like structure observed from the summit of Stromboli (looking NE). The northern rim of the Sciara del Fuoco is also visible on the right. Picture taken on 29 June 2007. Symbols as in (a).

eruptions, and were related to flank failure and slumping into the sea of portions of the Sciara del Fuoco, including its submerged part. To give an overview of how Stromboli volcano works during its major eruptive cycles, a summary of the last

two major eruptions is provided here. The earlier one started on 28 December 2002 with the eruption of a hot avalanche from the NE crater that preceded the emplacement of a lava flow onto Sciara del Fuoco. The onset of the effusive event was followed

38

C. CIGOLINI ET AL.

by a composite slump on the Sciara del Fuoco (30 December 2002), with flank failure including portions of the submerged part of the volcano, which generated a tsunami that affected the northern coast of the island (Bonaccorso et al. 2003). Almost continuous effusion of lava persisted until 21 July 2003 (Ripepe et al. 2005) and was interrupted by a major explosion on 5 April 2003, with the ejection of a 1 km high column (mainly consisting of ‘golden’ basaltic pumices) that expanded up to 4 km (Calvari et al. 2005). The total amount of lava erupted during this eruptive cycle has been estimated at 1– 1.3  107 m3 (Ripepe et al. 2005; Calvari et al. 2005). By the end of July 2003, the typical mild Strombolian activity resumed at the summit craters, and persisted until 27 February 2007, when a new lava flow effused from the NE crater (located at about 720 m a.s.l.). Lava discharge was continuous until 2 April 2007, and essentially from a new active vent that opened at 400 m a.s.l. in the evening of 27 February 2007. The opening of this vent replaced the lava outflow from the active NE crater. A new paroxysmal explosion (consisting essentially of lithic tephra, blocks and bombs with subordinated pumices) occurred on 15 March 2007, and was preceded by intense volcanic seismicity and followed by the vertical collapse (of about 150 m from its original altitude) of the summit crater’s floor (in the night of 24 March 2007). The mild Strombolian activity was then resumed by end of June/beginning of July, 2007. Preliminary and conservative estimates for the total amount of lava erupted in this phase are 6–9  106 m3 (S. Calvari, personal communication to C.C., 24 July 2007).

Analytical methods During field work we collected a total of 78 samples. We selected 21 of these for petrographic and geochemical investigations. Selected samples were analysed by means of a SEM Cambridge Instruments Stereoscan 360 equipped with an EDS Energy 200 and a Pentafet detector (Oxford Instruments). Operating conditions were 15 kV accelerating voltage, and a beam current of 2.68 A. Quantitative data (with spot sizes of 2 and 15 mm diameter for minerals and glasses respectively) were acquired and processed using the Microanalysis Suite Issue 12, INCA Suite version 4.01. Data were corrected for background, drift, mass absorption and secondary fluorescence using natural and synthetic standards. The relative errors are within 5% for most major elements and may go up to 10% for minor elements (with concentrations ,1.00 wt%). Bulk chemical (XRF) analyses were done in Zurich (Switzerland) at the Institute of Mineralogy

and Petrology (IMP-ETH, Swiss Federal Institute of Technology) by means of a wavelength dispersive X-ray fluorescence spectrometer (WD-XRF, Axios, PANalytical), equipped with five diffraction crystals. Analyses were determined on selected lavas, scoriae and pumices collected at the summit of Stromboli during the 2002–2003 eruptive cycles. For comparison, some scoria samples were also collected following this eruptive cycle. Major and minor element analyses were performed on fused glass-beads prepared from fired, finely ground rock powders mixed with lithiumtetraborate (1:5 mixture) using a Claisse M4w fluxer. Calibration is based on 30 certified international standards. The precision of analysed elemental abundances is better than +0.2% for SiO2 and +0.1% for the other major elements, with the exception of MnO and P2O5, which have concentration errors of c. +0.02%. Crystal size distribution data were collected by means of digital images of thin sections acquired by a JVC-3CCD camera applied to an Olympus BX60 optical microscope. Additional binary images of microlites were also obtained from images collected with an EDS microanalyser linked to a Cambridge stereoscan 360 SEM. Grain types were differentiated in a bidimentional plain in terms of number, area, equivalent radius and elongation by means of the computer code SCION IMAGE (Scion Corporation, Frederick, MD, USA), which allows separation of crystals from the background employing a segmentation technique. The images were also used to estimate bubble content as well as the average distance between bubbles. We performed automated measurements of single crystal areas as well as the major and minor axes of the best-fit ellipse to each crystal, to obtain the number of crystals per unit area (calculated on a bubble-free basis).

Petrography and mineral compositions In this section we briefly report the petrographic and minerochemical features of the lavas and juvenile ejecta erupted in 2002–2003 that will then be used in thermobarometry. We will also briefly revisit the petrography of San Bartolo basalts and their host mafic and ultramafic nodules (Laiolo & Cigolini 2006), since the P–T regimes for the crystallization of these materials will be used in constraining the downward extension of Stromboli magma chamber. A synopsis of the petrographic features of the lava and juvenile tephra together with the mafic and ultramafic nodules is given in Figure 4. Histograms for the mineral compositions found in pumices and scorias ejected during the 5 April 2003 explosion are reported in Figure 5,

STROMBOLI: FROM MANTLE TO ERUPTION

39

Fig. 4. Microphotographs of the petrographic and textural features of Stromboli lavas, scorias and pumices erupted in 2002–2003, together with mafic and ultramafic nodules included in San Bartolo lavas. (a) Lava erupted on 28 December 2002 with euhedral plagioclase, clinopyroxene and subordinate olivine in a fine-grained glassy matrix. Plane polarized light. (b) Scoria erupted a few days before the above lava, with euhedral to subhedral plagioclase and larger phenocrysts of olivine and clinopyroxene. Bubbles are also present (white vescicles). Plane polarized light. (c) Photomicrograph of a ‘golden pumice’ erupted during the paroxysmal explosion of 5 April 2003. Vescicularity is high (c. 50%) with bubbles variable in size and a very low degree of crystallinity (10% modal). Plane polarized light. (d) Partially disaggregated dunitic nodule in San Bartolo basalt: forsteritic olivine shows deformation bands (central-right part of the nodule) and is locally resorbed due to reaction with the host melt. Crossed polarizers. (e) Wherlitic nodule in San Bartolo basalts with olivine showing deformation bands and decompression rims (Wanamaker & Evans 1985) visible in the central part of the photomicrograph. Crossed polarizers. (f) Corona of orthopyroxene–clinopyroxene–magnetite rimming olivine in gabbronorite. Crossed polarizers.

40

C. CIGOLINI ET AL.

Fig. 5. Histograms showing the mineral compositions of lavas, scorias and pumices erupted at Stromboli during the 2002– 2003 major eruptive cycle. Data include both core and rim compositions of phenocrysts. Number of analyses is represented by bars stacked on top of each other.

Fig. 6. Histograms for the mineral compositions of San Bartolo basalts and host mafic and ultramafic nodules. The basalts show a higher compositional variability, with a higher frequency mode toward more ‘primitive’ phases, partially overlapping those of the nodules. Data include both core and rim compositions of phenocrysts. Bars stacked on top of each other refer to the number of analyses.

0.83 0.16 ¼ Total iron as FeO. T

100.1 101.3

0.69 0.30 0.88 0.10

100.6 99.9

0.70 0.29 0.69 0.29

100.0 99.8

0.69 0.29 0.71 0.27

99.2 100.7

0.71 0.27 0.70 0.29

99.5 99.3

0.69 0.29 0.70 0.29

101.0 101.2

0.70 0.29 Fo Fa

Total

39.5 15.4 0.50 44.7 bd 37.3 27.6 0.59 35.5 0.34 41.8 9.8 0.79 47.8 0.33 37.1 26.3 bd 36.1 0.36 37.3 26.5 0.64 35.2 0.42 37.0 26.2 0.58 35.5 0.50 36.7 25.1 0.60 36.5 0.40 37.1 25.5 0.67 37.0 0.46 37.5 26.0 0.52 35.2 0.34 37.4 26.2 0.46 36.6 0.39 37.6 26.3 0.45 36.4 0.44 SiO2 FeOT MnO MgO CaO

Sample Site Analysis no. Date

St55 rim 2 January 2003

36.9 26.4 0.55 35.1 0.38

St62 core 12 April 2003 St64 rim 11 April 2003

41

St55 core 1 January 2003

St61 core 3 February 2003

St61 rim 4 February 2003

St51 core 6 January 2002

St51 rim 7 January 2002

St59 core 8 December 2002

St59 rim 9 December 2002

St64 core 10 April 2003

Pumices Scorias Recent lavas Type

Table 1. Selected electron microprobe analyses for olivine found in lavas and tephra of the 2002 – 2003 Stromboli eruption

and are compared with those measured in 2002–2003 lava flows. Frequency modes for mineral phases of San Bartolo basalts and their host mafic and ultramafic materials are reported in Figure 6. Lavas erupted in 2002–2003 and crystal-rich scorias are characterized by a high degree of crystallinity (from 35 to 45% modal) and variable bubble content, 10–25 and 25–35% modal, respectively. They show an identical mineralogy and exhibit a porphyric to glomeroporphyric texture consisting of olivine, clinopyroxene and plagioclase on a hyalopilitic to pilotaxitic groundmass of plagioclase and pyroxene coexisting with a brownish glass (Fig. 4a & b). Accessory minerals are Ti-magnetite and apatite. Olivine is present as euhedral to subeuhedral microphenocrysts (0.2– 1 mm in size) locally enclosed within pyroxene. Within the lavas and scorias, olivine phenocrysts and microphenocrysts range from Fo73 to Fo68 (Table 1). Clinopyroxene phenocrysts, locally zoned, are euhedral to subeuhedral (0.5–4 mm in size). Clinopyroxene is salite and augite (Mg# 0.7–0.76) with average compositions of Wo36 – 43En39 – 48Fs5 – 15 (Table 2). Some phenocrysts show diopsidic cores (Mg#  0.80; Wo39 – 42 En40 – 46Fs5 – 7) and the highest Al2O3 content (up to 4 wt%). Similar compositions have been found for some matrix microlites, as well as for some rims of augite phenocrysts. Plagioclase phenocrysts (ranging from 0.3 to 3 mm in size) are idiomorphic to subidiomorphic showing normal-oscillatory zoning from bytownite to labradorite (Table 3). Some phenocrysts may show anorthitic preserved cores (An92 – 90). Matrix plagioclase is essentially labradorite (An65 – 74). Compared with lavas and scorias, crystal-poor pumices exhibit a nearly aphyric texture (Fig. 4c) together with a lower degree of crystallinity (up to 10% modal). Vescicularity is considerably higher (reaching 50% in volume). Mineral phases are microphenocrysts of olivine (0.3 mm in size), clinopyroxene and plagioclase (similar in size to those found within the scorias) in a glassy matrix. Euhedral phases are essentially olivine and clinopyroxene (diopside-salite with compositions nearly identical to those found in some phenocrystic cores of the scorias), the former being included within the latter. Microphenocrysts of augitic composition are also present. In this case, olivine shows a higher frequency at compositions Fo74 – 69, but some microcrystals are richer in MgO (up to Fo84, Table 1). Similar compositions have also been found in some olivine rims. Entrapped in these rims, Me`trich et al. (2005) found primitive melt inclusions. The olivine–clinopyroxene Fe–Mg Cpx Ol Cpx distribution coefficient (KD ¼ XOl Mg XFe /XFe XMg )

St62 rim 13 April 2003

STROMBOLI: FROM MANTLE TO ERUPTION

42

Table 2. Selected electron microprobe analyses for clinopyroxene found in lavas, scorias and pumices of the 2002 – 2003 Stromboli eruption – the recent materials erupted by Stromboli volcano Type

Scorias

Pumices

St53 core 1 December 2002

St53 rim 2 December 2002

St61 core 3 February 2003

St61 rim 4 February 2003

St51 core 5 June 2001

St51 int 6 June 2001

St51 rim 7 June 2001

St59 core 8 December 2002

St59 rim 9 December 2002

St62 core 10 April 2003

St62 rim 11 April 2003

St64 core 12 April 2003

St64 rim 13 April 2003

SiO2 TiO2 Al2O3 FeOT MgO CaO Na2O

50.8 0.90 3.39 9.46 14.1 20.6 0.37

50.7 0.97 3.47 8.93 14.3 21.5 0.37

53.8 0.01 2.29 3.11 17.2 22.7 n.d.

51.3 0.83 3.26 8.42 14.7 21.9 0.33

50.4 0.85 2.89 9.55 14.5 21.0 n.d.

50.0 0.62 4.63 4.92 15.7 22.6 n.d.

50.6 0.97 3.10 8.96 14.7 21.6 n.d.

52.6 0.01 3.92 5.31 16.3 22.8 n.d.

51.6 1.04 3.44 8.79 14.8 21.5 n.d.

50.1 1.19 3.30 11.5 13.8 19.7 0.35

51.3 0.50 4.04 5.22 15.9 22.5 n.d.

50.6 0.68 5.24 4.77 14.7 23.4 n.d.

50.5 0.80 5.22 5.30 14.9 23.4 n.d.

Total

99.7

99.2

98.6

99.5

99.3

Wo En Fs T

0.38 0.39 0.14

100.5 0.39 0.40 0.12

¼ Total iron as FeO; n.d. ¼ not detected.

99.1 0.42 0.47 0.05

100.8 0.39 0.41 0.11

0.39 0.41 0.14

0.39 0.44 0.06

100.0 0.39 0.41 0.13

101.0 0.39 0.44 0.07

101.2 0.39 0.41 0.13

100.1 0.36 0.39 0.16

0.40 0.44 0.07

0.40 0.40 0.07

100.2 0.40 0.41 0.07

C. CIGOLINI ET AL.

Sample Site Analysis no. Date

Recent lavas

Table 3. Selected electron microprobe analyses for plagioclase found in recent lavas and tephra of the 2002 – 2003 Stromboli eruption Recent lavas

Scorias

Pumices

Sample St53 St53 St55 St55 St61 St61 St59 St59 St59 St59 St62 Site core rim core rim core rim core rim core rim core Analysis no. 1 2 3 4 5 6 7 8 9 10 11 Date December December January January February February December December December December April 2002 2002 2003 2003 2003 2003 2002 2002 2002 2002 2003 SiO2 Al2O3 FeOT CaO Na2O K2O

46.2 32.6 0.83 17.3 1.94 0.20

50.0 30.4 0.91 15.0 3.12 0.37

47.9 31.4 1.02 16.3 2.39 0.32

50.3 30.1 0.90 14.3 3.37 0.54

48.2 31.4 0.95 15.8 2.58 0.32

50.5 30.1 0.57 14.5 3.17 0.64

Total

99.0

99.8

99.3

99.6

99.2

99.5

Ab An Or T

0.17 0.82 0.01

0.27 0.71 0.02

0.21 0.78 0.02

0.29 0.68 0.03

0.22 0.76 0.02

0.27 0.69 0.04

48.4 31.2 0.99 16.4 2.53 0.42 100.0 0.21 0.76 0.02

50.9 29.4 0.89 13.7 3.72 0.71 99.2 0.32 0.64 0.04

46.2 34.0 0.87 18.5 1.60 0.26 101.4 0.13 0.85 0.01

50.5 30.2 0.87 14.2 3.42 0.59

47.5 32.2 1.04 16.7 2.12 0.23

99.7

99.8

0.29 0.67 0.03

0.18 0.80 0.01

St62 rim 12 April 2003

St62 core 13 April 2003

St62 rim 14 April 2003

St64 core 15 April 2003

St64 rim 16 April 2003

50.6 46.5 30.2 32.6 0.84 0.94 14.5 17.6 3.45 1.81 0.57 n.d.

50.3 30.3 0.74 14.5 3.51 0.58

50.1 30.2 0.99 15.2 3.01 0.46

45.2 33.2 0.67 18.4 1.36 0.00

99.9

99.9

98.8

100.2 0.29 0.68 0.03

99.4 0.16 0.84 —

0.30 0.67 0.03

0.26 0.72 0.03

0.12 0.88 —

STROMBOLI: FROM MANTLE TO ERUPTION

Type

¼ Total iron as FeO; n.d. ¼ not detected.

43

44

C. CIGOLINI ET AL.

is approximately constant within the pumices (KD ¼ 0.5 + 0.06), indicating that olivine is in equilibrium with diopsidic clinopyroxene. Microphenocrysts of augitic composition (locally with diopsidic cores) are also present but probably inherited from the interaction with the crystal-rich magma (with a marked variability in KD, ranging from 1.2 to 1.8 in this specific case). Rare subhedral crystals of plagioclase (bytownite-labradorite, An90 to An64; Table 3) have been found. Most of the crystals exhibit distinct microcracking, particularly plagioclase, which shows reaction rims with the matrix glass (being partially and/or totally resorbed). The glass is rather heterogeneous with domains of dark brown glass straddling to a lighter glass. San Bartolo basalts exhibit a seriate porphyric to glomeroporphyric texture, with a high porphyrity index (up to 40 –45%) and a high crystallinity (c. 55 –60% modal). Groundmass textures are typically pilotaxitic to microfelsitic-granular textures. Plagioclase is the most abundant phase (An90 – 65) with subordinate clinopyroxene (diopside-salite and augite), olivine (with a higher frequency mode from Fo75 to Fo68) and orthopyroxene (hypersthene straddling to pigeonite). Accessory phases consist of abundant titano-magnetite and apatite. Crystal clots of gabbroic, gabbronoritic and ‘anorthositic’ type are abundant within San Bartolo lavas, together with xenocrysts of olivine and anorthitic plagioclase. Host nodules in San Bartolo lavas are mafic (gabbro, gabbronorite and anorthosite with distinct cumultic texture) as well as ultramafic (dunite, wehrlite and clinopyroxenite with porphyroclastic to mosaic-equigranular textures, see Fig. 4d & e) and have been described in detail by Laiolo & Cigolini (2006). According to them, gabbroic inclusions may be regarded as cumulates and represent crystallized portions of earlier Stromboli basalts. Some gabbroic and anorthositic samples show deformation twinning and weakly mylonitic textures, whereas gabbronorites may exhibit coronas of pyroxenes and titano-magnetite surrounding olivine (Fig. 4f). These textures have been interpreted as a product of subsolidus reactions typical of metagabbros of the granulite facies (Gardner & Robins 1974; Cigolini & Kudo 1987), probably representative of the middle–lower crust below Stromboli volcano. Conversely, ultramafic nodules represent mantle materials which have been captured during magma ascent. In summary, the compositional data obtained on mineral phases of recently erupted lavas and juvenile tephra lead to a rather similar frequency distribution, both for plagioclase and pyroxene (Fig. 5).

Conversely, olivine is somewhat ‘bimodal’, with the higher forsteritic content essentially restricted to the crystal poor ‘golden pumices’. However, similar Fo-rich compositions are found in mafic and ultramafic nodules hosted in San Bartolo lavas. In turn, these basalts show a higher compositional variability both for plagioclase and ferromagnesian phases (Fig. 6).

Bulk composition and glass chemistry Summaries of average bulk and glass compositions are given in Tables 4 and 5, respectively. Here we describe the major element compositions which will be used in thermobarometric calculations. Lavas and scorias are shoshonitic basalts, have nearly identical compositions and tend to be slightly richer in SiO2 and alkalies when compared with the pumices erupted on 5 April 2003 (Fig. 7). In turn, pumices exhibit higher MgO (Mg# 0.55–0.60) and CaO contents, whereas K2O is lower (c. 1.8 wt%). Glass compositions (Table 5) are more evolved within the lavas and scorias, as shown by the K2O vs SiO2 diagram of Peccerillo & Taylor (1976), where they show a clear compositional trend (Fig. 8). However, some melt inclusions found within the olivine crystals of golden pumices (Me´trich et al. 2005) are more primitive (with SiO2 ranging between 46.8 and 47.9 wt%) and exhibit an HKCA affinity (Fig. 7), with MgO, CaO and alkali contents slightly lower than those of the pumice bulk compositions. In comparison, San Bartolo basalts are richer in SiO2 (up to c. 51 wt%) and lower in alkali, thus falling into the HKCA field. It is likely that San Bartolo basalts originally had the same composition of those primitive melts, and their evolution was controlled by the interaction with lowercrust and mantle materials preserved as nodules (cf. Laiolo & Cigolini 2006). In this case, the temperature of the uprising melts was not high enough to melt the ‘foreign materials’ (since their minerals are nearly identical to the earlyformed liquidus phases in the differentiating San Bartolo basalts), but was sufficient to induce the onset of an ‘assimilation reaction’ within the melt (Bowen 1928; McBirney 1979). This process increased the efficiency of the system in fractionating plagioclase and the ferromagnesian phases, thus leading to silica enrichment coupled with a minor alkali content. This is also consistent with the presence of hypersthene (straddling toward pigeonite compositions) in San Bartolo Basalts (Kudo 1983).

Table 4. Bulk composition of the lavas, scorias and pumices recently erupted at Stromboli volcano St55 Lavaa 1 28/12/02

St61 Lavaa 2 19/02/03

St68 Lavaa 3 17/03/03

St69 Lavaa 4 26/06/03

St59 Scoriab 6 28/12/02

St67 Scoriac 7 12/11/03

St70 Scoriab 8 16/09/05

St71 Scoriab 9 18/04/06

St62 Pumicec 11 05/04/03

St64 Pumicec 12 05/04/03

SiO2 TiO2 Al2O3 Fe2OT3 MnO MgO CaO Na2O K2O P2O5 LOI

49.7 0.91 17.1 8.86 0.16 6.27 11.3 2.46 2.12 0.53 0.10

49.6 0.93 17.1 8.71 0.16 6.04 11.2 2.49 2.16 0.55 0.12

49.4 0.86 17.0 8.67 0.15 6.16 11.1 2.69 2.11 0.48 0.75

49.6 0.92 17.3 8.76 0.16 6.13 11.3 2.46 2.10 0.54 —

49.9 0.88 16.6 8.94 0.16 6.74 11.7 2.71 2.06 0.48 —

49.8 0.89 17.0 8.86 0.16 6.36 11.4 2.78 2.03 0.50 —

49.6 0.85 16.4 8.69 0.15 6.78 11.4 2.65 2.21 0.50 0.11

49.6 0.87 17.1 8.70 0.16 6.15 11.3 2.74 2.08 0.49 —

48.6 0.91 17.2 8.70 0.16 6.13 11.6 2.37 1.91 0.53 0.14

49.3 0.88 17.3 8.93 0.16 6.49 11.7 2.66 1.82 0.49 0.09

Total

99.5

99.1

99.3

99.3

99.7

99.4

99.2

98.3

99.9

Mg# CaO/Al2O3 K2O/Na2O

0.558 0.661 0.860

0.553 0.656 0.870

0.559 0.653 0.785

0.555 0.654 0.855

100.2 0.573 0.707 0.762

0.561 0.669 0.730

0.582 0.694 0.833

0.557 0.658 0.760

0.557 0.671 0.805

0.564 0.674 0.684

STROMBOLI: FROM MANTLE TO ERUPTION

Sample Type Analysis no. Date

T

¼ Total iron as Fe2O3; LOI, loss on ignition. sampled along the Sciara del Fuoco; bsampled near and below the crater area; csampled in the summit area.

a

45

46

Table 5. Selected electron microprobe analyses for interstitial glasses found in lavas and tephra of the 2002–2003 Stromboli eruption Type

Recent lavas

Scorias

Pumices

Sample Str53 Str53 Str53 Str53 Str55 Str55 Str059 Str059 Str063 Str063 Str063 St62 St62 St62 St62 St64 St64 Analysis no. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Date December December December December January January December December December December December April April April April April April 2002 2002 2002 2002 2003 2003 2002 2002 2002 2002 2002 2003 2003 2003 2003 2003 2003 51.7 1.64 15.1 9.40 n.d. 3.65 7.65 3.31 4.52 1.10

52.1 1.59 15.4 9.49 n.d. 3.56 7.49 3.24 4.29 1.13

52.1 1.44 15.2 9.29 n.d. 3.62 7.36 3.06 4.24 1.20

51.6 1.80 15.3 10.66 0.12 2.93 7.56 3.13 4.46 1.28

50.3 1.32 18.2 8.23 0.10 2.45 9.09 3.52 3.39 1.17

50.9 1.78 15.2 11.28 n.d. 3.22 5.68 3.35 5.13 1.33

51.2 1.64 14.6 9.72 n.d. 4.44 8.10 2.76 4.77 1.48

51.7 1.90 15.4 9.92 n.d. 3.65 8.03 3.17 4.25 1.41

52.7 1.54 15.6 9.16 n.d. 3.70 7.56 2.60 4.16 1.19

52.0 1.86 15.8 9.11 n.d. 3.55 7.60 2.55 4.17 1.53

50.8 1.68 15.3 9.53 n.d. 4.08 9.05 3.05 4.32 0.99

49.3 49.7 48.5 49.2 48.0 48.1 1.07 1.06 0.96 1.02 0.98 1.14 17.8 17.8 17.6 17.7 16.98 17.12 8.63 7.96 8.74 7.43 8.20 9.31 n.d. 0.10 n.d. n.d. n.d. n.d. 5.39 5.08 5.98 6.03 5.83 5.98 13.3 12.1 12.5 13.6 12.15 11.97 2.29 2.26 2.15 2.15 2.5 2.4 1.57 2.53 1.91 1.38 2.1 2.5 n.d. 0.93 0.57 0.88 0.78 0.68

Total

98.1

98.3

97.5

98.8

97.7

97.8

98.8

99.4

98.2

98.2

98.8

99.3

Mg# CaO/Al2O3 K2O/Na2O T

0.41 0.51 1.37

0.40 0.49 1.32

¼ Total iron as FeO; n.d. ¼ not detected.

0.41 0.48 1.39

0.33 0.49 1.42

0.35 0.50 0.96

0.34 0.37 1.53

0.45 0.55 1.73

0.40 0.52 1.34

0.42 0.48 1.60

0.41 0.48 1.64

0.43 0.59 1.42

0.53 0.75 0.69

99.5 0.53 0.68 1.12

98.9 0.55 0.71 0.89

99.3 0.59 0.77 0.64

97.4 0.56 0.72 0.83

99.2 0.53 0.70 1.02

C. CIGOLINI ET AL.

SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O P2O5

STROMBOLI: FROM MANTLE TO ERUPTION

47

Fig. 7. Low silica-moderate K2O sector of the diagram of Peccerillo & Taylor (1976) (reported in Fig. 8) with the compositions of materials erupted in 2002– 2003. The star is the average chemical composition of San Bartolo basalts. The grey field represents the distribution of primitive melt inclusions found in forsteritic olivines of Stromboli golden pumices (Me`trich et al. 2001; Bertagnini et al. 2003).

Fig. 8. K2O v. SiO2 diagram of Peccerillo & Taylor (1976) for interstitial glasses found in lavas and tephra erupted at Stromboli during the 2002–2003 eruptive cycle. The grey field shows the higher silica enrichment of the glasses of the mesostasis in San Bartolo basalts.

48

C. CIGOLINI ET AL.

Fig. 9. Binary versions of scanning electron microscope (SEM) images (a) and (b) analysed photomicrographs of plagioclase microlites of the 5 April 2003 pumice event. (c) Binary and (d) analysed version of vesicles of 5 April 2003 pumice samples. Images are 960 mm across.

Crystal size distribution and intrabubble distances The CSD method allows measurements of the population density (crystal or bubble number density of a given size in a given interval) and to acquire information about the kinetics and dynamics of crystallization (Marsh 1988; Cashman & Marsh 1988). Both crystals and bubbles were analysed through the segmentation of binary images (Fig. 9). Three-dimensional size distributions were obtained by calculating an average length of each crystal per size class (L) using the square root of individual crystal area measurement. We then calculated areal number densities for each size class (NAc), and converted them into a three-dimensional distribution (number per unit volume, NVc) according to (Cashman & McConnell 2005): NVc ¼ NAc =L

(1)

This procedure has been shown to be a reliable method to correct the ‘intersection probability’ since designated sampling planes are proportional

to crystal size (Underwood 1970). We took several digital images of each thin section in order to cover the entire area of the section (the average area per thin section analysed was c. 550.00 mm2 with c. 1800 crysts per thin section). Following Randolph & Larson (1971), the obtained NVc data may be converted to CSDs for each sample. For steady conditions of crystal nucleation and growth (e.g. Cashman 1990): n ¼ no exp (L=Ld )

(2)

where n o is the crystal number density and Ld is the number-referenced dominant size. If data are linearly distributed, the plot ln(Nv) v. L (mm) will yield a straight line with a slope of 21/Ld and intercept n o (Fig. 10). The total number of crystals per unit volume can be calculated as Nt ¼ n oLd. Assuming that the dominant size is a consequence of steady growth (G) over a certain period of time (t), Ld ¼ Gt and the residence time of crystals in different formation settings (magma chambers, conduit and external quenching) can be estimated. Table 6 summarizes all the measurements and

STROMBOLI: FROM MANTLE TO ERUPTION

Fig. 10. Crystal size distribution plots for selected samples of the 2002– 2003 eruptive crisis: (a) crystals of the June 2002 scoria, December 2002 scoria and 2003 lava flows. (b) CSD plot of 2003 lava flows: data distribution is well described by the two straight lines. (c) Microlites and phenocrysts of the 5 April 2003 golden pumice. Slopes, intercepts and linear correlation coefficients are quoted in Table 6. The data gap is due to the different image acquisitions, see text for details.

49

50

C. CIGOLINI ET AL.

Table 6. Morphometric data concerning both bubbles and crystals of the selected samples collected during the 2002 –2003 eruptive crisis Sample

2

Total area analysed (mm )

St51

St57

St55 and St61

St62

Scoria June 2002 230.16

Scoria December 2002 316.47

Lava 2003 542.81

Pumice April 2003 1096.74

Bubbles fba 0.21 Average bubble distance (mm) 323.59 s (standard deviation) 0.04 Time of exsolution 2 days Crystals fxxb c

0.31 549.42 0.38 7 days

0.38 22

Nac (mm ) Nvcd(mm23) n8 (L . 0.80 mm)e (mm24) n8 (L , 0.80 mm)e (mm24) n8e (mm24) Ld (L . 0.80 mm)f Ld (L , 0.80 mm)f Nt (L . 0.80 mm)g Nt (L , 0.80 mm)g (mm23) R (L . 0.80 mm)h R (L , 0.80 mm)h

0.03 500.03 0.08 1 day

0.45

0.53 186.8 0.04 15 hours

0.38

3

6.7  10 6.6  104 6.5  102 4.1  104

3

4

4.3  10 4.3  104 6.0  101 2.8  104

1.5  10 1.5  105 3.7  102 1.0  105

0.189 0.098 1.2  102 3.8  103 0.75 0.99

0.343 0.093 2.1  101 2.6  103 0.73 0.99

0.243 0.09 9.0  101 9.3  102 0.95 0.99

0.07 Phenocrysts Microlites 4 4.1  10 2.4  104 5 7.3  10 1.4  1011 1.3  104 0.110

3.4  108 0.003

1.4  103

9.4  105

0.98

0.92

a

Bubbles per unit area; brock crystallinity; careal number density per size class; dvolumetric number density; enumber density of crystal nuclei; fnumber-referenced dominant size; gnumber of crystal per unit volume; hlinear correlation coefficient; aand bmeasurements are directly determined from image analysis; all other parameters are determined from CSD plots.

characteristics determined from image analysis and CSD respectively. The time of exsolution of the gas phase has been constrained using the relation by Navon & Lyakhovsky (1998): d2  4Dt

(3)

where d is the distance between bubbles, D is alternatively the diffusion coefficient of CO2 (1.3  1029 cm2 s21; Zhang & Stolper 1991) and of H2O (1027 cm2 s21; Zhang 1999) and t is time (Table 6). To calculate the average bubble distance, we assumed that bubbles are spheres uniformly distributed in each portion of the thin section. Thus, depending on both number and dimensions of bubbles, we constructed a grid in which each single knot is the bubble centre; in this way the distance can be easily calculated as the difference between two knots and the bubble diameter (Table 6). We analysed several thin sections of the products erupted during the 2002– 2003 crisis that occurred at Stromboli volcano. In particular, we chose samples of scorias, golden pumice and

lava flows for detailed analysis of plagioclase size, shape, areal and volumetric number density (Table 6). To estimate magma residence times inferred from phenocryst crystallization, we considered two feasible growth rates introduced by Armienti et al. (2007) for Stromboli basalt: the ‘net average growth rate’ of 2  10210 mm s21 that takes into account crystal recycling within an active reservoir, and the so-called ‘balance growth rate’ of 1029 mm s21 determined on normally zoned plagioclase phenocrysts. For the microlites of the pumices, which probably crystallized in the very last time span that occurs from magma ascent into the conduit and quenching, we preferred to use a faster growth rate of 3.8  1027 mm s21 (Kirkpatrik 1977). We will discuss our estimates below.

Scorias and lavas The products erupted between June 2002 and February 2003 have crystallinities ranging from 0.38 (lavas) to 0.45 (scoria), and volumetric number densities (NVc) ranging from 4.3  104 to 1.5  105 mm23. The CSD plot shows that data

STROMBOLI: FROM MANTLE TO ERUPTION

are uniformly and linearly distributed (Fig. 10), with intercepts n o varying between 6.02  101 and 6.46  102 mm24 for crystals with equivalent diameter (L) . 0.80 mm and with intercepts n o ranging between 2.83  104 and 1.04  105 mm24 in the case of L , 0.80 mm (Fig. 10b has been selected for better describing this distribution). The downturn on the plot at small size may be due to post-nucleation growth. A summary of the results is given in Table 6. Assuming the net growth rate of 2  10210 mm 21 s , the phenocrysts’ (with L . 0.80 mm) residence time varies between 30 and 55 years, whereas for phenocrysts with L , 0.80 mm the residence time is 14 –15 years. We obtain more realistic estimates if we use the balanced growth rate of 1029 mm s21: in this case the residence times are 6–11 years for phenocrysts with L . 0.80 mm and c. 3 years in the case of L , 0.80 mm. The latter results seems to be roughly compatible with the c. 1 year residence time estimated by Gauthier & Condomines (1999) on short-lived isotopes. The average bubble distance measured varies between 500 mm (lavas) and 550.0 mm (scoria bomb). As a consequence, the time of exsolution of the gas phase results to be 2–7 days, assuming the gas phase to be predominantly CO2, and between 44 min and 2 h in the case of H2O only.

51

1981; McBirney & Murase 1984; Pinkerton & Stevenson 1992), which takes into account the rock crystallinity and the viscosity of the interstitial melt. By systematically applying image analyses and constraining crystal size distribution (CSD, e.g. Cashman 1990), we obtained an accurate estimate of the crystallinity of the erupted products. However, one of the major limits of applying these equations is the estimate of the viscosity of the interstitial melt. Previously, this was obtained by applying the models of Bottinga & Weill (1972) and Shaw (1972), which led to viscosity estimates that were systematically lower than field measurements and experimental data (particularly at subliquidus temperatures). The reason for this discrepancy lies in the fact that basaltic melts fail to reach the viscosity of about 1012 Pa s at the glass transition. Noticeably, the model of Hui & Zhang (2007) may satisfy this constraint at lower temperatures, but melt viscosities in the near-liquidus field are systematically underestimated and unrealistically low. However, Giordano & Dingwell (2003) stressed the non-Arrhenian behaviour for the viscosity of silicate melts by fitting their experimental data using the Tamman–Vogel– Fulcher (TVF) equation (Vogel 1921; Fulcher 1925; Tammann & Hesse 1926): log10 h ¼ ATVF þ BTVF =(T  T0 )

(4)

Golden Pumice In the case of the golden pumice it was possible to analyse both the phenocrysts and the microlites distributed in the interstitial glass. The crystallinity of the pumice erupted during 5 April 2003 is 0.07, and volumetric number densities (NVc) are 7.3  105 and 1.4  1011 mm23 for phenocrysts and microlites, respectively. Data are uniformly and linearly distributed in the CSD plot (Fig. 10c), the intercepts n8 of the trendlines are 1.3  104 mm24 (phenocrysts) and 3.4  108 mm24 (microlites), which leads to a residence time of 4 years in the case of phenocrysts (using an average growth rate of 1029 mm s21), and 2 h for the microlites (with a growth rate of 3.8  1027 mm s21). Compared with scorias and lavas, the vesicle content is much higher (c. 0.53) and the bubble distance is c. 170.0 mm, which leads to a time of exsolution of 15 h in the case of CO2 and of only 12 min in the case of H2O.

Melt and magma viscosity An indirect method for estimating viscosities is the application of the so-called modified Einstein– Roscoe equations (MER) (Einstein 1911; Marsh

where h is the viscosity of the melt (Pa s), T is the temperature (K) and ATVF, BTVF and T0 are adjustable parameters known as the shift factor, the nonArrhenian pseudo-activation energy and the TVF temperature, respectively. Their data also show that the liquid–glass transition can be easily achieved at relatively low temperatures. Unfortunately, their model does not fit the experimental data on Hawaiian tholeiite performed by Shaw (1969) at near-liquidus and superliquidus temperatures (Fig. 11b) and does not take into account the water content within the melt. Thus, assuming that the ratios of hydrous to anhydrous viscosities, calculated with the Shaw model hold within the temperature range considered, we calculated these parameters for Stromboli basalt at variable temperatures and H2O contents (up to 3.5 wt%). Then, from the ‘anhydrous viscosities’ obtained from the Giordano & Dingwell model at lower temperatures (below the glass transition), we calculated the values of the ‘corresponding’ viscosities of the hydrous melt. Next, we selected solely those viscosities at-and-below the glass transition (where h ¼ 1012 Pa s or higher) together with those of Shaw at T  12008C, and we interpolated, by multiple

52

C. CIGOLINI ET AL.

Fig. 11. Viscosity plot for the dry Hawaiian Tholeiite Basalt: (a) comparison of the model of Shaw (1972) (black line) with the one of Giordano & Dingwell (2003) (dashed line) and the experimental measurements of Shaw (1969) (grey squares). (b) Comparison with the modified TVF model with the one of Giordano & Dingwell (2003) and the experimental measurements of Shaw (1969). The crystallinity of the basalt is reported on the right side of each curve.

regression, a ‘modified’ Tamman –Vogel– Fulcher equation, which integrates the cited models, and includes a term for water content: log10 h ¼ ATVF þ ½BTVF =(T  T0 )  CXw

(5)

which is of the same type as Equation 4 with the additional parameter C, and where Xw is the mole fraction of water in the silicate melt. We expanded the model by considering the effect of crystallinity in the MER equation proposed by Pinkerton &

STROMBOLI: FROM MANTLE TO ERUPTION

Table 7. Parameters for the modified TVF equation for Stromboli basalts (see text) Sample

Pumice April 2003f

ATVFa BTVFb T0c Cd Re

21.65 2974.97 703.31 19.18 0.99

s 0.33 273.05 14.49 1.92

Lava 2003g 21.41 3136.98 691.37 19.77 0.99

s 0.35 300.95 16.05 1.85

a Shift factor; bnon-Arrhenian pseudo-activation energy; cTVF temperature; dnew adjustable parameter for the water content; e linear correlation coefficient; s, standard deviation. fAverage of analyses 12 –17 in Table 5. gAverage of analyses 1–6 in Table 5.

Stevenson (1992), which is preferred to the one proposed by Ishibashi & Sato (2007) due to the low average axial ratios of the plagioclase crystals in our samples (cf. Ishibashi & Sato 2007, for details). The plot for viscosities of the Hawaiian tholeiite is reported in Figure 11b, and shows a good agreement with the experimental data of Shaw for the given degree of crystallinity (Shaw 1969). Table 7 summarizes the parameters of Equation 5, which have been used to calculate the viscosities for the golden pumice (average of analyses 12 –17 in Table 5) and the lavas (average of analyses 1– 6 in Table 5) erupted during the 2003 events. The results are summarized in Table 8. The viscosity of the melts was initially calculated as anhydrous, using Equation 5 at three different temperatures (1200, 1150 and 1100 ºC), and the results range from 1.6  102 to 6.2  102 Pa s (golden pumice) and from 4.0  102 to 1.6  103 Pa s (recent lavas). Applying the equation proposed by Pinkerton & Stevenson (1992), we then calculated the viscosities considering the effect of the crystallinity of the rocks (0.07 for the golden pumice and 0.38 for recent lavas). As a consequence, the values rise by almost one order of magnitude, ranging from 2.0  103 to 7.6  103 Pa s (golden

53

pumice) and from 8.0  103 to 3.1  104 Pa s (recent lavas). The latter values are consistent with the estimated viscosities based on field measurements on active lava flows made in March 2007 and February– March 2003 (C.C. unpublished data). Finally, adding 2 wt% H2O to bulk compositions, viscosities decrease of one order of magnitude: the hydrous basaltic melt shows an extremely low viscosity from 1.1  101 to 4.1  101 Pa s (golden pumice) and from 2.2  101 to 8.7  101 Pa s (recent lavas) for the above temperatures. If we take into consideration the rock crystallinity, viscosities vary from 1.3  102 to 5.0  102 (golden pumice) and from 4.5  102 to 1.7  103 (recent lavas).

Background petrology and phase relationships Me`trich et al. (2001) and Bertagnini et al. (2003) found that primitive melt inclusions in olivine of the golden pumices contain 2–3.4 wt% H2O and 890–1700 ppm CO2. In their model, this gas-rich magma is stored within the middle-lower part of the chamber and during its ascent leads to the formation of crystal-richer horizons produced by decompression-driven crystallization. The materials which reflect this environment of formation are lapilli and scoria bombs ejected during the typically ‘mild and persistent’ Strombolian activity. Conversely, the undegassed magma residing at lower levels is represented by the ‘golden pumices’, which are solely erupted during major and paroxysmal explosions. However, mingling and interactions of dark-coloured and crystal-rich scoria with goldencoloured crystal-poor pumice have also been operative (e.g. Francalanci et al. 1999; Me`trich et al. 2001, 2005; Landi et al. 2004). On experimental grounds, Di Carlo et al. (2006) investigated phase relationships in melts of Stromboli golden pumices. Experimental runs were performed at temperatures of 1175–1050 8C for

Table 8. Viscosities (Pa s) calculated at 1200, 1150 and 1100 8C using the modified TVF Equation 5 Pumice April 2003d T (8C) h meltadry (Pa s) h meltdry þ fbxx (Pa s) h melthydrousc (Pa s) h melthydrous þ fbxx (Pa s)

1200 1.6  102 2.0  103 1.1  101 1.3  102

1150 3.0  102 3.7  103 2.0  101 2.5  102

1100 6.2  102 7.6  103 4.1  101 5.0  102

Lava 2003e 1200 4.0  102 8.0  103 2.2  101 4.5  102

1150 7.5  102 1.5  104 4.2  101 8.4  102

1100 1.6  103 3.1  104 8.7  101 1.7  103

a Viscosity of the anhydrous liquid phase calculated using the mod-TVF Equation 5; bviscosity calculated applying Pinkerton & Stevenson (1992), taking into account fxx (see Table 6); cviscosity of the hydrous liquid phase (2 wt% H2O) calculated using the mod-TVF Equation 5. dAverage of analyses 12 –17 in Table 5. eAverage of analyses 1–6 in Table 5.

54

C. CIGOLINI ET AL.

pressures ranging from 400 to 50 MPa, and variable H2O contents (5.5–1.2 wt%) and oxygen fugacities (within 2 log units above the Ni –NiO buffer). Their results show that diopsidic pyroxene is a liquidus phase from 0.4 GPa down to 0.15 GPa at temperatures of 1175–1150 8C, followed by olivine, and eventually crystallizing pyroxene and plagioclase + Fe–Ti oxide at lower pressures and temperatures. Olivine may be a liquidus phase between 0.15 and 0.10 GPa for temperatures reaching 1100 8C. This is essentially in agreement with the cooling experiments of Conte et al. (2004) performed at 1 atm on analogue bulk compositions. According to their data, olivine is a liquidus phase between 1175 and 1150 8C, and is followed by a Ca-poorer pyroxene coexisting with plagioclase at lower temperatures (down to 1100 8C). Di Carlo et al. (2006) further concludes that the magma representative of the ‘golden pumices’ is essentially an undegassed melt that, during decompression, may not have experienced crystallization on the way to the surface. In turn, lava and scorias represent a degassed magma as originally suggested by the previously cited authors. By testing the hydrous primitive melt compositions with the computer code MELTS (Ghiorso & Sack 1995), we confirmed that clinopyroxene is a liquidus phase above 0.25 GPa and 1175 8C, whereas olivine becomes a liquidus phase at lower pressure.

Mg2 SiO4 þ SiO2 ¼ Mg2 Si2 O6 liq CEn Fo (liq)

New thermobarometric constraints In the attempt to constrain the P –T regimes associated with the storage of primitive Stromboli magmas we first used the compositions of the ‘natural assemblage’ in solving the solid-melt equilibrium Mg2 SiO4 ¼ Mg2 SiO4 Fo (sol) Fo (liq)

Dixon’s model (see also Dixon & Stolper 1995) and refer to the fractions of water and carbon dioxide in the binary mixture dissolved in the m ¼ 1 (see the Appendix melt: so that XHm2 O þ XCO 2 for details). The degree of oxidation of the melt phase has been estimated according to Kress & Carmichael (1991) along the Ni–NiO buffer of Hu¨bner & Sato (1970), which has been shown to be appropriate for Stromboli magma (Laiolo & Cigolini 2006). Thermobarometric calculations indicate that forsteritic olivines (Fo88 – 84) are in equilibrium with their host hydrous high-K calc-alkaline basaltic melts (recalculated by Me`trich et al. 2001, to take into account postrapping crystallization of olivine) at pressures of 0.18 – 0.2 GPa and temperatures ranging from 1200 to 1220 8C (only one sample equilibrated at 0.32 GPa and similar temperatures). The results of these calculations are given in Table 9. The calculated mole fractions for the H2O–CO2 mixture dissolved in the melt shows that Strombolian m ¼ 0:72  0:57), fluids are enriched in CO2 (XCO 2 and are in good agreement with the data on dry gases sampled along summit fractures by H. Tazieff and M. Ripepe (Martini et al. 1996). To test the consistency of the above estimates we constructed a P– T grid by introducing the following reactions:

(6)

for primitive melt inclusions found in olivines of the golden pumice which have been reported by Me`trich et al. (2001). Their FTIR measurements show that H2O and CO2 range from 2.3 to 2.8, and from 0.089 to 0.187 wt%, respectively. Thermobarometric estimates for the above reaction have been obtained using the thermodynamic data of Berman (1988), the solution models of Sack & Ghiorso (1989) for olivine, and of Ghiorso & Sacks (1995) for forsterite as a liquid component. We solved for P and T at the given fluid concentrations (inserted in wt% as input data, together with the melt compositions) by minimizing the free energy of the above reaction and solving simultaneously the solubility equations of Dixon (1997). m can be then retrieved according to XHm2 O and XCO 2

(7)

CaAl2 Si2 O8 þ 2Mg2 SiO4 þ SiO2 An liq Fo (sol) ¼ 2Mg2 Si2 O6 þ CaAl2 SiO6 CaTs CEn

(8)

CaAl2 Si2 O8 ¼ CaAl2 SiO6 þ SiO2 An CaTs liq

(9)

for variable H2O contents within the melt (up to 3.5 wt%, which is the highest water content measured in primitive melt inclusions; Bertagnini et al. 2003). We used the pumice bulk composition (Table 4, an. 11) for estimating the liquid component in reactions (6) –(8) together with the appropriate mineral compositions consistent with the previously discussed frequency mode (Table 1 an. 13 for olivine; Table 2 an. 12 for clinopyroxene; and Table 3 an. 3 for plagioclase). Activities for pyroxene and plagioclase were calculated according to the solution models of Gasparik (1984, 1990) and Newton et al. (1980), respectively. The activity of the liquid components was calculated according to Ghiorso & Sack (1995). According to our calculations, diopsidic clinopyroxene is a liquidus phase above 0.28 GPa followed by olivine at lower pressures. The intersection

GS GS SG GS HS þ R Model

HS, Hu¨bner & Sato (1970) taking into account the effects of pressure on the Ni –NiO buffer according to the data of Robie et al. (1979). GS, Ghiorso & Sack (1995); SG, Sack & Ghiorso (1989); D, Dixon (1997).

D D

0.58 0.73 0.60 0.73 0.42 0.27 0.40 0.27 1216 1221 1199 1205 0.183 0.325 0.209 0.223 0.0832 0.0938 0.0904 0.0699 0.0504 0.0498 0.0469 0.0503 0.764 0.774 0.774 0.812 20.4477 20.4916 20.4904 20.5225 0.0649 0.1146 0.0748 0.0795 0.0894 0.1689 0.1031 0.1107 2.7 2.8 2.8 2.3 0.837 0.846 0.846 0.88 St82p-oln50 St82p-oln9 St82p-oln52a2 St79p-oln30

T(8C) P(GPa) aw aFo(liq) XFo

H2O%

CO2%

DNi –NiO

logaSiO2

aFo(sol)

Calculated parameters Input data Sample

Table 9. Summary of calculations on the data of Me`trich et al. (2001) by applying reaction (6). Melt compositions are part of the input data as well

XHm2 O

m XCO 2

STROMBOLI: FROM MANTLE TO ERUPTION

55

of reactions (6) and (7) is consistent with the assemblage olivine –Ca-rich clinopyroxene melt at c. 0.290 GPa and 1210 8C for water contents of 3.5 wt%. The use of the whole spectrum of other mineral phases, compatible with frequency modes reported in Fig. 5, together with the other pumice compositions reported in Table 4, will lead to similar temperatures (+10 8C) and pressures (+0.02 GPa). These phases (olivine and diopsidic pyroxene) will crystallize with plagioclase (An80 – 78) at c. 0.25 GPa and temperatures 1200 8C. These regimes are probably representative of the lower part of the reservoir intruded by the primitive magma. Reaction (8) is a reliable geobarometer and shows that magma batches may be stored and, undergoing degassing (thus lowering their H2O and CO2 contents), may be allowed to crystallize at nearly isobaric conditions before entering the conduit (Laiolo & Cigolini 2006). The use of the primitive melt compositions of Me`trich et al. (2001), together with the above mineral phases, will lead to nearly identical temperatures and slightly (c. 0.02 GPa) lower pressures. The lower P –T regimes are constrained by reaction (9) where an augitic pyroxene is crystallizing with plagioclase (An68) and olivine will not longer be stable at water concentrations 1 wt% (Stevenson & Blake 1998). In this case we used the average glass composition of the scorias (Table 5, an. 8) together with the mean mineral data reported in Tables 2 and 3 (an. 4 for the augitic clinopyroxene, and an. 4 for plagioclase, respectively), to obtain 0.15–0.17 GPa for temperatures approaching 1100 8C, which is the average effusion temperature of current Stromboli lava (M. Coltelli and P. Scarlato measured 1090 8C at the vent, in March 2003). The results are summarized in Fig. 12, where we also outline possible magma P–T paths. Lines a and b confine the possible paths within a given distinct P– T domain (Fig. 12a & b). The magma batch leading to the formation of lavas and scorias undergoes storage and degassing (down to c. 0.5 wt% H2O) at about 0.15 GPa during isothermal and/or adiabatic decompression (approaching the effusion temperature of 1100 8C, Fig. 12a). In contrast, hydrous pumices (2  H2O  3.5 wt%) are ejected during the isothermal decompression of a higher temperature batch (c. 1130 –1200 8C, Fig. 12b). In order to provide a reliable model for the Stromboli plumbing system, we revisited the thermobarometry of San Bartolo basalts. In this case we considered the interaction of ultramafic materials with primitive magma represented by the chemical composition of the golden pumices (Tables 4 and 11). The mineral phases are those reported in Table 12 of Laiolo & Cigolini (2006).

56

C. CIGOLINI ET AL.

Fig. 12. Thermobarometry of lavas, scorias and pumices erupted at Stromboli during the 2002–2003 eruptive cycle and P– T regimes for the interaction between primitive Stromboli lavas and the ultramafic materials. (a) P –T estimates for the crystallization of lava and scorias. The graph summarizes the P– T grid obtained by solving thermodynamically reactions (6)– (9), indicated by the large numbers in parentheses. Smaller numbers are the water contents within the melt (wt% H2O). Best solutions are outlined by rectangles (see text). The high-temperature and pressure rectangle identifies the initial P–T conditions for the crystallization of the ferromagnesian phases (olivine and clinopyroxene) the lower T and P rectangle identifies the equilibrium conditions for the rims of phenocrysts coexisting with the interstitial glass (with 0.5 wt% H2O). (b) P–T estimates for the crystallization of the mineral phases of the pumices.

STROMBOLI: FROM MANTLE TO ERUPTION

57

Thermodynamic solutions were obtained according to reaction

San Bartolo Basalts has been obtained by introducing the following reactions:

Mg2 SiO4 þ SiO2 ¼ Mg2 Si2 O6 liq CEn Fo (sol)

CaAl2 Si2 O8 þ Mg2 SiO4 ¼ Mg2 Si2 O6 þ CaAl2 SiO6 CaTs An Fo CEn

(10)

which is similar to reaction (7) but includes solid olivine (instead of the Fo component in the melt). We also considered the following reaction for diopside þ olivine interacting with a silicate melt CaMgSi2 O6 ¼ 1=2 SiO2 þ 1=2 Mg2 SiO4 þ CaSiO3 liq Fo (liq) Wo (liq) Di (11) Calculations show that a primitive Stromboli magma equilibrates with mantle wherlitic assemblages at temperatures of 1260 –1240 8C and pressures of 1.1–0.65 GPa, for water contents in the melt ranging from 3.5 to 0.5 wt% (Fig. 12c). These results give lower equilibration pressure than those obtained by Laiolo & Cigolini (2006), who considered the San Bartolo basalt as a melt phase. Although their earlier approach may at first appear more sensible, we have to point out that the petrochemical evolution of the San Bartolo basalts has been strongly affected by the onset of an ‘assimilation reaction’ (Bowen 1928; McBirney 1979) during the inclusion and disaggregation of mafic and ultramafic materials. Thus, the use of a more primitive melt in the above reactions is more appropriate for decoding melt-wallrock interactions within the upper mantle. Ultramafic material may have equilibrated at variable pressures, which are, in turn, negatively correlated with the water content of the basaltic melt: in case of ‘deep’ degassing through microcracking, ultramafic material may have equilibrated at lower P –T regimes. In any case, equilibration occurred within the upper mantle (the Moho being located at a depth of 15– 17 km, according to Pontevivo & Panza 2006; Barberi et al. 2007). Finally, the thermobarometric estimates for the crystallization of gabbroic nodules included in

(12) CaAl2 Si2 O8 þMg2 SiO4 ¼ Mg2 Si2 O6 þ CaAl2 SiO6 CaTs An Fo OEn (13) These do not include a melt phase. In this case equilibration temperatures range from 1150 to 1120 8C for pressures between 0.43 and 0.32 GPa (Fig. 12c). These regimes are basically consistent, at least in terms of pressure, with the lower part of the Stromboli magma reservoir. Therefore, it is further confirmed that these materials were originally cumulates of earlier Stromboli magmas which have been deposited at the base of the chamber (as shown by Laiolo & Cigolini 2006).

Stromboli plumbing system and the shape of the magma chamber In the light of the above, we may summarize decompression trajectories of Stromboli magma from the upper mantle to the surface. The deeper part of the feeding system is located at 24– 34 km within the upper mantle (Fig. 13). In this region, magma flow is probably channelled along fracture zones that may converge into a feeder dyke (e.g. Shaw 1980). However, the source region of Stromboli primitive melts is likely to be located deeper in the mantle wedge above the Benioff Zone, the latter being positioned at about 250–300 km depth (Panza & Pontevivo 2002). New analyses of teleseismic data locate the Moho at about 15 –17 km below the volcano (Pontevivo & Panza 2006; Barberi et al. 2007). During their ascent, basaltic magmas will interact with lower crust materials represented by cumulates of earlier Stromboli-type basalts (13 –10 km depth). This zone is also the section of the plumbing system where the feeder dyke is entering the chamber. Thermobarometric estimates on ‘golden pumices’

Fig. 12. (Continued) In this case we considered an undegassed hydrous melt (with 2  H2O  3.5 wt%). (c) P –T regimes for the equilibration of Stromboli ‘primitive’ basalt (represented by the melt of the pumices) with ultramafic materials and with mafic nodules. Dashed arrows in (a) and (b) outline the magma P –T path. Lines a and b confine the possible paths within a given domain. The magma batch leading to the formation of lavas and scorias undergoes storage and degassing (down to c. 0.5 wt% H2O) at about 0.15 GPa before isothermal and/or adiabatic decompression (at about 1100 8C). In contrast hydrous pumices (2  wt% H2O  3.5) are ejected during isothermal decompression of a higher temperature batch (c. 1130–1200 8C).

58

C. CIGOLINI ET AL.

Fig. 13. Summary of the whole P –T path for the ascent of Stromboli magma. Geophysical data are from Pontevivo & Panza (2006) and Barberi et al. (2007). See text for details.

suggest that the lower part of the magma chamber is located at about 11.0 km (+0.7) below the summit, based on a magmastatic gradient consistent with a density of 2700 kg m23 (calculated according to Lange & Carmichael 1990; for Stromboli basalts), so that 0.1 GPa is equivalent to 3800 m of lithostatic load. Thus, the current reservoir has a vertical extent of 5.4 –6 km (from about 0.29 + 0.02 to approximately 0.15 GPa, which is the lower estimate for the top of the reservoir) up to a depth of about 5.7 km below the summit, where the magma is entering the conduit. During the typical mild Strombolian activity the magma will crystallize and degas at depth before being erupted (approaching anhydrous conditions at c. 0.5 wt% H2O). In these cases, eruption rates are ruled by the steady-state intrusion of a primitive high-K basalt at the bottom of the chamber (approaching 0.3 m3 s21, according to Ripepe et al. 2005). During effusive cycles magma input dramatically increases, and the gas-rich component may not be accommodated within the reservoir, eventually generating paroxysmal explosions with the ejection of golden pumices. If we assume a spherical shape, the Stromboli magma chamber will have an approximate radius of about 2700– 3000 m, which is similar to that estimated by Murru et al. (1999) for the Etna upper reservoir. Thus the estimated volume would reach 8.2–11  1010 m3. Alternatively, we may assume an ellipsoidal shape for the reservoir, being geometrically concordant with the regional stress distribution. According to Tibaldi et al. (2003), the

summit ‘feeder’ dyke is concordant with the orientation of summit craters, located along the N408E normal fault. Therefore, smax ¼ s1 will be vertical, s2 is horizontal along the above direction, and s3 is the extensional component onto the horizontal plane (trending N508W, i.e. normal to the s1 – s2 plane). For pure extensional regimes in volcanic areas, consistent with the Griffith plus Coulomb criteria for the development of pure extensional fractures (e.g. Shaw 1980, pp. 218–225), s1 ¼ 3K, s2 ¼ K and s3 ¼ 2K, where K is the rock tensile strength. Thus, by assuming that the length of the ellipsoid’s axes will be proportional to the magnitude of the stress regimes, we may obtain a rough estimate for the geometry of an ellipsoidal reservoir. In the light of our thermobarometric estimates, the major axis would be vertical with a half-length of 2700–3000 m, being controlled by smax ¼ s1. The intermediate axis will be concordant with the orientation of summit craters (s2) and equal to about 900–1000 m (one-third of the vertical halflength; cf. Shaw 1980). The other axis, parallel to the extensional stress component s3 (N508W), would have the same half-length of the latter, leading to a total volume of c. 9– 12  109 m3 for the ellipsoidal magma chamber (Fig. 14a). However, several other ellipsoidal shapes are possible, depending on stress distributions and related geometries (Gudmundsson 2006). Assuming a prolate ellipsoidal chamber having a circular section with a diameter of 600 m (i.e. the approximate distance of active vents during the last two effusive phases) and the above vertical half-length

STROMBOLI: FROM MANTLE TO ERUPTION

59

Fig. 14. The plumbing system of Stromboli volcano. An ellipsoidal magma chamber geometrically concordant with the regional stress distribution seems to be the best candidate for the magma reservoir below Stromboli volcano. Alternatively we considered a feeder dyke. Dimensions of the feeder dyke have been exaggerated.

of 2700–3000 m, we get an approximate conservative volume of c. 1  109 m3. A third possibility is to consider a ‘reservoir’ being represented by a feeder dyke (Fig. 14b), also concordant with the local stress distribution. The maximum average width measured in the field for Stromboli dikes is c. 7 m (A. Renzullli, pers. comm., 18 December 2006). For an average feeder dyke 3.5 m wide, consistent with the estimates of Ripepe & Gordeev (1999) for the dyke feeding the active craters, and extending to a depth of 11 km with a lateral length of 600 m (which matches the cited distance of active vents), we get a total volume of 2.3  107 m3. This volume is nearly twice the total volume erupted during the 2002– 2003 effusive cycle (c. 107 m3; Ripepe et al. 2005; Calvari et al. 2005), but it would be only a minor fraction when compared with the volume of the ellipsoid magma reservoir. In the following, we will discuss these models in greater detail.

Discussion and conclusions We attempted to decode the inner architecture of the plumbing system beneath Stromboli volcano.

At deeper levels, magma ascent is probably channelled through a system of fractures that converge to form a feeder dyke before entering the crust, at about 17 km depth (Barberi et al. 2007). However, the source region of Stromboli primitive melts is well deeper, since these magmas probably originate in the mantle wedge above the Benioff Zone and/or within the subducting slab itself, or both (e.g. Peccerillo 2001). The lower crust consists of gabbroic –anorthosite layers that were formed by fractionation of earlier Stromboli type basalts (Laiolo & Cigolini 2006). This zone extends up to 12–11 km depth, where the feeder dyke is entering the base of the currently active chamber (located at about 11 km depth from the summit). The geometry of this reservoir, which may partially depend on the volumetric flux rate, is dominated by the stress field that approaches pure extensional regimes (Tibaldi et al. 2003), thus suggesting an ellipsoidal shape (Holohan et al. 2005). In the light of thermobarometric estimates and simple geometric relations, consistent with the Griffith plus Coulomb criteria for the development of pure extensional fractures, the volume of the chamber should be roughly comprised between 1  109 and 9–12  109 m3.

60

C. CIGOLINI ET AL.

Fig. 15. Details of the possible shape of the magma chamber below Stromboli volcano. The ellipsoid is injected by a feeder dyke that evolves into a magma column (of lower density and viscosity) across the whole chamber. The convective regions, locally affected by turbulence, are placed laterally in respect to the active column. Crystallization occurs preferentially in these sectors and in proximity of the wallrocks. Cumulates are stored at the base of the reservoir. Dimensions of the feeder dyke and magma column have been exaggerated (see text for details).

Alternatively, a single feeder dyke with a geometry consistent with field and geophysical data, as well as with reported thermobarometric constraints, will lead to a volume of c. 2.3  107 m3. Only about one-half of this volume was extruded during the 2002–2003 effusive cycle (Ripepe et al. 2005; Calvari et al. 2005); the remaining 107 m3 simply refilled the chamber and/or the evacuated part of the dyke. This possible scenario leads us to infer that the Stromboli magma reservoir may be realistically described by a convective ellipsoidal magma chamber ‘injected’ by an active feeder dyke of undegassed magma of higher temperature (c. 1200 8C), lower density and lower viscosity. This dyke will evolve into a magma column inside the chamber and subdivides the reservoir into two nearly symmetric sectors (Fig. 15). This region

will be affected by the drag force exerted by the ascending magma column and turbulent convection may take place, as postulated by Huppert & Sparks (1988) and Jellinek & Kerr (1999). Crystallization would occur preferentially in these convective regions and in proximity of the wallrocks. Under, steady-state regimes, these portions of the magma chamber may exchange matter with the inner column both by moderate mixing and/or diffusive phenomena, the latter being mainly restricted to the interface zone. This model seems to explain the coexistence of cumulitic materials, crystal mushes, crystal clots and glomerophorphyric aggregates in Stromboli basalts. It may explain complex zoning in plagioclase (e.g. Landi et al. 2004, 2006) as well as crystal recycling and, more importantly, it provides answers to the ‘apparent’ discrepancies

STROMBOLI: FROM MANTLE TO ERUPTION

connected with timescales of complex magmatic processes (Hawkesworth et al. 2004). It also gives a plausible explanation for the results obtained on CSD distributions since residence times for the nucleation of microlites within the ‘golden pumices’ (a few hours prior to their eruption) are consistent with the ascent of a melt through the active feeder dyke, directly connected to the conduit. In turn, the viscoelastic behaviour of magma chambers in response to seismic transients, which may eventually trigger volcanic eruptions (e.g. Manga & Brodsky 2006; Cigolini et al. 2007), is only compatible with the existence of a well developed magma reservoir. Thus, the most feasible answer is that these structural features both coexist and actively modulate volcanic activity. An apparently controversial issue is the lower estimate of the Stromboli magma chamber volumes provided by Gauthier & Condomines (1999), who analysed short-lived isotopes and calculated the average residence times for a set of samples. By assuming a steady-state magma flux (10 000– 50 000 m3/day, typical of the mild Strombolian activity, Harris & Stevenson 1997), they calculated a volume of 1.3+1.2  107 m3, which scales well with the erupted magma volume during the 2002– 2003 effusive cycle apparently supporting the ‘single’ dyke model. Similarly, if we doubled the magma flux (which is likely to occur during major effusive cycles), we would get a volume similar to the total volume of the dyke (the latter being roughly 2.3107 m3, as indicated above) down to a depth of 11 km. In our view this simply supports the idea that the feeder dyke, which penetrates as a magma column, is the most active part of the plumbing system. The chamber will simply modulate, through a minor expansion, the mechanical dynamics of the open system during ‘dyke’ injection. The bulk rheology of the chamber is estimated to be in the range of 103 Pa s (due to the higher crystallinity of the stored convective magma), whereas the magma in the feeder dyke would be, depending on the water content within the fresh magma, at least one order of magnitude less. Then, in terms of volume, the part of the dyke confined within the chamber is at the most 1% of the total volume of the ellipsoid. Similarly, the extruded magma volume during the major effusive cycles would also reach, at the most, 1% of the bulk chamber volume. Since its extrusion simply represents the dynamic response of the chamber to an overpressure, probably induced by increasing intrusion rates at the base of the reservoir, we may infer that this would be the ‘maximum’ bulk expansion of the chamber during major eruptive cycles. This estimate is in

61

good agreement with the increase in the injected magma mass (0.5% in terms of volume) during the transition from ‘closed’ to open system at Arenal volcano. This phase was followed by the onset of the vigorous Strombolian activity alternated with the extrusion of small lava flows (Cigolini 1998, p. 303). How do the paroxysmal explosions fit into this model? In our view the most feasible approach to explain these events is to consider the mechanical effects of elastic rebound occurring at the chamber walls. It is well known that magma chambers are viscoelastic bodies (e.g. Bonafede et al. 1986; Dragoni & Magnanesi 1989; Jellinek & De Paolo 2003) and their response to seismic triggering and/or magma injection is ruled by the rheological properties of the wallrocks, by the local stress field, and by the elastic properties of crustal rocks in the near field. The convective regions are located laterally with respect to the active column, and their efficiency will be increased due to the drag force exerted by the ascending magma column. Crystallization would occur preferentially in the convective regions and in proximity to the wallrocks. This explains the coexistence of cumulitic materials, crystal mushes, crystal clots and glomerophorphyric aggregates in Stromboli basalts. However, an increase of magma injection within the active dyke will cause an almost negligible ‘inflation’ of the bulk magma chamber (1% of its original volume at the most), since the increase in volume of the ‘inner dyke’ will be partitioned over the whole magma mass. During this process, the magma flux within the magma column is increased, and inflation is accompanied by a significant growth in the rate of exsolution of gaseous phases, because gases will become unsoluble within the upper portions of the column, resulting in a system that approaches volatile saturation. The system will now be pressurized: at Stromboli this normally coincides with the onset of lava effusion that may persist for a few weeks or months. Most of the overpressure will be selectively exerted on the chamber walls. The ceasing of ‘anomalous’ intrusion rates at the base of the chamber, coupled with higher discharge rates, will progressively depressurize the chamber. Consequently, a critical threshold is reached when the stress transferred to the chamber walls is dynamically released: at this point the walls themselves will undergo a nearly instantaneous elastic rebound, and contract in the attempt to recover their original pre-eruptive geometry (Murase 1962). These dynamics will squeeze up portions of the undegassed magma column, triggering a paroxysmal explosion with the ejection of ‘golden pumices’. Following these events, further depressurization of the feeding system will cause

62

C. CIGOLINI ET AL.

magma column instabilities associated with extensive mixing within the upper chamber, and apparently a prelude (due to an increase in viscosity and crystallinity of the magma) to the ceasing of the major eruptive cycle. Normal steady-state mild Strombolian activity will then be resumed as the inner equilibrium is finally restored. Future work should be concentrated in refining the modelling of these dynamic processes that we attempted to decode in probing this unique dynamic system. The research was funded by the National Institute for Geophysics and Volcanology (INGV, Project V2), and the Italian Civil Defence. Additional funds were provided by MIUR (Ministero dell’Universita` e della Ricerca). We also thank the staff of the Italian Civil Defence for their logistical support, in particular C. Cardaci and R. Colozza. We are indebted to M. Ripepe, K. Cashman, M. Rosi and S. Calvari for the stimulating discussions throughout these years. J. Hammer and G. Ventura provided brilliant reviews of an earlier draft of the paper. The volcano guides helped us in the field. We wish to thank M. Schmidt and P. Ulmer of the IMP-ETH Zurich (Swiss Federal Institute of Technology) for their hospitality to S. Bertolino and for the use of the analytical facilities.

Appendix As an example, we hereby outline the thermodynamic solution for the following solid– melt equilibrium [reaction (6) in the text]: Mg2 SiO4 ¼ Mg2 SiO4 Fo (sol) Fo (liq)

(A1)

which has been applied to investigate the equilibration pressure and temperature for the melt inclusions in olivine of the ‘golden pumices’. These juvenile tephra are ejected during major and/or paroxysmal eruptions of Stromboli volcano. The Gibbs free energy for the above reaction is represented by DGP;T ¼ DG1;T 0 þ RT ln K6 þ 0

ðP

DV dP

(A2)

1

where DG1,T 0 is the Gibbs free energy for the reaction calculated at 1 bar and at the temperature of interest (T ) by using the data of Berman (1988). R is the gas constant sol and K6 ¼ aliq Fo =aFo is the equilibrium constant which may be calculated from the appropriate solution models, knowing the chemical compositions of the phases considered. The last term is the difference of the molar volumes for the above phases integrated within the given pressure range. This can be solved numerically at given equilibrium conditions by using the data of Berman (1988), and those listed by Ghiorso & Sack for the melt component (Foliq), which account for thermal expansion and compressibility. In this specific case, activities for the solid phase were calculated according to the symmetric solution model of Sack & Ghiorso (1989), whereas the activity of olivine as a melt component was calculated following Ghiorso & Sack (1995). It should be emphasized that activities of the melt components change with the water content in the melt. The solution of the above equilibrium was obtained by minimizing the free energy DGP,T in Equation A2, which 0 goes to zero at equilibrium conditions, by fixing a given value of T and calculating P (or vice versa). The same procedure has been applied to the set of reactions presented in text, which have been solved for a given P –T range and have been used for estimating the depth and extension of the Stromboli magma reservoir. A summary with a list of equilibrium constants is given in Table A1. In addition, the above equilibrium [(A1), i.e. reaction m (6) in the text] can be solved for P, T, XHm2 O and XCO for 2 the given fluid concentrations (in wt% as input data) by minimizing the free energy of the above reaction and solving simultaneously the solubility equations of Dixon (1997). The insertion of the molar heat of solution for H2O in the melt (determined on albite melt according to Silver & Stolper 1985) for polythermal calculations has been shown to be negligible with respect to free energy m minimization. Notations XHm2 O and XCO refer to the frac2 tions of water and carbon dioxide in the binary mixture m dissolved in the melt, so that XHm2 O þ XCO ¼ 1 (Dixon & 2 Stolper 1995; and see text and Table 9 for the pertinent numerical solutions). In Table A2 and Table A3, we report the thermodynamic properties of end-member solids and liquids which have been used in calculations. For reference, we summarize (Tables A4–A6) the numerical solutions for the reactions considered in the present thermodynamic treatment.

Table A1. Summary of the equilibrium contants used in calculations. Subscripts represent the reactions numbered in the text sol K6 ¼ alip Fo =aFo

liq K10 ¼ aCEn =asol Fo aSiO2

liq K7 ¼ aCEn =aliq Fo aSiO2 2 2 liq K8 ¼ ðaCEn Þ aCaTs =aAn ðasol Fo Þ aSiO2 liq K9 ¼ aCaTs aSiO2 =aAn

1=2 liq 1=2 liq K11 ¼ ðaliq ðaFo Þ aWo =aDi SiO2 Þ K12 ¼ aCEn aCaTs =aAn asol Fo K13 ¼ aOEn aCaTs =aAn asol Fo

Table A2. Thermodynamic properties of endmember solids. Data are from Berman (1988) Phase

CaAl2Si2O8 CaAl2SiO6 Mg2SiO6 Mg2SiO4 (CaMg)Si2O6 Mg2SiO6 CaSiO3

24,228,730 23,298,767 23,091,852 22,174,420 23,200,583 23,091,104 21,627,427

DS0f (J/8K)

200.19 140.75 132.65 94.01 142.5 132.34 85.279

V0 (J bar21) 10.075 6.356 6.262 4.366 6.62 6.266 4.016

Heat capacity coefficients

Volume coefficients

k0

k1  1022

k2  1025

k3  1027

v1  106

v2  1012

v3  106

v4  1010

439.37 310.7 279.92 238.64 305.41 333.16 141.16

237.34 216.72 29.94 220.013 216.049 224.012 24.172

0.00 274.553 288.004 0.00 271.66 245.412 258.576

231.702 94.878 107.142 211.624 92.184 55.83 94.074

21.272 20.870 21.498 20.791 20.872 21.498 21.245

3.176 2.171 0.894 1.351 1.707 0.894 3.113

10.918 22.250 49.312 29.464 27.795 49.312 28.18

41.985 52.863 149.34 88.633 83.082 149.34 0.00

Table A3. Thermodynamic properties of liquid components. Data are from Ghiorso & Sack (1995, and references therein) Melt component

SiOa2 Mg2SiO4 CaSiO3

Reference solid

Amorphous silica Forsterite Pseudowollastonite

Tf

1999 2163 1817

DS0f (J/8K)

4.46 57.2 31.5

DC0p (J/8K)

81.373 271 172.4

V 0 (J bar21 1673 8C) 2.69 4.98 4.347

@V 0 @T

!

 104

P

0.00 5.24 2.92

@V 0 @P

!

 105

T

21.89 21.35 21.55

@2 V 0 @P@T

!

 108

T

1.30 21.30 21.60

@2V 0 @P2

!

 1010

STROMBOLI: FROM MANTLE TO ERUPTION

Anorthite Ca-Tschermak Clinoenstatite Forsterite Diopside Orthoenstatite Pseudowollastonite

DH0f (J)

Formula

3.60 4.10 3.90

p For amorphous silica below 1480 8C astonite, DH0f ¼ 901554; DS0f ¼ 48.475; DC0p ¼ 127.3 2 10.777  1023T þ 4.3127  105/T 2 2 14638/ T (Richet et al. 1982).

a

63

64

Table A4. Summary of selected thermodynamic solutions for solid– liquid equilibria for pumices and scorias erupted during the paroxysmal event of 5 April 2003 at Stromboli volcano. See text for details log aSiO2 (0.5 wt%)

log aSiO2 (3.5 wt%)

33.4680 33.1385 32.8032 32.1159 31.4063 29.9223

— — 20.4377 20.4379 20.4380 20.4383 Ghiorso & Sack (1995)

20.5148 20.5152 20.5157 20.5165 20.5174 20.5190 Ghiorso & Sack (1995)

0.7463 237.1423 — 0.7420 236.6721 — 0.7380 236.1950 20.4210 0.7343 235.2204 20.4208 0.7309 234.2190 20.4206 0.7277 232.1374 20.4203 Sack & Ghiorso Ghiorso & Sack (1989) (1995) XFo ¼ 0.83: Table 1, an. 13; XCEn ¼ 0.40: Table 2 an. 12; melt composition Table 4, an. 11

20.4976 20.4979 20.4982 20.4987 20.4992 20.5002 Ghiorso & Sack (1995)

Reaction (6) 1200 1210 1220 1240 1260 1300 References

aFo

aCEn

0.7396 0.5690 0.7390 0.5306 0.7384 0.4973 0.7373 0.4683 0.7361 0.4427 0.7340 0.4201 Sack & Ghiorso Gasparik (1989) (1984, 1990) XFo ¼ 0.83: Table 1 an. 13; melt compositions Table 4, an. 11

Reaction (7) 1200 1210 1220 1240 1260 1300 References

P (MPa) H2O ¼ 0.5 wt% Fo(sol) ¼ Fo(liq) 51 317 611 1303

Fo(liq) þ SiO2 ¼ CEn 41 139 244 471

P (MPa) H2O ¼ 3.5 wt%

41 172 310 607 935 1709

219 273 329 445 569 839

C. CIGOLINI ET AL.

DGI,T 0 (K J)

T(8C)

Table A5. Summary of selected thermodynamic solutions for solid– liquid equilibria, reactions (8) and (9), for pumices [reaction (8)] and scoriae [reaction (9)] erupted during the 2002 –2003 major eruptive cycle of Stromboli volcano. See text for details T(8C)

0.7465 0.7419 0.7376 0.7336 0.7300 0.7265 0.7234 Newton et al. (1980) XFo ¼ 0.83: Table 1, an. 13; Reaction (9) 1000 1050 1100 1130 1160 References

aFo

aCEn

aCaTs

0.7463 0.5690 0.0469 0.7420 0.5306 0.0465 0.7380 0.4973 0.0462 0.7343 0.4683 0.0459 0.7309 0.4427 0.0457 0.7277 0.4201 0.0454 0.7248 0.4000 0.0452 Sack & Gasparik Gasparik Ghiorso (1984, (1984, (1989) 1990) 1990) XCEn ¼ 0.40: Table 2, an. 12; XAn ¼ 0.78:

DGI,T 0 (K J)

log aSiO2 (0.5 wt%)

31.0489 32.6565 34.3173 36.0314 37.7992 39.6207 41.4960

log aSiO2 (3.5 wt%)

20.4970 20.4986 20.5001 20.5014 20.5027 20.5038 20.5049 Ghiorso & Sack (1995) Table 3, an. 3; melt composition:

0.0422 0.0424 43.4950 0.0427 0.0420 43.9192 0.0431 0.0416 44.3326 0.0434 0.0413 44.5763 0.0436 0.0411 44.8171 Newton Gasparik et al. (1984, (1980) 1990) XCaTs Table 2, an. 4; XAn ¼ 0.68: Table 3, an. 4; glass composition Table 5, an. 8

20.4183 20.4163 20.4145 20.4134 20.4123 Ghiorso & Sack (1995)

P (MPa) H2O ¼ 0.5 wt%

P (MPa) H2O ¼ 3.5 wt%

An þ 2Fo þ SiO2 ¼ 2Cen þ CaTs 20.4279 282 20.4271 252 20.4264 224 20.4257 198 20.4251 175 20.4244 153 20.4238 134 Ghiorso & Sack (1995) Table 4, an. 11

An ¼ CaTs þ SiO2 20.4311 20.4295 20.4280 20.4268 20.4262 Ghiorso & Sack (1995)

376 276 174 111 48

323 296 271 249 228 210 194

347 244 139 53 10

STROMBOLI: FROM MANTLE TO ERUPTION

Reaction (8) 1000 1050 1100 1150 1200 1250 1300 References

aAn

an. ¼ analysis number.

65

66

Table A6. Summary of thermobarometric estimates obtained on the ultramafic materials hosted in S.Bartolo lavas. See text for details T(8C)

aDi

aFo

aFo

aCEn

DGI,T 0 (K J)

log aSiO2 (0.5 wt%)

0.85203 0.73956 47.159 — 0.85145 0.73801 46.372 — 0.85100 0.73468 44.505 20.392 0.85048 0.73224 42.965 20.392 0.84939 0.72960 41.114 20.391 0.84939 0.72711 39.169 20.390 0.84901 0.72533 37.651 20.390 0.84882 0.72447 37.133 20.390 Ghiorso Sack & Sack & & Sack Ghiorso Ghiorso (1995) (1989) (1989) Melt composition Table 4, an. 11: mineral compositions previously reported in Laiolo & Cigolini (2006, Table 12) Fo(sol) þ SiO2 ¼ CEn 0.74631 0.74245 26.230 20.394 0.73801 0.64950 25.014 20.393 0.73576 0.62627 23.674 20.392 0.73432 0.61178 23.534 20.392 0.73224 0.59137 23.392 20.391 0.73090 0.57859 23.105 20.391 0.72834 0.55481 22.813 20.391 0.72592 0.53317 22.215 20.390 0.72447 0.52060 22.139 20.390 Ghiorso Gasparik Sack & & Sack (1984, Ghiorso (1995) 1990) (1989) Melt composition Table 4, an. 11: mineral compositions previously reported in Laiolo & Cigolini (2006, table 12)

Reaction (11) 1000 1100 1130 1150 1180 1200 1240 1280 1305 References

an. ¼ analysis number.

P (GPa) H2O ¼ 0.5 wt%

P (GPa) H2O ¼ 3.5wt%

Di ¼ 1/2SiO2(liq) þ 1/2Ol(liq) þ Wo(liq) 20.466 — 0.008 20.466 — 0.118 20.467 0.024 0.387 20.468 0.223 0.617 20.469 0.472 0.907 20.470 0.748 1.230 20.470 0.975 1.498 20.471 1.097 1.641 Ghiorso & Sack (1995)

20.463 20.466 20.467 20.467 20.468 20.468 20.469 20.470 20.470 Ghiorso & Sack (1995)

0.458 0.536 0.563 0.582 0.612 0.633 0.677 0.725 0.757

0.716 0.838 0.879 0.907 0.951 0.982 1.046 1.113 1.157

C. CIGOLINI ET AL.

Reaction (10) 1080 1100 1145 1180 1220 1260 1290 1305 References

log aSiO2 (3.5 wt%)

STROMBOLI: FROM MANTLE TO ERUPTION

References A COCELLA , V. & T IBALDI , A. 2005. Dike propagation driven by volcano collapse: a general model tested at Stromboli, Italy. Geophysical Research Letters, 32, L08308, DOI: 10.1029/2004GL022248. A COCELLA , V., N ERI , M. & S CARLATO , P. 2006. Understanding shallow magma emplacement at volcanoes: Orthogonal feeder dikes during the 2002–2003 Stromboli (Italy) eruption. Geophysical Research Letters, 33, L17310, DOI: 10.1029/ 2006GL026862. A RMIENTI , P., F RANCALANCI , L. & L ANDI , P. 2007. Textural effects of steady state behaviour of the Stromboli feeding system. Journal of Volcanology and Geothermal Research, 160, 86– 98. A RRIGHI , S., R OSI , M., T ANGUY , J. & C OURTILLOT , V. 2004. Recent eruptive history of Stromboli (Aeolian Islands, Italy) determined from high-accuracy archeomagnetic dating. Geophysical Research Letters, 31, DOI: 10.1029/2004GL020627. A UGER , A., G ASPARINI , P., V IRIEUX , J. & Z OLLO , A. 2001. Seismic evidence of an extended magmatic sill under Mt. Vesuvius. Science, 294, 1510–1512. B ARBERI , F., I NNOCENTI , F., F ERRARA , G., K ELLER , J. & V ILLARI , L. 1974. Evolution of Eolian Arc volcanism (Southern Tyrrhenian Sea). Earth and Planetary Science Letters, 21, 269– 276. B ARBERI , F., R OSI , M. & S ODI , A. 1993. Volcanic hazard assessment at Stromboli based on review of historical data. Acta Vulcanologica, 3, 173 –187. B ARBERI , G., Z HANG , H., S CARFI , L., C OCINA , O., C ASTELLANO , M., C HIARABBA , M. & P ATANE` , D. 2007. Crustal evidence of a low velocity Vp and Vs volume beneath Stromboli Volcano, Italy. Geophysical Research Abstract, 9, Sref-Id: 1607-7962/gra/ EGU2007-A-02621. B ECCALUVA , L., G ABBIANELLI , G., L UCCHINI , F., R OSSI , P. L. & S AVELLI , C. 1985. Petrology and K/Ar ages of volcanics dredged from the Eolian seamounts: implications for geodynamic evolution of the Tyrrhenian basin. Earth and Planetary Science Letters, 74, 187 –208. B ERMAN , R. G. 1988. Internally consistent thermodynamic data for stoichiometric minerals in the system Na2O–K2O–CaO–MgO–FeO–Fe2O3 –Al2O3 –SiO2 – TiO2 –H2O– CO2. Journal of Petrology, 29, 445–522. B ERTAGNINI , A., M E` TRICH , N., L ANDI , P. & R OSI , M. 2003. Stromboli volcano (Aeolian Archipelago, Italy): an open window on the deep-feeding system of a steady state basaltic volcano. Journal of Geophysical Research, 108, 2336–2351. B ONACCORSO , A., C ALVARI , S., G ARFI , G., L ODATO , L. & P ATANE` , D. 2003. Dynamics of the December 2002 flank failure and tsunami at Stromboli volcano inferred by volcanological and geophysical observations. Geophysical Research Letters, 30, DOI: 10.1029/ 2003GL017702. B ONAFEDE , M., D RAGONI , M. & Q UARENI , F. 1986. Displacement and stress field produced by a centre of dilation and by a pressure source in a viscoelastic half-space: application to the study of ground deformation and seismic activity at Campi Flegrei.

67

Geophysical Journal of the Royal Astronomical Society, 87, 455–485. B OTTINGA , Y. & W EILL , D. F. 1972. The viscosity of magmatic silicate liquids: a model for calculation. American Journal of Science, 272, 438– 475. B OWEN , N. L. 1928. The Evolution of Igneous Rocks. Princeton University Press, Princeton, NJ, 334. C ACCAMO , D., N ERI , G., S ARAO , A. & W ISS , M. 1996. Estimates of stress directions by inversion of earthquake fault-plane solutions in Sicily. Geophysical Journal International, 125, 857–858. C ALVARI , S., S PAMPINATO , L. ET AL . 2005. Chronology and complex volcanic processes during the 2002– 2003 flank eruption at Stromboli volcano (Italy) reconstructed from direct observations and surveys with a handheld thermal camera. Journal of Geophysical Research, 110, DOI: 10.1029/2004JB003129. C ASHMAN , K. V. 1990. Textural constrains on the kinetics of crystallization of igneous rocks. In: N ICHOLLS , J. & R USSEL , J. K. (eds) Modern Methods of Igneous Petrology: Understanding Magmatic Processes, Reviews in Mineralogy, 24, 259– 314. C ASHMAN , K. V. & M ARSH , B. D. 1988. Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization II: Makaopuhi lava lake. Contribution to Mineralogy and Petrology, 99, 292– 305. C ASHMAN , K. V. & M C C ONNELL , S. M. 2005 Multiple levels of magma storage during the 1980 summer eruptions of Mount St. Helens, WA. Bulletin of Volcanology, 68, 57– 75. C IGOLINI , C. 1998. Intracrustal origin of Arenal basaltic andesite in the light of solid-melt interactions and related compositional buffering. Journal of Volcanology and Geothermal Research, 86, 277 –310. C IGOLINI , C. & K UDO , A. M. 1987. Xenoliths in recent basaltic andesite flows from Arenal Volcano, Costa Rica: inference on the composition of the lower crust. Contribution to Mineralogy and Petrology, 96, 381– 390. C IGOLINI , C., G ERVINO , G., B ONETTI , R., C ONTE , F., L AIOLO , M., C OPPOLA , D. & M ANZONI , A. 2005. Tracking precursors and degassing by radon monitoring during major eruptions at Stromboli Volcano (Aeolian Islands, Italy). Geophysical Research Letters, 32, L12308. C IGOLINI , C., L AIOLO , M. & C OPPOLA , D. 2007. Earthquake– volcano interactions detected from radon degassing at Stromboli (Italy). Earth and Planetary Science Letters, 257, 511– 525. C ONDOMINES , M., T ANGUY , J. C., K IEBER , G. & A LLEGRE , C. J. 1982. Magmatic evolution of a volcano studied by 230Th– 238U disequilibrium and trace element systematics: the Etna Case. Geochimica and Cosmochimica Acta, 46, 1397–1416. C ONTE , A. M., P ERINELLI , C. & T RIGILA , R. 2004. Cooling kinetics experiments on different Stromboli lavas: effect on crystal morphologies and phases composition. Journal of Volcanology and Geothermal Research, 155, 179–200. D E A STIS , G., V ENTURA , G. & V ILARDO , G., 2003. Geodymanic significance of the Aeolian volcanism (Southern Tyrrhenian Sea, Italy) in light of structural, seismological and geochemical data. Tectonics, 22, DOI: 10.1029/2003TC001506,2003.

68

C. CIGOLINI ET AL.

D E F INO , M., L A V OLPE , L. ET AL . 1988. The Stromboli eruption of December 6, 1985– April 25, 1986: volcanological, petrochemical and seismological data. Rendiconti della Societa` Italiana di Mineralogia e Petrologia, 43, 1021– 1038. D I C ARLO , I., P ICHAVANT , M., R OTOLO , S. G. & S CAILLET , B. 2006. Experimental crystallization of a High K-arc basalt: the golden pumice, Stromboli Volcano. Journal of Petrology, 47, 1317–1343. D IXON , J. B. 1997. Degassing of alkalic basalts. American Mineralogist, 82, 368– 378. D IXON , J. B. & S TOLPER , E. M. 1995. An experimental study of water and carbon dioxide solubilities in Mid-Ocean Ridge basaltic liquids, Part II. Applications to degassing. Journal of Petrology, 36, 1633–1646. D RAGONI , M. & M AGNANESI , C. 1989. Displacement and stress produced by a pressurized, spherical magma chamber surrounded by a viscoelastic shell. Physics of the Earth and Planetary Interior, 56, 316–328. E INSTEIN , A. 1911. Berichtgung zu meiner Arbeit: eine neue Bestimmung der Moleku¨l-dimensionen, Annalen der Physik, 34, 591– 592. F INETTI , I. & D EL B EN , A. 1986. Geophysical study of the Tyrrhenian opening. Bollettino Geofisico Teorico Applicato, 28, 75– 156. F INIZOLA , A., S ORTINO , F., L ENAT , J. F. & V ALENZA , M. 2002. Fluid circulation at Stromboli volcano (Aeolian Islands, Italy) from self-potential and CO2 surveys. Journal of Volcanology and Geothermal Research, 116, 1– 18. F RANCALANCI , L., M ANETTI , P. & P ECCERILLO , A. 1989. Volcanological and magmatological evolution of Stromboli volcano (Aeolian Islands): the roles of fractional crystallisation, magma mixing, crustal contamination and source heterogeneity. Bulletin of Volcanology, 51, 355– 378. F RANCALANCI , L., M ANETTI , P., P ECCERILLO , A. & K ELLER , J. 1993. Magmatological evolution of the Stromboli volcano (Aeolian Arc, Italy): inferences from major and trace element and Sr isotopic composition of lavas and pyroclastic rocks. Acta Vulcanologica, 3, 127–151. F RANCALANCI , L., T OMMASINI , S., C ONTICELLI , S. & D AVIES , G. R. 1999. Sr isotope evidence for short magma residence time for the 20th century activity at Stromboli volcano, Italy. Earth and Planetary Science Letters, 167, 61–69. F ULCHER , G. S. 1925. Analysis of recent measurements of the viscosity of glasses, Journal of American Ceramic Society, 8, 339 –355. G ARDNER , P. M. & R OBINS , B. 1974. The olivine– plagioclase reaction: geological evidence from the Seiland petrographic province, northern Norway. Contribution to Mineralogy and Petrology, 44, 149–156. G ASPARIK , T. 1984. Two pyroxene thermobarometry with new experimental data in the CaO– MgO– Al2O3 –SiO2. Contribution to Mineralogy and Petrology, 87, 87– 97. G ASPARIK , T. 1990. A thermodynamic model for the enstatite-diopside join. American Mineralogist, 75, 1080–1091. G AUTHIER , P. J. & C ONDOMINES , C. 1999. 210Pb– 226Ra radioactive disequilibria in recent lavas and radon

deggassing: inferences on the magma chamber dynamics at Stromboli and Merapi volcanoes. Earth and Planetary Science Letters, 172, 111– 126. G AUTHIER , P. J., L E C LOAREC , M. F. & C ONDOMINES , C. 2000. Degassing processes at Stromboli volcano inferred from the short-lived disequilibria (210Pb– 210Bi – 210Po) in volcanic gases. Journal of Volcanology and Geothermal Research, 102, 1 –19. G HIORSO , M. S. & S ACK , R. O. 1995. Chemical mass transfer in magmatic processes: IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperature and pressures. Contribution to Mineralogy and Petrology, 119, 197– 212. G ILLOT , P. Y. & K ELLER , J. 1993. Radiochronological dating of Stromboli. Acta Vulcanologica, 3, 69–77. G IORDANO , D. & D INGWELL , D. B. 2003. NonArrhenian multicomponent melt viscosity: a model. Earth and Planetary Science Letters, 208, 337–349. G UDMUNDSSON , A. 2006. How local stresses control magma-chamber ruptures, dyke injections, and eruptions in composite volcanoes. Earth Science Reviews, 79, 1– 31. G VIRTZMAN , Z. & N UR , A. 1999. Formation of Mount Etna as a consequence of slab rollback. Nature, 401, 782–785. G VIRTZMAN , Z. & N UR , A. 2001. Residual topography, lithospheric thickness, and sunken slabs in the central Mediterranean. Earth and Planetary Science Letters, 187, 117– 130. H ARRIS , A. J. L. & S TEVENSON , D. S. 1997. Magma budgets and steadystate activity of Vulcano and Stromboli. Geophysical Research Letters, 24, 1043– 1046. H AWKESWORTH , C. J., G EORGE , R., T URNER , S. & Z ELLMER , G. 2004. Time scales of magmatic processes. Earth and Planetary Science Letters, 218, 1– 16. H OLOHAN , E. P., T ROLL , V. R., W ALTER , T. R., M UNN , S., M C D ONNELL , S. & S HIPTON , Z. K. 2005. Elliptical calderas in active tectonic settings: an experimental approach. Journal of Volcanology and Geothermal Research, 144, 119– 136. H ORNIG -K JARSGAARD , I., K ELLER , J., K OBERSKI , U., S TADBAUER , E., F RANCALANCI , L. & L ENHART , R. 1993. Geology, stratigraphy and volcanological evolution of the island of Stromboli, Aeolian arc, Italy. Acta Vulcanologica, 3, 21–68. H U¨ BNER , J. S. & S ATO , M. 1970. The oxygen fugacity– temperature relationships of manganese oxide and nickel oxide buffers. American Mineralogist, 55, 934–952. H UI , H. & Z HANG , Y. 2007. Toward a general viscosity equation for anhydrous and hydrous silicate melts. Geochimica et Cosmochimica Acta, 71, 403– 416. H UPPERT , H. E. & S PARKS , R. S. J. 1988. The generation of granitic magmas by intrusion of basalt into continental crust. Journal of Petrology, 29, 599–624. H YPPOLITE , J. C., A NGELIER , J. & R OURE , F. 1993. A major geodynamic change revealed by Quaternary stress patterns in the Souther Apennines (Italy). Tectonophysics, 230, 199–210.

STROMBOLI: FROM MANTLE TO ERUPTION I SHIBASHI , H. & S ATO , H. 2007. Viscosity measurements of subliquidus magmas: Alkali olivine basalt from the Higashi–Matsuura district, Southwest Japan. Journal of Volcanology and Geothermal Research, 160, 223–238. J ELLINEK , A. M. & D E P AOLO , D. J. 2003. A model for the origin of large silicic magma chambers: precursors of caldera-forming eruptions. Bulletin of Volcanology, 65, 363–381. J ELLINEK , A. M. & K ERR , R. C. 1999. Mixing and compositional stratification produce by natural convection. 2 Applications to the differentiation of basaltic and silicic magma chambers and komatiite lava flows. Journal of Geophysical Research, 104, 7203–7218. K ELLER , J., H ORNIG -K JARSGAARD , I., K OBERSKI , U., S TADLBAUER , E. & L ENHARD , R. 1993. Geological Map of Stromboli 1:10000. Acta Vulcanologica, 3. K IRKPATRIK , R. J. 1977. Nucleation and growth of plagioclase, Makaopuhi and Alae lava lakes, Kilauea Volcano, Hawaii. Geological Society of America Bulletin, 88, 78– 84. K RESS , V. C. & C ARMICHAEL , I. S. E. 1991. The compressibility of silicate liquids containing Fe2O3 and the effect of composition, temperature, oxygen fugacity and pressure on their redox states. Contribution to Mineralogy and Petrology, 108, 82– 92. K UDO , A. M. 1983. Origin of calc.alkaline andesites, Nasu zone, northeastern Japan: Kuno revisited. Geochemical Journal, 17, 51–62. L AIOLO , M. & C IGOLINI , C. 2006. Mafic and Ultramafic xenoliths of San Bartolo lava field: new insight on the ascent and storage of Strombolian magmas. Bulletin of Volcanology, 68, 653–670. L ANDI , P., M E` TRICH , N., B ERTAGNINI , A. & R OSI , M. 2004. Dynamics of magma mixing and degassing recorded in plagioclase at Stromboli (Aeolian Archipelago, Italy). Bulletin of Volcanology, 147, 213–227. L ANDI , P., F RANCALANCI , L. ET AL . 2006. The December 2002– July 2003 effusive event at Stromboli volcano, Italy: insight into the shallow plumbing system by petrochemical studies. Journal of Volcanology and Geothermal Research, 155, 263–284. L ANGE , R. A. & C ARMICHAEL , I. S. E. 1990. Thermodynamic properties of silicate liquids with emphasis on density, thermal expansion and compressibility. In: N ICHOLLS , J. & R USSELL , J. K. (eds) Modern Methods of Igneous Petrology: Understanding Magmatic Processes. Reviews in Mineralogy and Geochemistry, 24, 25– 59. M ANGA , M. & B RODSKY , E. 2006. Seismic triggering of eruptions in the far field: Volcanoes and Geysers. Annual Review of Earth and Planetary Sciences, 34, 263–291. M ARSH , B. D. 1981. On the crystallinity, probability of occurrence and rheology of lava and magma. Contribution to Mineralogy and Petrology, 78, 85– 98. M ARSH , B. D. 1988. Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization I: theory. Contribution to Mineralogy and Petrology, 99, 277–291.

69

M ARSH , B. D. 1989. Magma chambers. Annual Review of Earth and Planetary Sciences, 17, 439–472, DOI: 10.1146/annurev.ea.17.050189.002255. M ARTINI , M., B UCCIANTI , A., C APACCIONI , B., G IANNINI , L. & P RATI , F. 1996. Fumarole gas analysis (Stromboli). Acta Vulcanologica, 6, 53–54. M C B IRNEY , A. R. 1979. Effects of assimilation. In: Y ODER , H. S. (ed.) The Evolution of the Igneous Rocks. Princeton University Press, Princeton, NJ, 307– 329. M E` TRICH , N., B ERTAGNINI , A., L ANDI , P. & R OSI , M. 2001. Crystallization driven by decompression and water loss at Stromboli volcano (Aeolian Islands, Italy). Journal of Petrology, 42, 1471– 1490. M E` TRICH , N., B ERTAGNINI , A., L ANDI , P. & R OSI , M. 2005. Triggering mechanism at the origin of paroxysm at Stromboli (Aeolian Archipelago, Italy): the 5 April 2003 eruption. Geophysical Research Letters, 32, L10305, DOI: 10.10129/2004GL022257. M ORELLI , C., G IESE , P. ET AL . 1975. Crustal structure of Southern Italy. A seismic refraction profile between Puglia-Calabria-Sicily. Bollettino Geofisico Teorico Applicato, 18, 183– 210. M URASE , T. 1962. Viscosity and related properties of volcanic rocks at 800 to 1400. Journal Faculty of Science, Hokkaido University, VII-I, 487– 584. M URRU , M., M ONTUORI , C., W YSS , M. & P RIVITERA , E. (1999). The locations of magma chambers at Mt. Etna, Italy, mapped by b-values, Geophysical Research Letters, 26, 2553–2556, DOI: 10.10129/ 1999GL900568. N APPI , G. 1976. Recent activity of Stromboli (November 5– 24 1975). Nature, 261, 119–120. N AVON , O. & L YAKHOVSKY , V. 1998. Vesiculation processes in silicic magmas. In: G ILBERT , J. S. & S PARKS , R. S. J. (eds) The Physics of Explosive Volcanic Eruptions. Geological Society, London, Special Publications, 145, 27– 50. N EWHALL , C. & S ELF , S. 1982. The volcanic explosivity index (VEI) – an estimate of explosivemagnitude for hystorical volcanism. Journal of Geophysical Research, 87, 1231–1238. N EWTON , R. C., C HARLO , T. V. & K LEPPA , O. J. 1980. Thermochemistry of the high structural state of plagioclase. Geochimica et Cosmochimica Acta, 75, 369– 376. P ASQUARE , G., F RANCALANCI , L., G ARDUNO , V. H. & T IBALDI , A. 1993. Structure and geologic evolution of the Stromboli volcano, Aeolian Islands, Italy. Acta Vulcanologica, 3, 79–89. P ANZA , G. F. & P ONTEVIVO , A. 2002. The Lithosphere – Astenosphere System in the Calabrian Arc and Surrounding Seas. The Abdus Salam International Centre for Theoretical Physics (UNESCO- IAEA), 1– 38. P ECCERILLO , A. 2001. Geochemical similarities between the Vesuvius, Phlegraean Fields and Stromboli Volcanoes: petrogenetic, geodynamic and volcanological implications. Mineralogy and Petrology, 73, 93–105. P ECCERILLO , A. & T AYLOR , S. R. 1976. Geochemistry of Eocene Calc– Alkaline volcanic rocks from Kastamorun area, Northern Turkey. Contribution to Mineralogy and Petrology, 58, 63–81.

70

C. CIGOLINI ET AL.

P INKERTON , H. & S TEVENSON , R. J. 1992. Methods of determining the rheological properties of magmas at sub-liquidus temperatures. Journal of Volcanology and Geothermal Research, 53, 47–76. P ONTEVIVO , A. & P ANZA , G. F. 2006. The lithosphere– asthenosphere system in the Calabrian Arc and surrounding seas – Southern Italy. Pure and Applied Geophysics, 163, 1617–1659. R ANDOLPH , A. D. & L ARSON , M. A. 1971. Theory of Particule Processes. Academic Press, New York, 251. R ENZULLI , A., S ERRI , G., S ANTI , P., M ATTIOLI , M. & H OLM , P. M. 2001. Origin of high silica liquids at Stromboli volcano (Aeolian Islands, Italy) inferred from crustal xenoliths. Bulletin of Volcanology, 62, 400– 419. R ICHET , P., B OTTINGA , Y., D ENIELOU , L., P ETITET , J. P. & T EQUI , C. 1982. Thermodynamic properties of quartz, crystobalite and amorphous SiO2: drop calorimetry measurements between 1000 and 18008K and a review from 0 to 20008K. Geochimica et Cosmochimica Acta, 46, 2639–2658. R IPEPE , M. & G ORDEEV , E. 1999. Gas bubble dynamics model for Shallow volcanic tremor at Stromboli. Journal of Geophysical Research, 104, 10 635– 10 654. R IPEPE , M., M ARCHETTI , E. ET AL . 2005. Effusive to explosive transition during the 2003 eruption of Stromboli volcano. Geology, 33, 341 –344. R OBIE , R. A., H EMINGWAY , B. S. & F ISHER , J. R. 1979. Thermodynamic Properties of Minerals and Related Substances at 218.15 K and 1 Bar (105 Pascals) Pressure and Higher Temperatures. United States Geological Survey Bulletins, 1452, 456. R OSI , M., B ERTAGNINI , A. & L ANDI , P. 2000. Onset of the presistent activity at Stromboli Volcano (Italy). Bulletin of Volcanology, 62, 294 –300. R YAN , M. P. 1993. Neutral buoyancy and the structure of mid-ocean ridge magma reservoirs. Journal of Geophysical Research, 98, 22 321–22 338. R YAN , M. P. 1994. Neutral buoyancy and the structure of mid-ocean ridge magma reservoirs and their sheeteddike complex; a summary of basic relationships. In: R YAN , M. P. (ed.) Magmatic Systems. International Geophysics Series, 57, 97–138. S ACK , R. O. & G HIORSO , M. S. 1989. Importance of considerations of mixing properties in establishing an internally consistent thermodynamic database: thermochemistry of minerals in the system Mg2SiO4 – Fe2SiO4 – SiO2. Contributions to Mineralogy and Petrology, 102, 41–68. S HAW , H. R. 1969. Rheology of basalt in the melting range. Journal of Petrology, 10, 510 –535. S HAW , H. R. 1972. Viscosity of magmatic liquids: an empirical method of prediction. American Journal of Science, 272, 870–893. S HAW , H. R. 1980. The fracture mechanism of magma transport from the mantle to the surface. In: H ARGRAVES , R. B. (ed.) Physics of Magmatic Processes. Princeton University Press, Princeton, NJ, 201– 264.

S IGMARSSON , O. 1996. Short magma residence time beneath an Icelandic volcano inferred from U-series disequilibria. Nature, 382, 440–442. S ILVER , L. A. & S TOLPER , E. M. 1985. A thermodynamic model for hydrous silicate melts. Journal of Geology, 93, 161–178. S TEVENSON , D. S. & B LAKE , S. 1998. Modelling the dynamics and thermodynamics of volcanic degassing. Bulletin of Volcanology, 60, 307– 317. T AMMANN , G. & H ESSE , W. 1926. Die Abha¨ngigkeit der Viskosita¨t von der Temperatur bei unterkhu¨lten Flu¨ssigkeiten. Zeitschrift fu¨r Anorganischen und Allgemeine Chemie (Journal of Inorganic and General Chemistry), 156, 245 –257. T IBALDI , A. 1996. Mutual influence of dyking and collapses at Stromboli volcano, Aelioan Arc, Italy. In: MC QURRE , W. J., JONES , A. P. & NEUBERG , J. (eds) Volcano Instability on the Earth and Other Planets. Geological Society of London, Special Publications, 110, 55–63. T IBALDI , A. 2001. Multiple sector collapses at Stromboli volcano, Italy: how they work. Bulletin of Volcanology, 63, 112– 125. T IBALDI , A., C ORAZZATO , C., A PUANI , T. & C ANCELLI , A. 2003. Deformation at Stromboli volcano (Italy) revealed by rock mechanics and structural geology. Tectonophysics, 361, 187–204. U NDERWOOD , E. E. 1970. Quantitative Stereology. Addison-Wesley, Reading, MA, 274. V AGGELLI , G., F RANCALANCI , L., R UGGIERI , G. & T ESTI , S. 2003. Persistent polybaric rests of calcalkaline magma at Stromboli volcano, Italy: pressure data from fluid inclusions in restitic quartzite nodules. Bulletin of Volcanology, 65, 385–404. V ENTURA , G., V ILARDO , G., M ILANO , G. & P INO , N. A. 1999. Relationships among crustal structure, volcanism and strike–slip tectonics in the Lipari-Vulcano volcanic complex. Physics of the Earth and Planetary Interiors, 116, 31–52. V OGEL , D. H. 1921. Temperaturabha¨ngigkeitsgesetz der Viskosita¨t von Flv¨ssigkeiten, Zeitschrift fu¨r Physikalische Chemie (International Journal of Research in Physical Chemistry and Chemical Physics), 22, 645–646. W ANAMAKER , B. J. & E VANS , B. 1985. Experimental Diffusional Crack Healing in Olivine. In: S CHOCK , R. N. (ed.) American Geophysical Union Monographs, 21, 194–210. W ESTAWAY , R. 1993. Quaternary uplift of Southern Italy. Journal of Geophysical Research, 98, 21 741–21 772. Z HANG , Y. 1999. H2O in rhyolitic glasses and melts: measurements, speciation, solubility and diffusion. Reviews of Geophysics, 37, 493– 516. Z HANG , Y. & S TOLPER , E. M. 1991. Water diffusion in basaltic melts. Nature, 351, 306– 309. Z OLLO , A., G ASPARINI , P. ET AL . 1998. An image of the Mt. Vesuvius obtained by 2D seismic tomography. Journal of Volcanology and Geothermal Research, 82, 161–163.

A rigorous tool for evaluating the importance of viscous dissipation in sill formation: it’s in the tip ANDREW P. BUNGER CSIRO Petroleum Resources, Private Bag 10, Clayton South 3169, Australia (e-mail: [email protected]) Abstract: Crustal magma transport is typically described using a complex, non-linear model associated with fluid-driven fracturing, and therefore fundamentally sound modelling forms the basis for interpretation of magmatic intrusions. One of the most basic considerations is that magma-driven sills can be broadly categorized based on the energy dissipation mechanism that is predominant during intrusion growth. In cases where either viscous flow or overcoming fracture toughness strongly dominates fracture behaviour, it is typical to speak of viscosity-dominated or toughness-dominated regimes, each of which defines a class of fracture propagation with significant implications for modelling. This paper presents a straightforward and geometry-independent means for local determination of the expected propagation regime based on an experimentally verified mathematical analysis of the multi-scale, coupled mechanics that govern the near-tip region. The propagation regime is then related directly to the ratio between a characteristic length associated with the near-tip physics compared with the size of the fracture/sill. Sill growth is shown to be expected in or near the viscosity-dominated regime and hence modelling generally must take into account the complexity of the near-tip region rather than relying solely on the tip behaviour implied by linear elastic fracture, although toughness-dominated mafic intrusions can also be anticipated if fracture toughness increases sufficiently rapidly with the intrusion size.

Crustal magma transport is typically modelled as a fluid-driven fracturing process (e.g. Spence et al. 1987; Lister 1990; Lister & Kerr 1991). These systems are complex owing to the presence of multiple, coupled physical processes such as rock fracturing and viscous fluid flow in the growing fracture. They are also non-linear owing both to a nonlinear relationship between the fluid flow and the fracture opening and to the presence of moving boundaries associated with the fluid front and propagating fracture tip. Because of the complexity and non-linearity of these systems, the behaviour predicted by a given mechanical model can vary widely. One must therefore take great care in developing solution methods or approximations because, for example, a solution derived under the assumption that the viscous dissipation is negligible can be shown to give vastly different predictions of intrusion geometry or growth rate than would be derived under the assumption that the energy associated with fracture growth is negligible (e.g. Lister & Kerr 1991; Savitski & Detournay 2002). In devising a modelling approach upon which geological interpretation can be based, one of the basic questions is whether energy is primarily dissipated due to viscous fluid flow in the fracture or due to the process of breaking material bonds in the crack tip region, a process that is often expressed via a critical energy release rate or fracture toughness. Hence it is useful to speak of viscosity- and toughness-dominated regimes of

fracture/sill growth, where viscosity-dominated refers to the regime in which most of the energy dissipation is due to viscous fluid flow, and toughness-dominated refers to the regime in which most of the energy is dissipated due to the process of rock fracturing. These regimes are useful, because they define classes of fracture behaviour where the contribution of phenomena related to fracture toughness or fluid viscosity, respectively, can be neglected, thereby providing significant simplification for fracture modelling and analysis. In many cases it is these simplifications that make the underlying equations tractable (e.g. Spence & Sharp 1985; Detournay 2004), but it then remains vital that the propagation regime associated with a given approximate solution is well-understood, if it is to be applied in the interpretation of geological features. A number of past contributions present analyses of propagation regime based on relative values of quantities derived from a coupled elastic fracture/ lubrication theory mathematical model (after Khristianovic & Zheltov 1955) for a variety of idealized geometries (e.g. Spence & Sharp 1985; Lister & Kerr 1991; Adachi 2001; Savitski & Detournay 2002; Detournay 2004; Garagash & Detournay 2005). Yet the regime of propagation is still generally determined only from analysis of groups of parameters that are specific to particular idealized geometries.

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 71– 81. DOI: 10.1144/SP304.4 0305-8719/08/$15.00 # The Geological Society of London 2008.

72

A. P. BUNGER

A peculiar aspect of fluid-driven fractures is the multi-scale nature of the tip region. For example, the aperture w in the tip region has been shown to be given by linear elastic fracture mechanics (LEFM) in the toughness-dominated regime, so that w / s 1/2, where s is the distance from the tip. Conversely, for the viscosity-dominated regime, w / s 2/3 when viewed at the scale of the fracture (e.g. Detournay 2004). Each of the aforementioned contributions, at varying levels, discuss the dependence of the behaviour of the near-tip region on the regime of propagation. However, the details of the complex multi-scaled near-tip region require careful consideration, and it is the solution resulting from this tip analysis that leads to the theoretical foundation for the present work. The purpose, then, of the present work is to utilize the previously derived (Garagash & Detournay 2000, 2005) and recently experimentally verified (Bunger & Detournay 2008) near-tip solution in order to present a rigorous and straightforward method for determining fracture propagation regime in a manner that can be applied locally to a growing fluid-driven fracture, provided that the fracture’s leading edge is sufficiently smooth (i.e. there are no sharp corners in the overall fracture shape). After reviewing the relevant theoretical and experimental contributions and describing the new method for determining propagation regime, the paper concludes with a discussion of the implications of this analysis for numerical modelling of fluid-driven fractures, such as magma-driven sills, with the elastic fracture/lubrication theory model.

The sill growth model One of the most common approaches to modelling magma transport in Earth’s crust is to use a model that considers elastic fracture mechanics coupled with laminar fluid flow within the fracture as described by lubrication theory. This model can be traced back to the seminal work of Khristianovic & Zheltov (1955), and has been proposed for modelling magma-driven fractures in a number of contributions (e.g. Spence et al. 1987; Lister & Kerr 1991). A form of this model will be presented below. The viability of this model, in light of its simplifying assumptions, will also be discussed. At the outset the model will neglect thermal effects on the host rock, solidification processes in the magma, and the permeability of the host rock. This is akin to assuming sufficiently rapid propagation so that: (1) the fracture tip remains ahead of the thermal and fluid diffusion fronts; (2) nearly all of the fluid (i.e. magma) supplied, for example from a feeder dyke, is stored in the

growing sill rather than diffusing into the surrounding rock; and (3) insufficient time is allowed for the heat transfer required to bring about a phase transition in the magma or host rock during the time of sill extension. Of course these assumptions must be critically evaluated on a case-by-case basis. Further, it is assumed that the effects of gravity are negligible and therefore consideration does not cover the growth of features such as buoyancy-driven dykes. The mathematical model then begins with Reynold’s lubrication equation, which relates the fracture opening w to the fluid pressure pf by combining the mass balance equation for an incompressible fluid with the fluid flux expression given by the Poiseuille equation. Hence, @w 1 ¼ 0 r  (w3 rpf ) þ Qo c (t)d(x,y) m @t

(1)

where m0 ¼ 12 m with dynamic fluid viscosity m, d(x,y) denotes the Dirac delta function with the origin of the system of coordinates (x,y) taken to coincide with the injection point, Qo is the mean volumetric injection rate and c (t) expresses the variation of the injection rate with time (see Table 1). Although the Newtonian lubrication Table 1. Notation E E0 H KIc (K0 ) L N Qo V ‘ pf s st t w wo x,y ˆ V ˆ P

a 1 f (.,.) m (m 0 ) n c (. ) so jˆ

Young’s modulus Plane strain modulus Emplacement depth Fracture toughness (alternative form) Fracture extent Number of elements in a numerical simulation Mean/constant volumetric injection rate Fracture tip velocity Tip characteristic length Fluid pressure Distance from the fracture tip A fixed distance from the fracture tip Time Fracture opening An average fracture opening Spatial coordinates Dimensionless fracture opening in tip scaling Dimensionless fluid net pressure in tip scaling A prescribed accuracy Small parameter in tip scaling Variation function for the far-field stress Dynamic fluid viscosity (alternative form) Poisson’s ratio Variation function for the injection rate Mean/constant far-field stress Distance from fracture tip in tip scaling

VISCOUS DISSIPATION IN SILL GROWTH

model is an accepted approach for a wide range of applications, it should be noted that magmas containing a significant crystal fraction could be expected to deviate significantly from Newtonian behaviour. However, non-Newtonian versions of this model can be developed and the basic elements of the approach remain unchanged (e.g. Desroches et al. 1994). Additionally, compressibility is sometimes considered to be important to certain occurrences of magma-driven fracture behaviour (e.g. Woods et al. 2006); however, this is expected to be mainly relevant to the problem of magma ascent in dykes, which has already been omitted from consideration under the present assumptions. The lubrication equation is coupled to a description of the fracture opening based on linear elastic fracture mechanics (LEFM), which includes both a propagation condition (to follow) and a relation coupling the fracture opening with the internal pressure and far-field stress. Taking a mean far-field stress so acting perpendicular to a fracture with the a shape in plan view (i.e. the fracture footprint) denoted by S(t), the elasticity equation can be expressed as E0 pf  so f (x,y) ¼  8p ð w(x0 ,y0 ,t) dS(x0 ,y0 )  2 2 3=2 0 0 S(t) [(x  x) þ (y  y) ]

(2)

where E0 is the so-called plane strain elastic modulus of the solid, x0 and y0 are variables of integration, and f(x,y) embodies spatial variation in the far-field stress. Closed-form solutions have been derived for some very simple geometries and pressure/stress distributions; however, this relation usually requires numerical solution. For opening mode fractures, well-known asymptotic analysis of the elasticity equations imply that the near-tip opening is given by  1=2 32 KI 1=2 s , w p E0

s 1 L

73

an intermediate scale, outside of the damaged zone, but small relative to the fracture size at which the energy release rate associated with the fracture can be computed with a desired accuracy using only the leading order term of the asymptotic expansion of the elasticity equation, i.e. Equation 3 and its corresponding s 21/2 singular term for the near-tip stresses. A couple of additional points should be made regarding the application of LEFM in this context. Firstly, laboratory experiments and field observations give strong evidence that KIc is not a constant rock parameter, but instead it can increase significantly with increasing fracture size and in situ stress (e.g. Rubin 1993; Carpinteri 1994; Fialko & Rubin 1997). In spite of a number of important contributions, it remains difficult to confidently select, based on the available data, an appropriate value for KIc that would govern a large scale/large in situ stress process. One may hope that careful ongoing experimentation and modelling will eventually provide some insight; however until then, we move ahead using Equation 3 with KI ¼ KIc. Future work may lead to important refinements to this approach. While this first issue is primarily concerned with the physics of rock fracturing, a second issue is related to the mathematical properties of the elasticity equations leading to Equation 3. For illustration, consider a two-dimensional fracture of length 2L under plane strain conditions, which is subjected to an internal uniform fluid pressure pf and an isotropic in situ compressive stress so (Fig. 1). The Williams (1957) type expansion of the near tip stresses is given by ( pf  so )(pL)1=2 sij (s; u) ¼ fij (u) s1=2   pf 0 þ ( pf  so )O[(s=L)1=2 ] þ 0 pf

(4)

where fij(u) is a known function of the angular coordinate u and O is used to indicate terms which are of (3)

where KI is the mode I stress intensity factor, s the distance from the fracture tip (with the s-axis directed inwards), and L is the fracture length. Fracture propagation is then taken to require the availability of sufficient energy to break the rock just a head of the fracture tip. Within LEFM this is conveniently expressed by KI ¼ KIc, where KIc is the fracture toughness of the material. The basic assumption for this approach to fracture propagation is that the inelastic damaged zone near the fracture tip is sufficiently small so that there exists

Fig. 1. Sketch of plane strain pressurized fracture.

74

A. P. BUNGER

this order and higher in s/L. This representation of the near-tip stresses has been used to argue that the region that is dominated by leading term, which is the one retained in LEFM analysis, requires that s/L  (pf 2 so)2/p2f , which, for typical values of the in situ stress and fluid (magma) pressure, would render the zone governed by the LEFM term so small that it would typically be irrelevant (Rubin 1993, 1995). However, it is easily shown using the well-known J-integral method Rice (1968, 1974) that the uniform stress term (second term on the right-hand side of Equation 4) does not contribute to the computation of the energy release rate. Therefore, provided that one is considering a problem for which LEFM considerations are used solely as a tool for determining whether a fracture has sufficient energy for propagation (i.e. to enforce the condition KI ¼ KIc), the applicability of this theory is not undermined by the presence of large in situ stresses, provided that the inelastic zone near the tip is sufficiently small and that it is possible to appropriately account for (e.g. Khazan & Fialko 1995) the expected dependence of the fracture toughness on the in situ stress.” Returning to the mathematical description, it can be seen immediately that this problem involves at least one moving boundary. In fact, generally the fluid and fracture fronts do not coincide with one another, thus requiring the tracking of two moving boundaries (e.g. Bunger 2005; Lecampion & Detournay 2007). However, it can be demonstrated (Garagash & Detournay 2000) that satisfying the condition  1=2 so K 02 1 (5) m0 VE02 with V the fracture tip velocity, implies that the lag between the fluid and fracture fronts can be neglected. In this case the mathematical model is completed by the addition of the boundary conditions at the moving front K0 lim 1=2 ¼ 0 s!0 s E @p f ¼0 lim w3 s!0 @s w

(6)

Fig. 2. Sketch of the fracture tip region.

can be demonstrated that the region close to the fracture tip (Fig. 2) has a universal form (Garagash & Detournay 2000, 2005). This universal form will eventually be used to obtain a geometryindependent method for determining the propagation regime, but its basic results must first be examined. Non-dimensional forms of the distance from the ˆ , and the internal net tip jˆ, the fracture opening V ˆ may be introduced according to pressure P ^ pf  so ¼ 1E0 P ^ s ¼ ‘^j, w ¼ 1‘V, with ‘¼

K 06 E04 m02 V 2

, 1¼

E0 m0 V : K 02

where an alternative form of the fracture toughness K0 ¼ Klc(32/p )1/2 has been introduced to keep the equations uncluttered. Here the first equation expresses the LEFM propagation condition discussed above, while the second simply expresses a no-flux boundary condition at the fracture tip.

The near-tip region In spite of the geometric specificity required in order to solve the above system of equations, it

(9)

Noting that variation in the far-field stress can typically be neglected when examining only the near-tip region, it can be shown rigorously (Bunger & Detournay 2008) that the above governing equations for fluid-driven fracture growth degenerate near the fracture tip to ð1 ^ ^ 0 dV(j ) d^j 0 ^ ^j ) ¼ 1 (10) P( 4p 0 d^j 0 ^j  ^j 0 ^ ^ 2 dP ¼ 1 V ^ dj

(11)

^ V ¼ 1: 1=2 ^j!0 ^ j

(12)

lim

(7)

(8)

Obviously, the fracture opening nearest the tip is described by the limit in Equation 12. However, simultaneous solution of Equations 10 and 11 implies another asymptotic form given by ^ ¼ 21=3  35=6 ^j2=3 V

(13)

which in dimensional form is given by w  2  37=6

 m 1=3 E0

V 1=3 s2=3 ,

s  1: L

(14)

VISCOUS DISSIPATION IN SILL GROWTH

This 2/3 asymptote of the fracture tip opening is recognized as an asymptotic form associated with fluid-driven fracture propagation when KIc is equal to zero (Spence & Sharp 1985; Lister & Kerr 1991; Desroches et al. 1994). Its reconciliation with the LEFM asymptote Equation 12 under conditions of ‘small toughness’ relies on the realization that the near-tip region has the form of a boundary layer solution when the fracture is propagating in the viscosity-dominated regime, with the LEFM asymptote providing the inner behaviour (jˆ ! 0, closest to the tip) and the 2/3 asymptote providing the outer behaviour (jˆ ! 1, but still near the tip relative to the rest of the fracture) (Garagash & Detournay 2000). Hence, Garagash & Detournay (2005) present a numerical solution connecting the inner and outer asymptotics so that the universal form of the fluid-driven fracture tip is known (Fig. 3). The validity of the autonomous generalized ˆ (jˆ ) has recently been confirmed tip asymptote V with a series of laboratory experiments that consisted of driving circular hydraulic fractures through polymethyl methacrylate (PMMA) or glass specimens using fluids which were solutions of water, blue food dye and either glycerine or glucose (see Bunger & Detournay 2008 for details of the design and results of these experiments). Experimental data pertaining to 10 different tests are presented in Figure 4 together with the tip ˆ (jˆ ). The experimental results exhibit solution V some scatter, mainly due to the fact that the fracture opening becomes very small in the tip region, which can be to the detriment of the signal to noise ratio for the measurements. Nonetheless, the close agreement between the experimental and analytical results for the fracture tip opening uphold the multiscale tip asymptote that has

75

Fig. 4. Experimental verification of the boundary layer solution showing data from 10 experiments, after Bunger & Detournay (2008).

been developed to describe the tip region of hydraulic fractures. Physical understanding of the near-tip behaviour is aided by recognition of the fact that the fracture tip region is governed by two physical processes, which in this case happen to be energy dissipation mechanisms. Here energy is dissipated both in viscous flow of the fluid and in breaking of bonds just ahead of the fracture tip. As is typically the case when multiple physical processes are coupled, each process carries with it its own characteristic length scale (or scales), so that a multi-scale problem naturally arises. For the present problem, LEFM processes occupy the region closest to the tip, with coupling between the fluid and solid governing the outer tip behaviour. Furthermore, the tip length scale ‘ gives a characteristic length associated with the transition between the region that is LEFM dominated and the one that is dominated by fluid– solid coupling (and hence the 2/3 asymptote).

Determining propagation regime

Fig. 3. Numerical solution (bold line) shown with ˆ ¼ jˆ1/2) and outer fluid–solid coupling inner LEFM (V 1/3 5/6 ˆ 2/3 ˆ (V –2 3 j ) asymptotic solutions.

As will be demonstrated in the sections to follow, the regime of fluid-driven fracture propagation has a profound impact on approaches to modelling fracture growth numerically. Furthermore, for simplified geometries it may be desirable to use an analytical solution based on some simplified form of the governing equations; however, the development and application of these analytical solutions depends crucially on the propagation regime. Hence the motivation is to devise a method for estimating the propagation regime that is both straightforward and that does not rely on quantities that are associated with a particular idealized geometry. A few details of the criterion proposed here depend on how one defines the tip region. Here a

76

A. P. BUNGER

definition of the fracture tip is considered that is relevant to numerical modelling. In particular, the tip region is taken as a single point located at st ¼ 0.01 L. For example, from a numerical modelling perspective this could correspond to the centre of the discretization element nearest the fracture tip, if the fracture is meshed using 50 equal-sized elements. The question in determining propagation regime then becomes which, if either, of the above tip asymptote Equations 6 or 14 give an accurate estimate of the fracture opening at st. The propagation regimes can then be defined in terms of the thresholds    1=2  32 KIc 1=2   s  w   p E0 t  ,a (15) w for the toughness-dominated regime and    1=3   1=3 2=3  w  2 37=6 m V s t   0 E ,a w

(16)

for the viscosity-dominated regime. Choosing the desired accuracy a ¼ 0.02, making use of the numerical tip solution (Fig. 3) and the tip scaling Equation 9, one arrives directly at conditions in terms of the tip length scale ‘ and an estimate of the overall fracture extent L, namely † toughness-dominated, ‘ . 3800L; † viscosity-dominated, ‘ , 0.03L. A graphical representation of these criteria is portrayed in Figure 5. Determining the propagation

Table 2. Coefficient ck

a

st

0.01 0.02 0.05

0.02

0.05

0.1

3800 7700 19,000

480 960 2400

100 210 520

regime then requires estimating ‘ and L, either a priori from physical reasoning and parameter value estimates, or a posteriori from the solution to a numerical model or analysis of data from analogue experiments. It is important to note that this method relies on a local estimate of the propagation regime based on a local value of the fracture tip velocity V, which may be different at different locations around the fracture front. However, for reasonably simple geometries it may be valid to estimate the overall propagation regime based on an average expected tip velocity. Alternatively, bounds could be placed on the regime of propagation based on upper and lower bound estimates on the local tip velocity. It is apparent that the definitions of the regimes of propagation are specific to the definition of the tip region st and the desired accuracy a. Hence these regimes are not defined in absolute terms, but instead require one to decide what is meant by ‘/L being either large enough or small enough to warrant a given classification. This issue is not unique to the method of determining regime presented here. It also arises in methods that are based on parameters derived from the global geometry (e.g. Lister & Kerr 1991; Detournay 2004), in that one must always decide how small or large the value a group of parameters must be in order for the fracture to be in the viscosity or toughness dominated regime. In some situations a different choice of st or a may be appropriate. Tables 2 and 3 gives the coefficients ck and cm for different combinations of st and a, where ‘ . ckL and ‘ , cmL define the toughness- and viscosity-dominated regimes, respectively.

Table 3. Coefficient cm

a

st

Fig. 5. Diagram for estimating propagation regime based on overall extent L and the tip length scale ‘ from Equation 9.

0.01 0.02 0.05

0.02

0.05

0.1

0.03 0.06 0.2

0.2 0.5 1.1

1.1 2.2 5.6

VISCOUS DISSIPATION IN SILL GROWTH

Expected regime for sill growth The challenge remains to acquire appropriate estimates of the necessary parameters. At this point such choices, and even the degree to which the feature of interest may satisfy the necessary assumptions for application of the theory, are specific to each field case. However, one can attempt to make a few general, albeit somewhat cautious, statements regarding sill growth based on expected ranges of parameter values. First a range of magma viscosity values is taken as 100  m  2000 Pa s based on the discussion of Lister & Kerr (1991). Here the selection is essentially limited to relatively low viscosity basaltic magmas. In the case of granitic magmas, which are expected to have considerably higher viscosity values than the range considered here, significant questions remain regarding how to interpret the mechanism of emplacement based on field data. Whether one is considering diapirism or episodic tabular injection (e.g. Petford et al. 2000), it leads to the same conclusion that the emplacement mechanism for granitic bodies may in many cases lie outside of what can be captured by the mathematical model under consideration here. The host rock is then characterized by a plane strain modulus taken from the range 10  E0  70 GPa and a fracture toughness taken from the range 1  KIc  120 MPa m1/2. When discussing the mechanical properties of rocks it is impossible to escape the issue of size dependance. As previously discussed, there remains large uncertainty regarding how to quantify the increase of KIc with fracture size and in situ stress as one pushes the theories to geological scales. For the basic estimates desired, the considered range accounts for a possible factor of 1000 increase in the critical fracture energy Gc (noting the well known relation KIc / G1/2 c ) at the scale of sill growth relative to the laboratory-scale range, which is taken up to c. 4 MPa m1/2. Finally, the tip velocity V and fracture extent L must be considered. In field settings the fracture extent L is typically known from observation. However, estimating the tip velocity requires one to assume a model. For example one could make a rough estimate based on a known L and bounds (e.g. Petford et al. 2000) on the total emplacement time. Alternately, in this case V will be estimated based on a solution for a viscosity-dominated, penny-shaped hydraulic fracture presented by Savitski & Detournay (2002) in which the fracture extent (here the radius) is given by  0 3 4 1=9 E Qo t : L ¼ 0:696 m0

(17)

77

Here the magma supply rate Qo is considered to be constant and in the range 1  Qo  105 m3 s21 (after Lister & Kerr 1991). The tip velocity is, of course, given by simple differentiation of Equation 17 with respect to time. The time of injection t is then selected so as to give sills which fall within the range 10 m  L  10 000 m based on Equation 17. Upon subsequent computation of the tip length scale ‘ for 40 000 randomly selected combinations of parameters from within the ranges listed above, one can identify the area in the propagation regime diagram in which natural sills are expected to lie. Tabulation of these test cases is presented in Figure 6. Here we see that approximately 90% of the realizations fall in the viscositydominated regime; a fraction of 1% fall in the toughness-dominated regime, with the remainder lying in the transitional regime. Note that a solution for a viscosity-dominated fracture was chosen at the outset for estimation of the fracture size L and tip velocity V. On the other hand, one could have chosen a toughnessdominated solution presented by Savitski & Detournay (2002). In this case one would have predicted, for a given injection rate and time, a larger fracture size and tip velocity than with the viscosity-dominated model. This would drive the propagation regime estimates for the parameter ranges given here even further to the viscositydominated regime so that, on examination of these results in a diagram such as Figure 6, one would immediately see that the viscosity-dominated fracture model is more appropriate for the majority of the realizations. It is important to reiterate that the current results have been obtained based on a somewhat arbitrary assumption about how fracture energy increases with fracture size. For example, if KIc is taken in

Fig. 6. Evaluation of expected propagation regime for sills based on 40 000 realizations from estimated parameter ranges.

78

A. P. BUNGER

a range up to 400 MPa m1/2, then only 50% of the realizations lie in the viscosity dominated regime, 1% lie in the toughness-dominated regime, and the rest are in the transitional regime. It should be further noted that it is possible, then, that the lower end of the fracture toughness range may be irrelevant at the scale considered, in which case toughness-dominated behaviour could in fact be more important than is indicated by this analysis. Nonetheless, it is clear that analysis requires careful consideration of regime, and that in most cases it will be required to consider the effects of viscous dissipation. This work has also made it clear that development of models for analysis of geological data is reliant on careful experimentation and analysis in order to better understand how fracture energy/toughness scales fracture size.

Implications for numerical modelling In order to track the moving fracture front, fluiddriven fracture models almost invariably involve imposing a condition of the fracture tip opening (or near tip stress field) based on an expected asymptotic behaviour. These methods are well accepted in fracture mechanics modelling, where models based on the LEFM asymptote (Equation 6) form an important and very commonly employed class of propagation criteria. One temptation that follows from this history is to bring an approach from LEFM to the problem of fluid-driven fractures without accounting for the consequences of fluid– solid coupling in the fracture tip region. Of course, if the fluid can be assumed inviscid, this approach is valid. However, it has been demonstrated in detail above that, for many, if not the majority of the cases which are of interest to the geosciences (and indeed industry also), fluid – solid coupling is important, and as a result the region that is governed by the LEFM asymptote may be very, very small relative to the size of the fracture that is being modelled. One further temptation is to attempt to address the fact that the LEFM asymptote may only govern a very small region near the fracture tip by simply using a large number of elements. Based on the analysis presented above, one can find directly that maintaining 2% accuracy at the tip element with a uniform meshing of the fracture requires N  4:5  105

L ‘

(18)

elements. Hence it can be seen that accurate modelling of the magma-driven sills using LEFM considerations only would typically (based on 90% of

the realizations) require N , 108 elements in each modelling dimension, assuming a uniform mesh. Obviously, grading the element size from the tip towards the inlet would significantly decrease the number of elements needed, although at the expense of substantial algorithmic complexities. Also, the number of elements required would, of course, decrease upon relaxation of the accuracy requirement, but only by about an order of magnitude if one were to require only 10% accuracy. Furthermore, it is important to realize that a strong increase in the fracture toughness with fracture size could also drive the sills towards the toughness-dominated regime and thus reduce the number of elements required for accurate modelling based on LEFM considerations. Nonetheless, it is clear that in most cases modelling fluid-driven fractures using a tip condition based on LEFM is expected to be, at best, computationally impractical and at worst a brute forcing of the wrong physics at the wrong scale. In contrast, it has been demonstrated that accurate numerical solutions (compared with known analytical solutions for simple geometries) can be obtained for viscosity-dominated fractures using very coarse meshing, i.e. N  10 elements in each dimension for the two-dimensional model, provided that the 2/3 (Equation 14) asymptote is imposed at the tip elements rather than the LEFM solution (Bunger et al. 2007).

Implications for predicting the thickness of sills Although modelling the aspect ratios observed in particular natural sills is beyond the scope of this paper, it is useful to determine whether the present model is able to produce aspect ratios which are within the same orders of magnitude as are observed in the field. According to graphical catalogs (McCaffrey & Cruden 2002; Breitkreuz & Petford 2004; Cruden & McCaffrey 2006) of natural intrusions spanning the classifications of minor intrusions, sills, laccoliths, and plutons, the characteristic thickness wo to length L ratios (or just ‘thickness ratios’) are expected to lie in the range 0.0005 , wo/L , 0.5. Recall that the present analysis neglects: (1) formation of tabular bodies by periodic injection of magma; (2) the potential for complex, non-elastic deformation of the host rock; and (3) the large viscosity range which would typically be associated with the magmas that form granitic intrusions. Hence the present analysis is expected to provide a lower bound to the observed range of thickness ratios. As discussed above, provided that KIc does not become too large as the fracture size and in situ

VISCOUS DISSIPATION IN SILL GROWTH

stress increase, one would expect the majority of sills to grow in the viscosity-dominated regime. In this case, using the scaling relations derived by Savitski & Detournay (2002), it is easily shown that the relationship between wo and L is estimated by wo 



m 0 Qo E0

1=4

L1=4 :

(19)

It is shown here that fracture growth is expected to be characterized by decreasing thickness ratios or lateral spreading, as its length L increases. Taking the parameter ranges above and L ¼ 10 m, one finds that 0.001 , wo/L , 0.01, which is consistent with the available data for very small intrusions. However, if one considers cases for which L ¼ 10 km, one finds that wo/L , 0.0004 so that the predicted range of thickness ratios is smaller than what has been measured on natural sills. Interestingly, the departure of the predictions based on viscosity-dominated fractures from the available data corresponds approximately to the point at which intrusions can safely be expected to have grown to a lateral extent that is significantly greater than the depth of emplacement. Again, making use of previously derived scaling relationships, in this case for a shallow (i.e. L . 5H, where H is the emplacement depth) circular fracture that is governed by the elastic fracture/lubrication model and is propagating in the viscositydominated regime, one finds that (Bunger 2005) wo 



m0 Qo E0 H 3

1=4

L:

(20)

For the parameter ranges used above one can demonstrate thickness ratios up to wo/L  0.01. Recall, however, that the ranges above do not account for the large viscosities associated with granitic magmas. Taking, for example, m ¼ 106 Pa s one obtains a thickness ratio of wo/L  0.07. So consideration of the effect of the Earth’s surface on fracture growth would be expected to give sill thickness ratios that are within the range of the observed data. Furthermore, one finds that these shallow, viscosity-dominated fractures are expected to be characterized by growth with a constant thickness ratio wo / L, which is reasonably consistent with the scaling observed for large intrusions (McCaffrey & Petford 1997). Promising comparisons such as this should drive a methodical and rigorous improvement of the mathematical model for the emplacement of sills and other intrusions. For example, one may hope that as we come to better understand large-scale

79

rock fracture and the formation of tabular/sheeted intrusions, this model may be extended to encompass the range of intrusion geometries observed in Nature. Others may suggest that elastic fracture mechanics should be replaced by approaches that account for the details of the plastic deformation in the fracture tip region, but it is important to recognize that such an approach requires plasticity at a sufficiently large scale relative to the fracture size. Hence one must use caution owing to the fact that it is possible to have inelasticity large relative to the observational (i.e. outcrop) scale, but still small relative to a kilometre-sized intrusion, so that elastic fracture mechanics may in fact still be applicable. Specifically, the present theory is built on the premise that there exists a lag region between the fluid and fracture fronts which is very small compared with the fracture length, but much larger than the zone of near-tip plastic deformation. Although the interplay among length scales in the near-tip region is a complex matter, a reasonable starting point is to require that the intrusion length be about 1000 times or more than the observable zone of plastic deformation.

Conclusions The lubrication/elastic fracture model, in various forms, is perhaps the most common approach to modelling magma transfer in the Earth’s crust. While solving this coupled, non-linear system of equations for non-ideal geometries is in general an arduous task, the near-tip region possesses a universal form that is leveraged in the present work in order to devise a geometry-independent method for determining the regime of propagation for magma-driven sills. In so doing, it has been demonstrated that viscous dissipation is expected to be an important consideration in modelling sill formation, owing to the fact that most of the combinations of relevant governing parameters lead to sills that are viscosity-dominated or in the transition regime. As a result, numerical solution methods must be devised so as to account for the effect of fluid –solid coupling on the near-tip behaviour, for example by imposing the 2/3 asymptote of Equation 14 rather than the LEFM asymptote (Equation 6) at the fracture tip in the case of viscosity-dominated fractures. A number of implications for evaluation of geological data based on mechanical modelling have been discussed. Firstly, it has been demonstrated that a full understanding of the propagation regime relevant to field cases requires a clearer understanding of how the fracture energy/ toughness scales with fracture size. Secondly, it

80

A. P. BUNGER

has been shown that naive application of the LEFM tip asymptote in modelling intrusion growth can lead to application of the wrong physical process governing fracture extension, and therefore to erroneous results. Finally, promising results have been presented, which indicate that the linear elasticity/lubrication model considered here is likely to be capable of producing realistic predictions of the thickness to length ratios for magmatic intrusions, provided that viscous dissipation and nearsurface effects are considered. The present work arose out of a series of discussions with Emmanuel Detournay, for which I am sincerely grateful. Funding for this work has been provided by the Petroleum Research Fund administered by the American Chemical Society (grant no. ACS-PRF 43081-AC8), with additional support from Schlumberger. This support is gratefully acknowledged.

References A DACHI , J. 2001. Fluid-driven Fracture in Permeable Rock. Ph.D. thesis, University of Minnesota, Minneapolis, MN. B REITKREUZ , C. & P ETFORD , N. 2004. Physical geology of high-level magmatic systems: introduction. In: P ETFORD , N. & B REITKREUZ , C. (eds) Physical Geology of High-Level Magmatic Systems. Geological Society, London, 1– 4. B UNGER , A., D ETOURNAY , E., G ARAGASH , D. & P EIRCE , A. 2007. Numerical simulation of hydraulic fracturing in the viscosity dominated regime. In: Proceedings SPE Hydraulic Fracturing Technology Conference, College Station, Texas, SPE 106115. B UNGER , A. P. 2005. Near-Surface Hydraulic Fracture. Ph.D. thesis, University of Minnesota, Minneapolis, MN. B UNGER , A. P. & D ETOURNAY , E. 2008. Experimental validation of the tip asymptotics for a fluid-driven fracture. Journal of the Mechanics and Physics of Solids (submitted). C ARPINTERI , A. 1994. Scaling laws and renormalization groups for strength and toughness of disordered materials. International Journal of Solids and Structures, 31, 291– 302. C RUDEN , A. & M C C AFFREY , K. 2006. Dimensional scaling relationships of tabular igneous intrusions and their implications for a size, depth and compositionally dependent spectrum of emplacement processes in the crust. EOS Transactions of the AGU, 87, Abstract V12B– 06. D ESROCHES , J., D ETOURNAY , E., L ENOACH , B., P APANASTASIOU , P., P EARSON , J., T HIERCELIN , M. & C HENG , A.-D. 1994. The crack tip region in hydraulic fracturing. Proceedings of the Royal Society of London A, 447, 39–48. D ETOURNAY , E. 2004. Propagation regimes of fluiddriven fractures in impermeable rocks. International Journal of Geomechanics, 4, 1– 11.

F IALKO , Y. A. & R UBIN , A. M. 1997. Numerical simulation of high-pressure rock tensile fracture experiments: evidence of an increase in fracture energy with pressure? Journal of Geophysical Research, 102, 5231–5242. G ARAGASH , D. & D ETOURNAY , E. 2000. The tip region of a fluid-driven fracture in an elastic medium. ASME Journal of Applied Mechanics, 67, 183– 192. G ARAGASH , D. I. & D ETOURNAY , E. 2005. Plane strain propagation of a hydraulic fracture: small-toughness solution. ASME Journal of Applied Mechanics, 72, 916–928. K AZAN , Y. M. & F IALKO , Y. A. 1995. Fracture criteria at the tip of fluid-driven cracks in the earth. Geophysical Research Letters, 22, 2541–2544. K HRISTIANOVIC , S. & Z HELTOV , Y. 1955. Formation of vertical fractures by means of highly viscous fluids. In: Proceedings of 4th World Petroleum Congress, Carlo Colombo, Rome, 579– 586. L ECAMPION , B. & D ETOURNAY , E. 2007. An implicit algorithm for the propagation of a hydraulic fracture with a fluid lag. Computer Methods in Applied Mechanics and Engineering, 196, 4863–4880. L ISTER , J. 1990. Buoyancy-driven fluid fracture: The effects of material toughness and of low-viscosity precursors. Journal of Fluid Mechanics, 210, 263–280. L ISTER , J. R. & K ERR , R. C. 1991. Fluid-mechanical models of crack propagation and their application to magma transport in dykes. Journal of Geophysical Research, 96, 10,049–10,077. M C C AFFREY , K. & C RUDEN , A. 2002. Dimensional data and growth models for intrusions. In: B REITKREUZ , C., M OCK , C. & P ETFORD , N. (eds) First International Workshop: Physical Geology of Subvolcanic Systems – Laccoliths, Sills, and Dykes (LASI). Wissenschaftl. Mitt. Inst. Geol, TU Freiberg, Freiberg, 37–39. M C C AFFREY , K. & P ETFORD , N. 1997. Are granitic plutons scale invariant. Journal of the Geological Society, London, 154, 1 –4. P ETFORD , N., C RUDEN , A., M C C AFFREY , K. & V IGNERESSE , J.-L. 2000. Granite magma formation, transport and emplacement in the earth’s crust. Nature, 408, 669– 673. R ICE , J. 1968. Mathematical analysis in the mechanics of fracture. In: L IEBOWITZ , H. (ed.) Fracture, an Advanced Treatise, Vol. II. Academic Press, New York, 191 –311. R ICE , J. 1974. Limitations to the small scale yielding approximation for crack tip plasticity. Journal of the Mechanics and Physics of Solids, 22, 17–26. R UBIN , A. M. 1993. Tensile fracture of rock at high confining pressure: implications for dike propagation. Journal of Geophysical Research, 98, 15 919–15 935. R UBIN , A. M. 1995. Why geologists should avoid using ‘fracture toughness’ (at least for dykes). In: B AER , G. & H EIMANN , A. (eds) Physics and Chemistry of Dykes. Balkema, Rotterdam, 53– 63. S AVITSKI , A. & D ETOURNAY , E. 2002. Propagation of a penny-shaped fluid-driven fracture in an impermeable rock: asymptotic solutions. International Journal of Solids and Structures, 39, 6311–6337.

VISCOUS DISSIPATION IN SILL GROWTH S PENCE , D. & S HARP , P. 1985. Self-similar solution for elastohydrodynamic cavity flow. Proceedings of the Royal Society of London A, 400, 289– 313. S PENCE , D., S HARP , P. & T URCOTTE , D. 1987. Buoyancy-driven crack propagation: a mechanism for magma migration. Journal of Fluid Mechanics, 174, 135– 153.

81

W ILLIAMS , M. L. 1957. On the stress distribution at the base of a stationary crack. ASME Journal of Applied Mechanics, 24, 109–114. W OODS , A., B OKHOVE , O., DE B OER , A. & H ILL , B. 2006. Compressible magma flow in a two-dimensional elastic-walled dike. Earth and Planetary Science Letters, 246, 241– 250.

Dynamics of magma supply to Kı¯lauea volcano, Hawai‘i: integrating seismic, geodetic and eruption data THOMAS L. WRIGHT1 & F. W. KLEIN2 1

US Geological Survey, Johns Hopkins University Department of Earth and Planetary Sciences, Baltimore, MD 21218, USA (e-mail: [email protected]) 2

US Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025, USA (e-mail: [email protected])

Abstract: We focus on movement of magma beneath Kı¯lauea from the long summit eruption in 1967– 1968 through the first historical sustained eruption on the east rift zone (Mauna Ulu 1969– 1974), ending with the occurrence of a magnitude 7.2 earthquake beneath Kı¯lauea’s eastern south flank. Magma from the Hawai‘iian hot spot continuously moves upward to summit storage and drives seaward spreading of Kı¯lauea’s south flank on a 10– 12 km deep de´collement. Spreading creates dilation in Kı¯lauea’s rift zones and provides room to store magma at depths extending to the de´collement surface. During the period of study three types of eruptions – normal (shortlived), episodic and sustained – and three types of intrusions – traditional (summit to rift), inflationary and slow – are classified. Rates of sustained eruption are governed by the geometry of the magmatic plumbing. Swarms of earthquakes beneath the south flank signal increased pressure from magma entering Kı¯lauea’s adjacent rift zone. Magma supply rates are obtained by combining the volume of magma transferred to sites of eruption or intrusion with the volume opened by seaward spreading over the same increment of time. In our interpretation the varying character of eruptions and intrusions requires a gradual increase in magma supply rate throughout the period augmented by incremental increases in spreading rate. The three types of eruptions result from different combinations of magma supply and spreading rate.

Magma supply at the active Kı¯lauea volcano is considered from first principles, i.e. addressing the means by which magma makes its way from a shallow reservoir located 1–6 km below Kı¯lauea’s summit to eruption or shallow intrusion at Kı¯lauea’s summit or rift zones. Magma is transported in a resistant edifice of rock to fill space created by dilation of Kı¯lauea’s rift zones during seaward spreading of the adjacent south flank. Limitations that the geometry of the magma delivery system may have on eruption rates are considered, and times of initiation of swarm seismicity and ground deformation (the start of magma movement) relative to times of arrival of magma at the surface are compared. Further understanding of magma transport and spreading is obtained through an understanding of these time differences. Seismic, geodetic and geological data are used to relate rates of supply to both rates of eruption and rates of spreading, and to address variation of spreading and supply rates over various timescales. The ultimate goal of studies like this is the time history of absolute magma supply volumes at Kı¯lauea volcano. The Mauna Ulu eruption occurred from 25 May 1969 to 18 June 1974, and represents the first attempt within Kı¯lauea’s modern history at continuous eruption on the east rift zone. The time

period encompassing this eruption, 1967–1976, was chosen for study, based on the great variation in eruption styles, the frequency of intrusion, the very high rates of seismicity and the dramatic occurrence of the M 7.2 south flank earthquake that closes the period. Since 1983 Kı¯lauea has been in continuous eruption for nearly 25 years at essentially the same vent location. During the first 10 years there were no intrusions and no eruptions elsewhere on the volcano. Following the beginning of eruption, the east rift zone and summit were seismically quiet. By contrast, the Mauna Ulu eruption could not maintain continuous activity for more than about two years at a time, and its activity was accompanied by many intrusion-related earthquake swarms beneath the rift zones and by eruptions at other locations. The Mauna Ulu period is studied to evaluate the evidence for changes in the rates of magma supply and spreading over time, including an explicit assessment of earthquake swarms and their relationship to magma transfer and south flank spreading.

Background and previous work Kı¯lauea volcano is built on the flanks of the much larger Mauna Loa volcano and consists of a

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 83– 116. DOI: 10.1144/SP304.5 0305-8719/08/$15.00 # The Geological Society of London 2008.

84

T. L. WRIGHT & F. W. KLEIN

summit caldera containing a smaller pit crater (Halema‘uma‘u crater) and two rift zones extending east and SW from the summit (Fig. 1). Between the rift zones is a zone of north-facing grabens, the Koae fault zone. Kı¯lauea’s south flank is located seaward of the two rift zones and Koae fault zone and is moving seaward along sub-horizontal slip surfaces extending to 10–12 km depth (e.g. Swanson et al. 1976a; Ando 1979; Hill & Zucca 1987; Ryan 1988) at rates of up to 10 cm year21 (e.g. Owen et al. 1995; Delaney et al. 1998). Magma is supplied for eruption or intrusion from a reservoir defined by seismic and deformation data to be located 1–6 km beneath Kı¯lauea’s summit (Eaton 1962; Koyanagi et al. 1974; Cervelli & Miklius 2003). Magma transport paths have been defined from the Hawai‘ian hotspot to the shallow reservoir (Wright & Klein 2006), and from the reservoir laterally to the rift zones (Klein et al. 1987). Magma transfer above 5 km is accompanied by rapid deflation measured by the tiltmeter at Uwekahuna (Fig. 1, top left panel), and by earthquake swarms beneath the rift zones and summit (Klein et al. 1987). An early study considered the mechanics of magma transport from the melting source to shallow storage (Shaw 1980). Recent work postulates an expanded zone of lateral magma transport extending to the de´collement (Ryan 1988; Delaney et al. 1990; Johnson 1995; Wyss et al. 2001). Magma transfer deeper than 5 km is represented by the infrequent occurrence of earthquakes beneath the rift zones at depths extending from 5 km to the de´collement. Some deeper magma transport may be aseismic, as evidenced by the absence of earthquakes associated with the resupply of the reservoir in Kı¯lauea’s lower east rift zone which was evacuated in 1955 (Macdonald & Eaton 1964). Crystallization of magma during upward transport results in a dense pile of accumulated olivine that serves as a gravitational aid to seaward spreading (Clague & Denlinger 1994). Work on magmatic gases indicates that little CO2 is detected during rift activity, indicating that almost all of the CO2 is lost from the magma during upward transport and released beneath Kı¯lauea’s summit (Gerlach & Graeber 1985; Greenland et al. 1985; Gerlach et al. 2002). Most magma for rift eruptions must therefore experience some residence time in Kı¯lauea’s summit reservoir before being erupted or intruded. Modern study of Kı¯lauea volcano begins in 1952 with the return of activity to Halema‘uma‘u crater. Subsequent summit eruptions took place in 1954, 1959, 1967–1968, 1971 and 1982 interspersed with eruptions and intrusions on and beneath the two rift zones, the east rift zone being the more

active. Summit eruptions in 1952 and 1967– 1968 were sustained for more than 6 months. East rift activity includes eruptions on the lower east rift zone in 1955 and 1960, several eruptions on the middle and upper east rift zone in the 1960s, and two periods of sustained activity, 1969– 1974 ¯ ‘o¯ (Mauna Ulu) and 1983–present (Pu‘u ‘O Kupaianaha). Both of these eruptions began with periods of episodic high-fountaining followed by periods of sustained effusion from one or more vent locations. Rift activity is preceded and accompanied by shallow (,5 km) earthquake swarms beneath the site of eruptions and intrusions (Klein et al. 1987). Seismic activity beneath Kı¯lauea’s south flank shows a continuous background (M  2) of less than one earthquake per day punctuated by small seismic swarms (M  3) and by larger earthquakes (M . 4), some with aftershocks (Klein et al. 2006). Table 1 summarizes eruptive activity interpreted in this paper, beginning 5 November 1967 and ending with a small eruption and large intrusion beneath the SW rift zone on 31 December 1974 (Fiske & Koyanagi 1968; Kinoshita et al. 1969; Jackson et al. 1975; Swanson et al. 1976b, 1979; Duffield et al. 1982; Tilling et al. 1987; Lockwood et al. 1999). At the time of the 1952 eruption there was only one seismometer in operation. Beginning with the arrival of Jerry Eaton in 1953 a seismic network was put in place and expanded over the next two decades (Klein & Koyanagi 1980). Eaton also set up a network to measure ground tilt using a waterlevel system (Eaton 1959) and in 1957 a tiltmeter was installed in the Uwekahuna vault at Kı¯lauea’s summit. The tilt was read each morning and, up to the establishment of telemetered GPS in the 1990s, provided the only continuous measure of ground deformation during the years covered in this paper. The earliest quantitative estimate of Kı¯lauea’s magma supply (Swanson 1972) was made before the recognition that Kı¯lauea’s south flank was spreading seaward. Swanson concluded that magma supply was equivalent to the amount of lava extruded during three periods of sustained activity (1952 and 1967 – 1968 Halema‘uma‘u eruptions and the first 8 months of the Mauna Ulu eruption) after a correction of lava volumes for an assumed 15% vesicularity. Later estimates of magma supply (Dzurisin et al. 1984; Dvorak & Dzurisin 1993) still accepted the rate of lava emission during sustained eruptions (now corrected for 20% vesicularity) as a direct measure of total magma supply for those periods (Dvorak & Dzurisin 1993, Discussion, p. 22,265). The latter authors consider that magma supply has varied over both short and long periods of time. Both sets of references

KI¯LAUEA VOLCANO MAGMA SUPPLY

85

Fig. 1. Elements of Kı¯lauea’s tectonics and plumbing system taken from previously published work (Ryan 1988, plate 1; Delaney et al. 1990, figs 1A & 6A; Tilling & Dvorak 1993, fig. 1). The large lower figure outlines the overall geometry of the plumbing featuring a shallow reservoir beneath Kı¯lauea’s summit connected to magma-bearing regions extending vertically beneath Kı¯lauea’s two rift zones to a detachment surface (de´collement) at 10–12 km depth. The smaller figure on the upper left outlines the two rifts extending east and SW from Kı¯lauea’s summit. Arrows (lower left) show directions of shallow magma transport and (upper right) seaward spreading of Kı¯lauea’s south flank away from the two rift zones and connecting the Koae fault system. The deeper part of the magmatic system beneath the east rift zone is drawn on the lower right and shows a multi-level accommodation of Kı¯lauea’s south flank spreading away from the rift zone.

86

T. L. WRIGHT & F. W. KLEIN

Table 1. Eruptions at Kilauea between 1966 and 1976 Date* 5 November 1967 – 15 July 1968 22– 26 August 1968 7 –22 October 1968 22– 28 February 1969 25 May 1969 – 1 January 1970 24– 25 May 1969 27– 29 May 1969 12– 13 June 1969 25– 26 June 1969 15 July 1969 3 –4 August 1969 5 –6 August 1969 22 August 1969 6 –7 September 1969 10– 13 October 1969 20 October 1969 30 December 1969 1 January 1970 – 18 June 1971‡ 18 June 1971 – 5 February 1972 14 August 1971 24– 29 September 1971 5 February 1972 – 26 April 1973 26 April 1973 – 18 June 1974‡ 5 May 1973 10 November 1973 – 10 December 1973

Location†

References

ks: Halema‘uma‘u

Kinoshita, et al. (1969)

erz: Hiiaka erz: Napau erz: Aloi erz: Mauna Ulu Ia

Jackson et al. (1975) Jackson et al. (1975) Swanson et al. (1976) Swanson et al. (1979)

ep 1 ep 2 ep 3 ep 4 ep 5 ep 6 ep 7 ep 8 ep 9 ep 10 ep 11 ep 12 erz: Mauna Ulu Ib

Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson Swanson

Mauna Ulu pause

Duffield et al. (1982)

ks; erz ks; swr erz: Mauna Ulu IIa

Duffield et al. (1982) Duffield et al. (1982) Tilling et al. (1987)

erz: Mauna Ulu Iib

Tilling et al. (1987)

et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al.

(1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979)

erz: Pauahi erz: Pauahi

19 July 1974 19 September 1974 31 December 1974

ks, erz ks, swr swr

Lockwood, et al. (1999) Lockwood, et al. (1999) Lockwood, et al. (1999)

29 November 1975

Ks: Halema‘uma‘u

Tilling (1976)

Comment

Episodic high fountaining episodes (numbered) interspersed with intrusions

Continuous eruption from two vents (Alae and Mauna Ulu) interspersed with intrusions Mauna Ulu pauses; eruptions occur elsewhere on the volcano interspersed with or accompanied by intrusions Mauna Ulu resumes continuous activity – no intrusions Honomu earthquake changes character of eruption – two eruptions elsewhere on east rift zone are interspersed with continuous activity at Mauna Ulu; resumption of intrusions Mauna Ulu ends: activity is similar to Mauna Ulu pause with eruptions and intrusions elsewhere on volcano M7.2 Kilauea south flank earthquake triggered small eruption at Kilauea’s summit

*

Beginning and ending dates of activity. Locations: ks ¼ Kilauea’s summit; erz ¼ east rift zone; swr ¼ SW rift zone. Ending date is approximate, based on seismicity and observations at the vent.

† ‡

conclude that Kı¯lauea has been supplied at an average rate of about 0.1 km3 year21. Dvorak & Dzurisin also evaluated the amount of intrusion beneath Kı¯lauea’s rift zones, using the variation of ground tilt measured in the Uwekahuna vault at Kı¯lauea’s summit as a proxy for magma added to summit storage or transferred to the rift zones. They concluded that about 65% of Kı¯lauea’s magma supply is left underground between periods of sustained eruption (Dzurisin et al. 1984). One recent paper addresses both

magma supply rates and volume of Kı¯lauea’s summit reservoir (Denlinger 1997). Seismicity associated with eruption and intrusion was not considered in formulating the magma supply estimates given in the above studies. Several recent papers address the tectonics of dike emplacement (Gillard et al. 1996; Rubin & Gillard 1998) and the tectonic implications of the seismicity beneath Kı¯lauea’s south flank (Dieterich 1994; Dieterich et al. 2000). Recent work (Cayol et al. 2000) recognizes that quantifying the

KI¯LAUEA VOLCANO MAGMA SUPPLY

magma supply must take into account deep rift opening caused by the spreading of Kı¯lauea’s south flank, and their estimate of total magma supply for the period before the beginning of Kı¯lauea’s current eruption (0.19 km3 year21) is nearly twice the previous estimate.

Assumptions In this study we accept as working principles some conclusions reached by earlier workers, question others, and make some additional assumptions of our own: † Seaward spreading of Kı¯lauea’s south flank is assumed to be driven by magma supply and rift expansion, as noted by numerous authors beginning with Swanson et al. (1976a). The inactive Hawai‘ian volcanoes show no evidence of seaward movement, although some of these volcanoes experience occasional earthquakes at the base of the volcanic pile that have a ‘flat fault’ de´collement-type mechanism. The slopes of the older volcanoes are the same or steeper than Kı¯lauea’s, suggesting that gravity alone cannot drive seaward spreading. Numerical simulations of gravitational spreading have suggested that gravitational loading is a necessary component of spreading (Borgia 1994; Morgan & McGovern 2005). In this paper we consider that pressure exerted by magma supplied adjacent to Kı¯lauea’s south flank and gravitational loading are coupled as part of a single process. Normal faulting of the edifice, as represented by Kı¯lauea’s Hilina system, is considered to be a secondary effect. † The proposal by Cayol et al. (2000) that volume created by rift dilation is an important component of magma supply is accepted. † The earlier assertion that sustained eruption rates are constrained by the geometry of the magma delivery system (Wright & Fiske 1996) and cannot be an accurate proxy for magma supply is adopted as a working hypothesis. In addition, we note that magma moving underground is a viscous mixture of crystals, liquid and dissolved and exsolving gas that cools during transport, thus necessitating less than 100% transfer to the surface from any transport geometry because of crystal settling and cooling against the dyke wall. † Changes in tilt magnitude derived from daily measurement of east –west and north–south components at Uwekahuna are assumed to adequately represent magma addition to or transfer out of Kı¯lauea’s summit reservoir complex.

87

† The concept that earthquake rate changes beneath Kı¯lauea’s south flank can be used as a ‘stress meter’ (Dieterich et al. 2000) is accepted. Specifically, we use swarms of south flank earthquakes to reflect changes in magma pressure exerted on the south flank, and explore the possibility of using swarms as potential harbingers of change in magma supply or spreading rates. † A constant or gradually increasing melting rate at Kı¯lauea’s deep source in the Hawai‘ian hotspot is assumed. Therefore the long-term magma supplied from depth is expected to show little variation. However, no assumptions are made regarding short-term magma supply.

Presentation and use of seismic and tilt data Figure 2 shows a geographic classification of Kı¯lauea earthquakes developed earlier (Wright & Klein 2006). Earthquakes occurring along the magma supply path are distinguished from those occurring beneath the rift zones (sites of eruptions and intrusions), beneath the Koae fault system and beneath the seaward-moving south flank. The south flank is divided into segments that adjoin the different rift segments. Although the boundaries are often crossed, the subdivisions are useful in distinguishing periods of activity dominated by south flank seismicity in the eastern sectors from those dominated by south flank seismicity in the western sectors. With the exception of the magma supply path (see below), plotted earthquakes occur between 0 and 15 km depth. Rift earthquakes are mostly shallower than 5 km; south flank earthquakes extend from the surface to the base of the volcanic pile at about 12 km. Earthquake swarms are defined as consisting of at least three earthquakes within 6 h. All earthquake swarms of three or more events are plotted. Within the defined regions we consider only time, not location. A single stress event may activate an earthquake swarm that covers a wide area within that region. Many swarms occur over more than one region. For example, the northernmost east rift zone (er3uer) and seismic SW rift zone (er4sswr) overlap the southern part of Kı¯lauea caldera. Swarms classified as shallow caldera (ms1) occur north of the zones of overlap. Isolated earthquakes in the areas of overlap are also classified as summit events (ms1) whereas earthquake swarms in the region of overlap are classified as rift events (er3uer or er4sswr). Koae swarms are located mainly beneath the Koae fault system, but may extend laterally into the upper east rift zone or seismic SW rift zone. In all instances where an

88

T. L. WRIGHT & F. W. KLEIN

Fig. 2. Index map showing Kı¯lauea regions used to classify the seismicity and locations of the Kı¯lauea eruptions shown in Table 1. The east rift zone is divided arbitrarily somewhat east of the point where the rift trend changes from SE and east (upper east rift zone) to more east-northeasterly (middle east rift zone), and at the point beyond which eruptions are less frequent (lower east rift zone). The traditional SW rift zone defined by open cracks and pit craters is distinguished from the seismic SW rift zone, so named because it is the site of earthquake swarms associated with intrusions on the SW side of the volcano. The shaded magma supply path is taken from Wright & Klein (2006, fig. 9).

earthquake swarm extends into more than one region, the swarm activity for each of the affected regions is plotted. The magma supply earthquakes (ms1–5) are subdivided by depth. Magma supply earthquakes above 20 km (mostly above 15 km) are located directly beneath Kı¯lauea’s summit. Magma supply earthquakes deeper than 20 km spread to the south and most earthquakes deeper than 35 km are located off Kı¯lauea’s south shore. The magma supply earthquakes are treated in detail by Wright & Klein (2006). During the time period studied the seismic network operated by the US Geological Survey’s Hawai‘ian Volcano Observatory (HVO) was expanding. As a result the catalogue completeness magnitude decreased from approximately 2.7 in 1966 to 2.0 in 1969. For the south flank we want to make the long-term record comparable and also emphasize earthquake swarms. For comparability among south flank earthquakes occurring between 1967 and 1976 we plot all earthquakes that have magnitudes  2.6, the catalogue completeness magnitude for 1967. Some of these occur within

swarms. We believe the magnitude cutoff ensures that differences in frequency of isolated south flank earthquakes are real, even though the seismic processing sensitivity was increasing during this time. Swarm data for all regions occur over short periods of time. There may be some earthquakes of M , 2.6 missing from swarms in the earliest periods, but this does not compromise the identifcation of the larger swarms as we are concerned with numbers of events rather than magnitudes. Figure 3b provides a key to the presentation of seismic data.

Analysis of Uwekahuna tilt data The Uwekahuna tilt data is used as a guide to the location of centres of inflation and deflation of Kı¯lauea’s shallow summit reservoir. These were determined for the buildup to the 1967 –1968 eruption (Fiske & Kinoshita 1969, fig. 5). Depths to inflation and deflation centres were determined to be 2 –4 km, increasing to the east, as modelled for the period 1966–1970 (Dvorak et al. 1983, fig. 5). The latter authors also determined that model fits

KI¯LAUEA VOLCANO MAGMA SUPPLY

89

Fig. 3. Interpretation of Uwekahuna tilt magnitude and seismic data. The date format is mm/dd/yyyy. (a) Time series tilt data for summit and rift eruptions. See text for explanation. (b) Time series seismic data. Labelled regions are keyed to abbreviations on Figure 2. Long-period earthquakes are distinguished in the magma supply regions. Earthquakes beneath Kı¯lauea’s south flank are divided into swarms and isolated earthquakes of magnitude  2.6. See text for further explanation.

90

T. L. WRIGHT & F. W. KLEIN

obtained assuming a Mogi point-source were as good as fits assuming other source geometries. Subsequent work showed a linear correlation between volumes calculated assuming a Mogi source and the tilt change recorded at Uwekahuna (Dvorak & Okamura 1985, fig. 2), and this value was subsequently refined and used as a basis for subsequent magma supply calculations (Dzurisin et al. 1984; Dvorak & Dzurisin 1993). Mogi modelled volumes can vary by nearly an order of magnitude according to distance from the Uwekahuna tilt station and depth ranges of 2–4 km (Dzurisin et al. 1984, table II, p. 192). The azimuths of tilt changes shown in the Appendix, Table A1 agree within about 208, suggesting that the modelled depth is near-constant. A time plot of the daily tilt magnitude recorded in Uwekahuna vault at Kı¯lauea’s summit is shown in Figure 3a. Tilt value is a surrogate for magma volume stored in the summit reservoir. Thus, tilt data are treated differently for summit and rift eruptions in the magma supply calculations. During a summit eruption an increase of tilt magnitude is a positive magma volume quantity reflecting addition of magma to Kı¯lauea’s shallow reservoir, even as eruption is ongoing, meaning that the eruption rate is less than the rate of magma supplied from depth. Likewise, a decrease in tilt magnitude reflects drawdown of magma in the shallow

reservoir, meaning that the eruption rate exceeds the rate of magma supply from depth and is thus treated as a negative magma volume quantity. Very rapid decrease of tilt magnitude may occur at the beginning of eruption. A similar rapid increase of tilt magnitude can indicate shallow intrusion between the summit reservoir and the surface. During periods of rift eruptions, an increase of tilt magnitude is likewise interpreted as addition of magma to Kı¯lauea’s summit reservoir, treated as a positive magma volume quantity. However all decreases of tilt magnitude are also treated as positive magma volume quantities, reflecting transfer of magma from the summit reservoir to the rift plumbing. Rift eruptions and traditional intrusions (see intrusion type 1 of Table 2) are accompanied by very rapid decreases in tilt magnitude, reflecting transfer of magma from the summit reservoir to the site of eruption or intrusion. Thus, for rift eruption periods all tilt volumes are positive. Net volume changes based on tilt magnitude differences over a period of time covering many events is likewise positive. A positive tilt change indicates addition of magma to Kı¯lauea’s shallow reservoir. A negative tilt change indicates additional transfer of magma to the rift zones. Magma volumes cannot be calculated from

Table 2. Classification of intrusions Type

Tilt change at Uwekahuna

Seismicity Summit

1

.1 s deflation

1a

.1 s deflation

2

,1 s deflation or inflation

3

,1 s deflation or inflation

Variable

Rift

Description South flank

High

High

Mod– low

Mod – low

Variable

High

Low

Variable

Low

High

Traditional intrusion: Occurs alone or accompanying rift eruption. Strong seismic response to strong deflation Weak seismic response to strong deflation Inflationary intrusion: Rift swarm accompanied by low south flank seismicity and small summit tilt response. Some inflationary events of Klein et al. (1987) are of this type Slow intrusion: Periods of intense south flank seismicity accompanied by little or no seismic activity elsewhere, and by little or no tilt change. Most slow intrusions of Klein et al. (1987) are of this type

Note: variants of these major types are seen and, over a period of a few weeks, one type may follow or precede another. Swarms beneath Kı¯lauea’s summit may be important during some periods, and absent during others. See text for further explanation and interpretation.

KI¯LAUEA VOLCANO MAGMA SUPPLY

periods of no tilt change when magma is inferred to leave the summit reservoir for the rift at the same rate it is supplied from depth.

Classification of eruptions and intrusions Eruption styles. The most common rift eruptions at Kı¯lauea are short (days to weeks) and begin along a line that quickly settles on a single vent. Subsequent rift eruptions generally occur several months later on different parts of the rift zone. These are denoted as normal eruptions. The three east rift eruptions that preceded Mauna Ulu (Table 1) are typical of normal eruptions. A second type of eruption, denoted episodic is marked by return to the same vent at intervals of days to weeks. An entire sequence may last weeks to years. Examples are the first eight months of the Mauna Ulu eruption (Swanson et al. 1979) and the first two-and-a-half years of the current ¯ ‘o¯ eruption (Heliker et al. 2003). Pu‘u ‘O A third type of eruption, denoted sustained is marked by uninterrupted activity at a single vent. These eruptions typically last for months to years. Examples are parts of the Mauna Ulu eruption (Swanson et al. 1979; Tilling et al. 1987) and the ¯ ‘o¯ eruption (Heliker et al. 2003). current Pu‘u ‘O Summit eruptions may occur in and around Halema‘uma‘u crater, separated in time by months to years. They include all three types defined above. Seismic classification of intrusions. Rift eruptions and intrusions have been traditionally associated with deflationary tilt at Kı¯lauea’s summit and often by uplift and/or graben formation near the site of eruption. Such events are nearly always accompanied by seismic swarms beneath the summit, rift zones, south flank or all three. Many examples are tabulated and illustrated in Klein et al. (1987). In the same paper two other types of events occur with little summit tilt change. Major south flank swarms of tens to hundreds of events are described as ‘slow intrusions’ and major rift swarms as ‘inflationary intrusions’. In Table 2 we formalize a classification of intrusions, taking into account the various combinations of tilt change and seismic activity in different regions. The relationship of seismicity to Uwekahuna tilt changes for each type of intrusion is shown in Figure 4.

Limitations on the quantitative analysis of magma supply and spreading In the two decades following the 1952 eruption, both volumes of erupted lava and the results of geodetic analysis are relatively imprecise. Over time, eruption observations were made by many

91

different observers and geodetic data were collected using a number of different methods. Estimation of lava effusion rates during the period covered in this paper depend on estimates of erupted lava volumes made by mapping the new flow area and estimating lava thickness. A factor is applied (0.8 in this paper) to convert lava volumes to equivalent magma volumes. Errors in estimated magma volume are probably on the order of +20%. Early geodetic measurements of ground tilt, vertical changes over a levelling network and distance measurement using a geodimeter were reasonably precise, but calculation of deformation volume was hampered by non-continuous measurement of vertical and horizontal changes. In particular, surveys were consistently made following an eruption or large earthquake, but the previous survey data were almost always made many months prior to the event so that both precursory deformation and deformation related only to the event are not distinguishable. The deployment of continuous GPS networks beginning in 1997 has made it possible to accurately model volumetric changes over the whole of Kı¯lauea (e.g. Cayol et al. 2000; Owen et al. 2000) and to precisely correlate seismicity with ground movement. Deformation networks before 1975 were not designed to measure spreading of Kı¯lauea’s south flank. In this paper we turn to seismicity to infer possible changes of spreading rate. In our analysis of Uwekahuna tilt we rely on two factors that help overcome the uncertainties. First, we use internally consistent measurements from the same or closely spaced eruptions. As in the analysis of Dvorak & Dzurisin (1993), we model the Uwekahuna tilt assuming a Mogi model of a magma chamber whose centre remains a constant distance from Uwekahuna, even as the azimuth to that centre changes over 10 – 20 degrees, and consider the relative deflation and inflation volumes among several eruptions and intrusions to be accurate and interpretable. Errors in relative volume of magma transfer calculated from tilt changes are on the order of +20%.

Observations Eruption periods In this paper we divide the time period from 29 December 1965 to the M7.2 south flank earthquake on 29 November 1975 into nine broad periods based on changes in eruptive behaviour (Table 3). Seismic and tilt data for all periods are summarized in Table A1 and presented in Figures 5–8.

92

T. L. WRIGHT & F. W. KLEIN

Fig. 4. Three types of intrusions at Kı¯lauea, distinguished according to tilt change and location of seismic swarms. See text and Table 2 for further explanation.

Precursory sequences to rift zone eruption and intrusion Increased activity on Kı¯lauea’s south flank has long been known to follow increased activity on

Kı¯lauea’s rift zones (Koyanagi et al. 1972, discussion p. D91 ff ). It was potentially difficult to understand how intrusion of magma at depths less than 5 km beneath a rift zone could trigger south flank earthquake activity at depths extending down to at

Table 3. Eruptive periods 1967 – 1976 Eruptive period Pre-1967 – 68 1967 – 68 eruption

Pre-Mauna Ulu

Mauna Ulu period Ib

29 December 1965 5 November 1967 5 November 1967 11 January 1968 14 July 1968 14 July 1968 22 May 1969

Eruption style* No eruption

Normal

Sustained

Inflation Deflation

Increasing

High Deflation

18 June 1971†

High Inflation

Mauna Ulu pause

18 June 1971 5 February 1972

Normal

Mauna Ulu period IIa

5 February 1972 26 April 1973

Sustained

26 April 1973 18 June 1974§

Low Deflation

1 January 1970

Mauna Ulu period IIb

Low

Sustained

Episodic

26 April 1973

Seismicity

Inflation

22 May 1969 1 January 1970

Distant earthquake

Net tilt change

Inflation‡

High: swarms dominate

Low Deflation

Sustained

Inflation

Moderate

Comment No earthquake swarms No intrusions No traditional or inflationary intrusions Slow intrusion beneath Kı¯lauea’s south flank 7 –10 January 1968

(Continued)

93

East rift eruptions on 22– 26 August 1968, 7 – 22 October 1968 and 22– 28 February 1969 Intrusive activity accelerates after 3/1969 Seismic SW rift zone activated during 2/1969 eruption 12 episodes of high-fountaining at Mauna Ulu Slow intrusions on 3 – 9 June 1969 and 30 September – 2 October 1969 Inflationary intrusion on 3 July 1969 Seismic SW rift zone activity accelerates before episode 12 Effusion at two neighbouring vents, Mauna Ulu and Alae Five paired inflationary intrusions beneath upper east rift zone and seismic SW rift zone, the last heralding the close of period Ib Eruptions at Kı¯lauea’s summit (14 August 1971) and SW rift zone (24 – 29 September 1971) Intrusion beneath Kı¯lauea’s summit on 24 September 1971 Slow intrusion 22 – 29 September 1971 No activity beneath either rift zone from c. 3 October 1971 to 13 November 1971 Virtual absence of seismic activity beneath the rift zones and greatly reduced activity beneath the south flank Deep M6.2 earthquake off the east flank of Mauna Kea triggered changes in the Mauna Ulu eruption Mauna Ulu continuity interrupted in May and November 1973 by two eruptions in Pauahi crater East rift inflationary swarms return

KI¯LAUEA VOLCANO MAGMA SUPPLY

Mauna Ulu period Ia

Begin/end

T. L. WRIGHT & F. W. KLEIN

Eruptions at Kı¯lauea’s summit and upper east rift zone (19 July 1974) and SW rift zone (19 September 1974), the latter accompanied by summit intrusion Small eruption (31 December 1974) and large intrusion beneath Kı¯lauea’s SW rift zone; East rift seismicity ends on 12/31, seismic SW rift seismicity ends two weeks later. Both pick up again 3 – 4 weeks before earthquake M7.2 eastern south flank earthquake

least 10 km. Figure 5 shows precursory earthquake sequences to three rift zone eruptions and four intrusions preceding the Mauna Ulu eruption. Figure 5 includes the onset of swarms of earthquakes too small to locate (earthquake counts) and swarms of located earthquakes beneath different regions, onset of deflation measured at Uwekahuna and the beginning of eruption. Similar earthquake sequences are observed for episodes 1–12 of the Mauna Ulu eruption. For nearly all events the order of occurrence is earthquake counts ! deflation ! eruption. For most eruptions the located south flank swarm activity precedes in time swarm activity located beneath the rift zones, both of which precede eruption. Intrusions without eruption also commonly show south flank activity preceding rift activity, onset of summit deflation or both. These observations are entirely consistent with work cited above in support of magma transfer extending to 10 –12 km depth beneath the rift zones. They indicate that magma movement is initially sensed at the depths at which south flank seismicity occurs, and that movement of magma toward the eventual site of eruption triggers lateral transfer of magma from Kı¯lauea’s summit reservoir.

High: many swarms

Seismicity

Comment

94

See text. Period end based on tilt and seismicity. Lava remained visible in Mauna Ulu through October 1971. Inflation dominated by shallow intrusion beneath Kı¯lauea’s summit on 9/24/1971. § Period end based on tilt and seismicity. Lava remained visible in Mauna Ulu through July 22, 1974. k Inflation dominated by shallow intrusion beneath Kı¯lauea’s summit beginning on 9/18/1974. ‡

†

*

29 November 1975 Earthquake

Inflation:k Deflation Inflation Normal 18 June 1974 31 December 1974 17 January 1975 29 November 1975 29 November 1975 Post-Mauna Ulu

Net tilt change Eruption style* Begin/end Eruptive period

Table 3. Continued

Eruption efficiency Eruption efficiencies are calculated as the ratio of volume erupted to volume of magma transferred during summit deflation (Table A1). Eruption efficiencies for rift eruptions from August 1968 through Mauna Ulu period Ia are plotted in Figure 6. The efficiencies increase during 1968 and 1969, the period preceding Mauna Ulu, and again during Mauna Ulu period Ia. Over the short periods of time represented, these are one representation of the increased ability of the east rift magma plumbing to deliver magma to the surface.

Seismic and tilt sequences Seismic sequences from 1 January 1967 to the earthquake on 29 November 1975 are compared in Figures 7 and 8. Figure 7 summarizes data for entire periods, referenced to an expanded explanation shown in Figure 3. The time axis on these plots is scaled similarly in order to facilitate comparison of rates of seismic activity. Figure 8 shows 10-day windows surrounding times of intensified seismic activity. Tilt magnitudes measured at Kı¯lauea’s summit are shown on the same timescale in the upper part of Figure 7. Tilt magnitudes are in seconds of arc (1 s  5 mrad) and are referenced to an arbitrary 0. Upward trends record expansion or inflation of Kı¯lauea’s summit reservoir. Downward

KI¯LAUEA VOLCANO MAGMA SUPPLY

95

Fig. 5. Onset times for (1) harmonic tremor and swarms of earthquakes too small to locate (tabulated ‘earthquake counts’ in HVO seismic summaries) at Kı¯lauea’s summit, (2) located earthquake swarms beneath Kı¯lauea’s summit, rift zones, Koae fault zone and south flank, and (3) deflation at Kı¯lauea’s summit. These times are compared with the beginning of rift eruptions and intrusions in the pre-Mauna Ulu period. The date format is mm/dd/yyyy. The vertical axis is an unlabelled arbitrary scale used to separate the different labelled regions. Symbols and colours are identical for rift segments and the adjacent south flank, but they are plotted on different lines corresponding to each region.

trends represent magma transfer out of the summit reservoir. An expanded explanation of the tilt plots is given above (‘Analysis of Uwekahuna tilt data’).

Seismic and deformation history: 1967 – 1976 This section summarizes seismicity and deformation during the time periods shown in Table 3.

Fig. 6. Eruption efficiencies calculated as the ratio of vesicle-corrected eruption volumes to the volume of magma transferred during sharp deflations at Kı¯lauea’s summit. The date format is mm/dd/yyyy. See text for further explanation.

96

T. L. WRIGHT & F. W. KLEIN

Fig. 7. Time plots showing tilt, seismicity and incidence of eruption and intrusion. The time axis is similar for all plots so that shorter time intervals are represented by a narrower plot. Large tickmarks are placed every three months; small tickmarks are at intervals of one week. The date format is mm/dd/yyyy. The upper panels show tilt magnitude in seconds measured at Uwekahuna. The lower panels show seismicity along the magma supply path, beneath the rift zones, Koae fault zone and the south flank. See introduction and Figure 2 for explanation of region abbreviations. Seismicity beneath the south flank is shown by two kinds of symbols. Isolated earthquakes of M  2.6 are shown as solid symbols, as indicated in the legend above each plot. Times of earthquake swarms are shown by lines without symbols plotted at the same vertical position as isolated earthquakes in the same region. Because of the short duration of earthquake swarms relative to the total timescale for each plot, the earthquake swarms appear as narrow vertical bars: (a) 1967–1968 eruption and its

KI¯LAUEA VOLCANO MAGMA SUPPLY

97

Fig. 7. (Continued) prelude (1 January 1967 to 15 July 1968); (b) the end of the 1967–1968 eruption to the end of Mauna Ulu period Ia (15 July 1968 to 1 January 1970); (c) Mauna Ulu period Ib (1 January 1970 to 18 June 1971); (d) Mauna Ulu pause (18 June 1971 to 5 February 1972); (e) Mauna Ulu period IIa (5 February 1972 to 26 April 1973); (f) Mauna Ulu period IIb (26 April 1973 to 18 June 1974); (g) Mauna Ulu end to 1975 earthquake (18 June 1974 to 1 December 1975).

98 T. L. WRIGHT & F. W. KLEIN

Fig. 8. Ten day time plots showing details of seismic swarm activity (lower panel) and tilt change (upper panel). Large tickmarks and labels are placed every 5 days; small tickmarks are placed at the beginning of each day. Swarm data are connected by lines when time separating individual events is less than 6 h. Numbers of earthquakes are shown only when they comprise three or more events Eruptions and intrusions are shown at the top of the lower panels: (a) 1968– 1969 eruptions and intrusions preceding the Mauna Ulu eruption; (b) Mauna Ulu period Ia – precursory seismicity and slow intrusion between episodes 2 and 3;

KI¯LAUEA VOLCANO MAGMA SUPPLY

Fig. 8. (Continued) (c) Mauna Ulu period Ia – slow intrusion and end of high-fountaining activity; (d) Mauna Ulu period Ib – selected inflationary intrusions;

99

100 T. L. WRIGHT & F. W. KLEIN

Fig. 8. (Continued) (e) Mauna Ulu pause – intense seismicity near end of pause and precursory seismicity to the return of Mauna Ulu activity in period IIa; (f) Mauna Ulu period IIb – effects of Honomu earthquake, May 1973 eruption and one selected intrusion;

KI¯LAUEA VOLCANO MAGMA SUPPLY

Fig. 8. (Continued) (g) Mauna Ulu period IIb – seismicity preceding the end of eruption; (h) Mauna Ulu post – December 1974 eruption and intrusion.

101

102

T. L. WRIGHT & F. W. KLEIN

Information is summarized in Figures 7 and 8, which is then interpreted in the next section. Changes in eruption style, frequency of intrusions not associated with eruption, frequency and intensity of earthquake swarms and whether the longterm changes of Uwekahuna tilt mark deflation or inflation of Kı¯lauea’s summit are all empahasized. Seismicity increases toward the end of all periods, which helps to define the periods. 1 January 1967 to 14 July 1968. The 1967–1968 eruption was preceded by summit inflation accompanied by very low seismicity (background) in all regions (Fig. 7a). The summit deflated at the beginning of eruption and continued to deflate until late January 1968 (Kinoshita et al. 1969, fig. 2A). Reinflation began following a slow intrusion defined by significant swarm activity in the central and eastern south flank (Fig. 9). Eruption rates vary significantly (Kinoshita et al. 1969, table 2) and decrease when low fountaining and small tilt cycles gave way to continuous filling of the Halema‘uma‘u lava lake (Kinoshita et al. 1969, fig. 2). 14 July 1968 to 22 May 1969. Three east rift eruptions preceded the beginning of eruption at Mauna Ulu. The eruption of 22 –26 August 1968 was unusual in its lack of swarm seismicity (Figs 7b & 8a), even though a previously unbroken part of the rift zone was opened during the latter part of the eruption (Jackson et al. 1975, pp. 7–8). In contrast, the following eruption on 7–22 October 1968 was accompanied in the initial 4 days by intense earthquake swarms beneath Kı¯lauea’s

summit, east rift zone and adjacent south flank (Figs 7b & 8a). A third eruption on 22–28 February 1969 was accompanied by moderate seismicity (Figs 7b & 8a). Each eruption was accompanied by significant ground cracking and uplift, interpreted as intrusion beneath the rift zone (Jackson et al. 1975, figs 13, 18 and 32; Swanson et al. 1976b, fig. 20). Seismicity is low between eruptions, but increases with intrusive activity beginning in March 1969, premonitory to the beginning of the Mauna Ulu eruption. The period is marked by net deflation of Kı¯lauea’s summit. Paired inflationary intrusions beneath the northern termination of the seismic SW rift zone and the uppermost east rift zone begin toward the end of this period (Fig. 8b), and persist through the entire Mauna Ulu eruption (Klein et al. 1987). 22 May 1969 to 31 December 1969. The Mauna Ulu eruption began on 24 May 1969 with the first of 12 episodes of high fountaining (Mauna Ulu Ia). A prolonged slow intrusion follows episode 2 (Fig. 8b). Episodes 3 and 4 are aseismic even though summit deflation magnitudes are similar to those of previous episodes. An inflationary intrusion follows episode 4. A second prolonged long slow intrusion precedes episode 10 (Fig. 8c). Koae seismicity is intermittent. The western south flank is activated for the first time preceding and following episode 10 (Figs 7b & 8c). 31 December 1969 to 18 June 1971. Activity shifts from episodic to sustained (Mauna Ulu Ib). Five paired inflationary intrusions occur during this

Fig. 9. South flank swarm seismicity during the early stages of the 1967–1968 Halema‘uma‘u eruption. These swarms occurred toward the end of episodic tilt changes and a change from net deflation to net inflation for the remainder of the eruption (Kinoshita et al. 1969, fig. 2a). The date format is mm/dd/yyyy.

KI¯LAUEA VOLCANO MAGMA SUPPLY

period (Fig. 8d), accompanied by net inflation of Kı¯lauea’s summit (Fig. 7c). For many intrusions south flank seismicity is very low and a tilt response is very small, in contrast to the eruption and intrusion episodes in the previous period. Where deflationary tilt is detected, it occurs at the end rather than the beginning of the seismic sequences (Fig. 8d). 18 June 1971 to 5 Febuary 1972. The temporary end of sustained overflows at Mauna Ulu (Mauna Ulu pause) is marked by increased seismicity coupled with an increased rate of inflation, leading to eruption beneath Kı¯lauea’s summit and upper east rift zone on 14 August 1971 and eruption and intrusion beneath Kı¯lauea’s summit and SW rift zone on 24 –29 September 1971. Several paired inflationary intrusions occur between eruptions (Klein et al. 1987, figs 43.50, 43.51, 45.53 and 43.54). A prolonged slow intrusion occurs toward the end of the period, marked by intense seismicity beneath the central south flank, followed by an equally intense western south flank response (Figs 7d & 8e; Klein et al. 1987, fig. 43.56). Increased seismicity beneath the east rift zone and summit (Fig. 8e) precedes the return of activity to Mauna Ulu c. 5 February 1972. 5 February 1972 to 26 April 1973. The dramatic events of the preceding period are followed by the calmest period in the Mauna Ulu eruption (Mauna Ulu IIa, Fig. 7e). Quite in contrast to previous periods, there are no intrusions and Kı¯lauea’s summit shifts from slight inflation to slight deflation. A small deflation accompanies a short eruption on the west flank of Mauna Ulu in March 1972. 26 April 1973 to 18 June 1974. The Honomu earthquake of 26 April 1973 caused significant changes in the Mauna Ulu eruption (Fig. 7f; Tilling et al. 1987), including an immediate inflation at Kı¯lauea’s summit and increases in magma supply seismicity deeper than 10 km (Fig. 8f). Subsequently Mauna Ulu activity was supplanted by two eruption/intrusion sequences at the nearby Pauahi crater marked by sharp summit deflations (Table 1, Fig. 7f). The May eruption and a subsequent intrusion in June were also marked by earthquake swarms extending beneath the Koae fault zone (Fig. 8f), not seen since Mauna Ulu period Ia. As in Mauna Ulu period Ib intense seismicity beneath the rift zones was accompanied by very low rates of flank seismicity (Fig. 7f). Eruption rates at Mauna Ulu (Tilling et al. 1987, table 16.2) are greatly diminished compared with the preceding period. 18 June 1974 to 29 November 1975. The end of the Mauna Ulu eruption was remarkably similar to the end of Mauna Ulu period I, and the events following

103

cessation of Mauna Ulu activity were likewise a repeat of activity during the Mauna Ulu pause (Fig 7g). Two eruptions occurred, the first at Kı¯lauea’s summit and upper east rift zone (19–22 July 1974), and the second at the summit and upper SW rift zone (21 September 1974). As in 1971 the second eruption was accompanied by shallow intrusion beneath Kı¯lauea’s summit (Fig. 7g). Seismicity beneath the east rift zone increased again in October, shifting to the seismic SW rift zone in November and back to the east rift zone in early December. Seismicity beneath both rift zones rapidly escalated throughout the month (Fig. 8h), culminating in a short eruption on the SW rift zone on 31 December accompanied by intrusion extending almost two weeks into the new year (Fig. 8h). Seismicity beneath the east rift zone abruptly ended before the beginning of the eruption. This intrusion was accompanied by higher central and western south flank seismicity than in any of the preceding Mauna Ulu stages. Seismic activity beneath all regions continued at relatively high levels over the next several months. On 29 November 1975 the eastern south flank ruptured in an M7.2 earthquake.

Interpretation In this section we address the relationships among eruption rate, magma supply rate and spreading or rift dilation rate and how they influence the styles of activity that characterize each of the periods outlined in Table 3. We begin with an analysis of the 1967–1968 eruption (Table 4).

1967 – 1968 Halema‘uma‘u eruption Eruption rates are calculated for different subperiods of the eruption using data from Kinoshita et al. (1969, table 2 and fig. 2). Eruption volumes are converted to equivalent volumes of magma by correcting the published values by 0.8 (20% vesicularity). Inflation and deflation volumes are calculated from tilt changes measured in seconds using the factor of 0.00218 km3 s21 (Dvorak & Dzurisin 1993). In our calculations we use the daily north– south and east– west readings of tilt measured at Uwekahuna. These differ from the continuously recorded east– west component of tilt shown in figure 2 of Kinoshita et al. (1969), which are more precise, but are uncorrected for drift. Magma supply volumes are calculated by adding or subtracting the volume of tilt change (Table 4, column Tiltv) to the volume erupted (Table 4, column Ev) and then divided by the number of years to get the rate in column MSr. The periods

104

Table 4. 1967 – 1968 eruption: critical parameters* Period

Begin

End

Dt (years) Ev(km3)†

Er‡

DT (s)§

Tiltv(km3)** §§

MSr††

RZdr‡‡

Comment

Pre-eruption Phase 1

5 January 1966 25 October 1967 5 November 1967 9 November 1967

1.8015 0.01095

0.001592 0.1454

þ32.8066 0.071518 21.3082 20.00067

0.0397 0.01 0.0840 0.01

Phases 2 – 7 (6?)

9 November 1967 1 December 1967

0.06023

0.004752 0.0789

21.1302 20.002464

0.0380 0.01

Phases 8 (7?) – 23 1 December 1967 11 January 1968 7 January 1968 10 January 1968

0.11225

0.014104 0.1287

22.0276 20.006152

0.0863 0.01

Phases 1 – 23 5 November 1967 11 January 1968 Phases 24 – 28*** 11 January 1968 2 February 1968

0.18343 0.06023

0.020448 0.1115 0.002816 0.0468

24.2936 20.00936 þ1.0168 þ0.002217

0.0604 0.01 0.0836 0.011?

Phases 28 – 30 Entire eruption 1983 – present‡‡‡

0.42984 0.68993

0.035138 0.0818 0.060742 0.088 0.1264 0.1052

þ0.7348 þ0.00160 24.0920 20.00892

0.0859 0.011? 0.0751 0.025– 0.065§§§ Pu‘u‘O‘o: episodic fountaining do Kupaianaha: continuous

8 July 1968 14 July 1968††† 18 July 1986 7 February 1992

South flank earthquake swarm Increase in rift dilation rate following slow intrusion (south flank)

Data from Kinoshita et al. (1969, table 2). Eruption volume in Halema‘uma‘u corrected for vesicularity by a factor of 0.8 and given in km3. ‡ Eruption rate in km3 year21. § Net change of daily tilt at Uwekahuna during the times listed. þ ¼ summit inflation; 2 ¼ summit deflation. Differences compared with the continuously recorded east–west tilt shown in Kinoshita et al. 1969 (fig. 2) are ascribed to uncorrected drift of the continuously recording instrument. ** Volume of magma in km3 added (þ) or withdrawn (2) from Kilauea’s summit reservoir using factor of 0.00218. †† Magma supply rate in km3 year21 ¼ [Ev + Tiltv]/time in years ‡‡ Assumed rate of rift dilation rate due to spreading (km3/year21). §§ Comparison with volume in cubic kilometres calculated from Fiske and Kinoshita (1969, fig. 3A) 0.0183 (cone); 0.0367 (oblate spheroid). *** Before 2 February 1968. ††† Renewal of activity on 13 July added no new lava. ‡‡‡ Data from Heliker et al. (2003, table 1). §§§ Results of modelling rift dilation using GPS; see text for references. †

T. L. WRIGHT & F. W. KLEIN

*

2 February 1968 5 November 1967 3 January 1983 18 July 1986

Pre-eruption inflation Initial deflation of 22.77 s followed by re-inflation Tilt/phase description disagree

KI¯LAUEA VOLCANO MAGMA SUPPLY

of continuous activity with no fountaining yield similar rates 0.85 km3 year21 (column MSr, phases 24 –28 and 28–30) and we take this to be the equilibrium rate at which the sub-surface plumbing was able to provide magma for eruption. The true magma supply rate is obtained by adding an additional component of volumetric rift dilation related to seaward spreading of Kı¯lauea’s south flank. An initial value of 0.01 km3 year21, is assumed, significantly lower than values modelled for the current eruption from continuous GPS data (Table 4, bottom rows, column Rzdr, 1983–present). The volumetric rate of magma addition to Kı¯lauea’s summit during the period preceding the summit eruption is considerably less than the rate of eruption (Table 4 pre-eruption), suggesting that the opening of vents lowers the pressure on the magmatic system, resulting in a temporary increase in magma supply (Koyanagi et al. 1987, conclusion (7), p. 1256). Another way of expressing this is that the magma system is ‘throttled’ when the passages leading to the surface are sealed. Koyanagi’s conclusion was accepted by Dvorak & Dzurisin (1993, p. 22,265). This behaviour makes it impossible to quantify short-term magma supply rates. Even as magma may be delivered from the mantle at a near-constant rate, once entering the Kı¯lauea edifice its further rate of upward transport is subject to pressure fluctuations related to eruptive activity or lack thereof, in addition to the effects from rift dilation due to spreading. Why did the 1967–1968 eruption end, and why was it followed by an east rift eruption accompanied by unusually low seismicity? We hypothesize that an incremental increase in magma supply during the summit eruption triggered a south flank earthquake swarm (slow intrusion) on 7– 10 January 1968 (Fig. 9). The increased magma pressure caused an increase in spreading rate that dilated the upper east rift zone, forming an area of subsidence near the site of the August 1968 flank eruption (Kinoshita et al. 1969, fig. 6a). Dilation of the rift zone resulted in a diminished supply of magma to the summit eruption and created a favourable condition for a subsequent rift eruption to be largely aseismic. The analysis of the 1967–1968 eruption and its aftermath provides a framework within which to understand the period between the 1967–1968 summit eruption and the 1975 earthquake. Eruption and intrusion styles are interpreted in terms of stress applied to Kı¯lauea’s edifice, particularly the south flank, by incremental increases in magma supply from the mantle. On the principle that nature abhors a vacuum, we assume that the rift dilation volume will be occupied by incoming magma. A magma supply rate equal to the volumetric rate of dilation will suppress eruptions or

105

intrusions as well as inflation of Kı¯lauea’s summit. This is probably the case during extended periods of Kı¯lauea quiescence. The probability of eruption may also be reduced following periods of sudden dilation such as, for example, abrupt seaward motion following a large south flank earthquake or a major draining of the Kı¯lauea plumbing system, either of which requires some time to refill the magma system between 5 and 12 km depth. Near the surface, the rate at which magma can be delivered to the site of an eruption is controlled by the geometry of the magma plumbing. Intrusions, eruptions with a high component of intrusion, and inflation of Kı¯lauea’s summit are favoured when the magma supply rate exceeds the rate at which magma can be delivered to the surface. Over a long time span of frequent eruption the plumbing may evolve through heating and widening to allow a greater rate of eruption. The SW rift zone and associated seismicity beneath the western south flank is an important indicator of magma supply and the ability of the summit and east rift to accept magma. In a period of frequent eruptions on the east rift zone, we consider the western extension of the magmatic system in the SW rift zone to act like an overflow tank. Its activation is an indication that the magma supply cannot be accommodated solely by eruption or intrusion at the summit or east rift zone. Our analysis is consistent with the concept of Kı¯lauea’s south flank as a ‘stress meter’ (Dieterich et al. 2000) formulated for activity occurring during a more recent (1983 and later) period of time. We hypothesize that incremental increases in magma supply during a time when Kı¯lauea is erupting are first manifested as heightened seismicity beneath Kı¯lauea’s south flank. Increases of magma supply can be accommodated in one or more of the following: (1) increased rate of spreading; (2) inflation of Kı¯lauea’s summit; or (3) increased incidence of intrusion. The stress within the south flank manifested as slow intrusions can induce an increase of spreading rate. Volcanic history is an all-important variable in our analysis. Spreading rates and the geometry of the magma delivery system may evolve and change. Volcanic events, notably large eruptions, intrusions and earthquakes, change the ground state of the volcano and have a significant influence on succeeding events. Thus, the nature of each event in Kı¯lauea’s history must be interpreted in terms of both driving factors such as magma supply rate and strain accumulation, but also by what occurred during the preceding event or events. All the evidence in Figures 7 and 8 leads us to conclude that magma supply to Kı¯lauea was at

106

T. L. WRIGHT & F. W. KLEIN

least incrementally increasing during the entire period leading up to the 1975 earthquake. The increasing frequency at which earthquakes occur in all sectors of the volcano following the end of the 1967–1968 summit eruption suggests that significant magma pressure was being applied. This is particularly true in the period immediately preceding the Mauna Ulu eruption and in the periods following the end of continuous eruption at Mauna Ulu in the summers of 1971 and 1974. The seismic evidence is bolstered by the frequent intrusions, net inflation of Kı¯lauea’s summit during most periods, and by the activation of the western side of the magmatic plumbing.

Magmatic history: 14 July 1968 to 29 November 1975 Magma supply rates calculated from the data in Appendix A1 are summarized in Table 5 and Figure 10 for eruption cycles beginning with the period of inflation preceding the 1967–1968 summit eruption and ending with the 1975 earthquake. For the periods of sustained east rift eruption (Mauna Ulu periods Ib and II) we sum the eruption volumes, the net volumetric change measured over the entire period and the incremental magma transfer volumes associated with intrusion or eruption outside of Mauna Ulu. For periods of normal and episodic eruption where eruption efficiencies are less than unity, we use the magma transfer volumes calculated from the tilt for each event and add these quantities to the net change of tilt volume for the period. The volumes, expressed in cubic kilometres, are divided by the time in years to obtain the minimum magma supply rate (lower part of the bars shown in Fig. 10). An assumed component of rift dilation is added (upper parts of the bars in Fig. 10). Volumetric increases in rift dilation are based on the occurrence of slow intrusions. The spreading rate of Kı¯lauea’s south flank prior to the 1967–1968 eruption is not known. A baseline rift dilation rate of 0.01 km3 year21 in 1967 is assumed in our interpretations. The values in Table 5 include a component of rift dilation that increases from 0.01 to 0.05. Incremental increases in rate follow slow intrusions. All values shown in Table 5 and Figure 10 are assumed to approach the rates of rift dilation (0.25 –0.65 km3 year21) estimated for the current eruption (Delaney et al. 1993; Owen et al. 1995, 2000). The interpreted maximum rates at which the shallow plumbing could deliver magma to the summit and east rift zone in this period are labelled ‘S’ and ‘R’ on the plot. All of the estimates of total magma supply rate shown in Figure 10 are less than the rate of 0.19 km3 year21 calculated for Kı¯lauea’s current

eruption (Cayol et al. 2000), suggesting that increases in magma supply continued past the time period considered here. 14 July 1968 to 22 May 1969 (pre-Mauna Ulu). In this period we hypothesize that locations of prior eruptions on Kı¯lauea’s east zone affect the location and the degree of seismic activity in subsequent eruptions. Locations of the 1967–1968 summit eruption, the three eruptions on Kı¯lauea’s east rift zone that preceded Mauna Ulu, and Mauna Ulu are shown in Figure 11. We hypothesize that, following the aseismic eruption in August 1968 (see above), the east rift zone beneath the vents was sealed by intrusion, setting the stage for the highly seismic middle east rift eruption in October whose vents were located between the two sets of August vents. The vents of the moderately seismic upper east rift eruption in February 1969 overlap those of the preceding two eruptions. The Mauna Ulu eruption begins near the upper end of the February vents in an area not occupied during eruptions in the preceding year. The many intrusions and increasing seismicity preceding the beginning of the Mauna Ulu eruption are ascribed to an increase in magma supply rate over the value of 0.088 km3 year21 that the rift plumbing was capable of delivering to the surface. 22 May 1969 to 31 December 1969 (Mauna Ulu Ia). The intense south flank swarms following episode 2 are interpreted to have triggered an increase in spreading rate, allowing subsequent episodes 3 and 4 to be aseismic even as magma transfer rates as measured by summit deflation remained similar. The remainder of the Mauna Ulu stage Ia is marked by variable seismic response to different eruptive episodes, which we ascribe to continuing increase in magma supply. The long slow intrusion in October (Fig. 7c) is interpreted to have triggered a second increase in spreading rate, which is felt in the following period, Mauna Ulu stage Ib, where it can explain the very low south flank response to rather intense inflationary intrusions occurring beneath both rift zones (Fig. 8d). 31 December 1969 to 18 June 1971 (Mauna Ulu Ib). The combination of increased magma supply and increased spreading rate produced a critical balance that allowed a shift from episodic to sustained activity. The continuation of paired inflationary intrusions with little flank seismic response, combined with slight net inflation of Kı¯lauea’s summit, indicates that the magma supply rate during this period continued to exceed that necessary for sustained eruption. However the eruption rate for this period is half that of the previous period, and yields a low value for the magma

Table 5. Magma supply summary Eruptive period Pre-1967 – 68 1967 – 68 eruption

Mauna Ulu period Ia Mauna Ulu period Ib Mauna Ulu pause Mauna Ulu period IIa Distant earthquake Mauna Ulu period IIb Post-Mauna Ulu Earthquake

29 December 1965 5 November 1967 5 November 1967 14 July 1968 14 July 1968 22 May 1969 22 May 1969 31 December 1969 31 December 1969 18 June 1971†† 18 June 1971 5 February 1972 5 February 1972 26 April 1973 26 April 1973 26 April 1973 18 June 1974 18 June 1974 28 November 1975 29 November 1975

Dt (years)

V E/I (km3)*

1.8480

0

0.6899

0.6074 E

0.8542

V inf/def (km3)†

Msr (km3/ year21)§

0.01

0.0487

20.0089 def

0.011

0.0861

0.0772 I

0.0030 def

0.011

0.1272

0.6133

0.0948

0.0050 def

0.02

0.1777

1.4593

0.0174 inf

0.02

0.0857

0.6352

0.0792 E 0.0066 I 0.0174 I rift

0.025

1.2211

0.1000 E

0.0842 inf 0.0410 inf§§ 0.0081 def

0.035

0.1799‡‡ 0.1235§§ 0.1085

1.1444

0.0298 E 0.0273 I 0.0771

0.0177 inf

0.04

0.0854

0.0219 inf

0.05

0.0885

1.4456

0.0715 inf**

RD (km3/ year21)‡

‡‡

Comment

Preferred magma supply rate for sustained summit eruption

Adjusted for shallow intrusion Preferred magma supply rate for sustained east rift eruption Honomu earthquake Ms 6.2

Kalapana earthquake Ms 7.2

KI¯LAUEA VOLCANO MAGMA SUPPLY

Pre-Mauna Ulu

Begin/end

*Volume of eruption (measured) or intrusion (calculated from tilt). See text for further explanation. Volume of inflation or deflation measured from Uwekahuna tilt differences as shown in Figure 3. See text for further explanation. Assumed rate of rift dilation; increases follow slow intrusions. § Magma supply rate including rift dilation. ** Agrees with volume of uplift (Fiske & Kinoshita, fig. 3A) calculated as cone (0.0183 km3) or oblate spheroid (0.0367 km3). †† Period end based on tilt and seismicity. Lava remained visible in Mauna Ulu through October 1971. ‡‡ Large effect of summit intrusion on 9/24/1971. §§ After reducing volume of 9/24 intrusion by a factor of 4 assuming depth of intrusion c. 1km. † ‡

107

108

T. L. WRIGHT & F. W. KLEIN

Fig. 10. Magma supply rates (km3 yr21) for eruption periods with and without an assumed additional volumetric rate of dilation from spreading (Table 5). See text and Table 5 footnotes for calculation procedures. S and R denote preferred rates of sustained eruption at Kı¯lauea’s summit and east rift zone, respectively. The high value for the Mauna Ulu pause is probably exaggerated by overestimation of the volume of intrusion that occurred beneath Kı¯lauea’s summit but above the shallow magma reservoir. The arrow shows the estimated increase of mantle magma supply rate during the period between the sustained eruptions of 1967– 68 (Summit) and Mauna Ulu IIA (Rift). P denotes the magma supply rate calculated by Cayol et al. (2000) for the period near the beginning of the still-ongoing eruption that began in 1983.

Fig. 11. Locations of eruptions from 1967 to 1969. Shifting locations of eruptions on the east rift zone are interpreted to explain variations in the intensity of the accompanying seismicity.

KI¯LAUEA VOLCANO MAGMA SUPPLY

supply rate (Fig. 10). This indicates that either (1) the eruption rate is underestimated and/or (2) the spreading rate was higher than estimated in Table 5. A spreading rate of 0.035 km3 year21 during this period would reconcile the magma supply rates of Mauna Ulu periods Ib and IIa. If so, this suggests that our assumption of low spreading rate is wrong and that spreading rates could have been comparable to those observed during the current eruption. Unfortunately there is no relevant ground deformation data from 1969–1971 to validate this suggestion. 18 June 1971 to 5 February 1972 (Mauna Ulu pause). Increased rate of inflation toward the end of period Ib requires another increase in magma supply. Because of limitations on the amount of magma that could be erupted at Mauna Ulu, an unanticipated cessation of Mauna Ulu activity occurred as magma was diverted to the summit and the western side of the volcano, leading to intrusion and eruption beneath Kı¯lauea’s summit and SW rift zone. Toward the end of the period an increase in spreading rate is signalled by a slow intrusion. Magma pressure was relieved, leading to a renewal of intrusion beneath the east rift zone, preceding the return of magma to Mauna Ulu. The volume of intrusion beneath Kı¯lauea’s summit in September 1971 (Fig. 7d) used to calculate the magma supply may be overestimated as the intrusion was above the summit reservoir. A magma supply value of 0.1235 is obtained if the intrusion is assumed to be shallower by half the depth assumed in the Dvorak–Dzurison calculations, yielding an intrusion volume of one-quarter the value calculated using their factor. 5 February 1972 to 18 June 1974 (Mauna Ulu II). A shift from inflation to deflation at Kı¯lauea’s summit combined with the absence of intrusions and diminished south flank swarm seismicity suggest that, until 26 April 1973 (Mauna Ulu period IIa), magma supply and spreading rate were in balance with the maximum permissible eruption rate, represented by ‘R’ in Figure 10. The suggestion that the Honomu earthquake on 26 April 1973 significantly upset the balance described for Mauna Ulu IIa (Tilling et al. 1987) is supported by our analysis. The shaking in the Kı¯lauea summit region was considerable: the felt intensity was VII because water tanks were shifted off their foundations, numerous cracks and rockfalls occurred, and peak acceleration was 0.17g at 0.15 period (Nielsen et al. 1977). The eruption rate significantly decreased following the earthquake (Mauna Ulu period IIb). This suggests that the earthquake reduced the ability of the east rift magma plumbing to deliver magma to the

109

surface, and perhaps new underground pathways were shaken open. The occurrence immediately following the earthquake of a small swarm of earthquakes at 30 km depth beneath Kı¯lauea’s summit (Fig. 8f) suggests that this distant earthquake also triggered a surge of new magma from the mantle that caused (1) eruption elsewhere on the East rift zone, (2) increased intrusive activity and (3) a shift back to net summit inflation. 18 June 1974 to 29 November 1975 (post-Mauna Ulu). The end of the Mauna Ulu eruption can again be interpreted to be associated with an incremental increase in magma supply compared with that of the preceding period. As in the summer of 1971, we suggest that this led to eruption and intrusion at Kı¯lauea’s summit and a similar slow intrusion on the SW side of the volcano. The slow intrusion rapidly escalated into a major intrusion into the seismic SW rift zone, accompanied by a higher rate of south flank seismic release than in any of the preceding Mauna Ulu periods. Unlike the Mauna Ulu pause, the intrusion on the SW side of the volcano resulted in no further supply to the east rift zone, preventing another return to eruption at Mauna Ulu. Earlier analysis of the south flank tectonics led Swanson et al. (1976a) to forecast ‘a subsidence (strain-release) event of unknown magnitude in the not too distant future’. In our interpretation the weakening of the south flank during the long period of intrusion from late 1974 through the first quarter of 1975 was a proximate cause for the failure of the south flank on 29 November 1975, when an M7.2 earthquake occurred beneath the eastern south flank and caused much of the south flank to slide seaward.

Seismicity along the magma supply path from the mantle Statistics on the number of earthquakes occurring along the magma supply path (Fig. 1 and Wright & Klein 2006) in swarms are contrasted with the number not in swarms for the periods of Table 3 (Fig. 12). There is a striking increase in both the background seismicity and the number of swarm earthquakes in all regions through the Mauna Ulu pause. In Mauna Ulu period IIa numbers return to values similar to those characterizing the period preceding the 1967–1968 summit eruption. Numbers then increase again through the postMauna Ulu period, and remain high, particularly those deeper than 20 km, during the run-up to the 1975 earthquake. Mauna Ulu period Ib fits within the trend of increasing magma supply and seismicity, supporting the supposition above of a much higher spreading rate during this period. The

110

T. L. WRIGHT & F. W. KLEIN

Fig. 12. Histograms showing numbers of earthquakes in the magma supply path for different eruption cycles. Swarm seicmicity consists of sequences of contiguous days averaging greater than one event per day over a 5-day period, at least two earthquakes per day for shorter time intervals, and a minimum of three earthquakes if the time interval is only one day. Background includes all other events. Swarms are easy to pick out as they are concentrated in time, whereas there may be many days or even weeks separating background earthquakes: (a) earthquake swarms; (b) background seismicity.

similarity of the trends in Figures 12 and 10 suggest that magma supply variation within the Kı¯lauea edifice could reflect an incremental increase of melting rate within the mantle source. The magma supply from depth was estimated by Aki & Koyanagi (1981) to be 90% aseismic. The data of Figure 12 suggest that, even though sparse, the seismicity along the magma transport path carries important information regarding upward magma transfer below and within the Kı¯lauea edifice.

Significance of different kinds of intrusions The three types of intrusions described in Table 2 are interpreted to result from different aspects of the magma transport process. Traditional intrusions occur with or without accompanying eruption and represent shallow and rapid magma transfer from Kı¯lauea’s summit to the site of eruption or intrusion (cf. Klein et al. 1987). Inflationary intrusions extend over a longer time and may represent tectonic adjustments of the rift zones to the stresses generated by spreading, but are not necessarily triggered by direct magma transfer. These are consistent with the conclusions reached by Gillard et al. (1996) on the basis of their study of relocated Kı¯lauea earthquakes, i.e. ‘The shallow background seismicity

within Kı¯lauea’s rifts can therefore be tied to large-scale motions of the volcanic edifice, and need not be interpreted as resulting from magma migration’. Slow intrusions represent release of strain accumulated during magma pressure from the deeper, aseismic part of Kı¯lauea’s rift system, and we interpret them as precursory to changes in spreading rate. They may involve slow magma flow in existing channels as well as spreading. This interpretation is consistent with studies of Dieterich (Dieterich 1994; Dieterich et al. 2000) that relate changes in earthquake rate as a stress metre within the Kı¯lauea edifice.

Thoughts on eruption types and a new understanding of eruption rate Eruptions are initiated when pressure exerted from exsolution of volatile components in magma moving toward the surface exceeds the threshold beyond which magma can no longer be contained within the edifice. A trigger for some eruptions may be an incremental increase in magma supply. Initial eruption rates are very high, driven by volatile overpressures. Subsequently, the eruption rate decreases to a value sustainable by the delivery system. The short-lived eruptions will end because

KI¯LAUEA VOLCANO MAGMA SUPPLY

the magma supply rate is not sufficient to counter the drawdown in the magma plumbing before the delivery system closes through cooling against the walls. Episodic eruptions, as in Mauna Ulu period Ia, exhibit a geyser-like behaviour, which in our interpretation happens when the magma supply rate is high enough to keep the plumbing open in between eruptive episodes. Our ideas are consistent with those expressed in a detailed study of the episodic stage of Kı¯lauea’s current east rift eruption (Parfitt & Wilson 1994). Sustained eruption occurs when the magma supply is more than sufficient to keep the delivery system intact. The three types of eruptions could represent three levels of magma supply in excess of the volumetric spreading rate. In other words, if the spreading rate is constant, then sustained eruption would require a higher rate of magma supply than episodic eruption and short-lived eruptions would occur when the plumbing is not well-developed (cooler?) and magma must break a new pathway, causing an intense earthquake swarm, or during a period of relatively low magma supply rate above what is needed to drive spreading. The preceding analysis leads us to reconsider what constitutes an equilibrium eruption rate during sustained activity. Previous analyses (e.g. Swanson 1972) calculate eruption rates for an entire eruption cycle, including the 1967–1968 Halema‘uma‘u eruption and the high-fountaining period of Mauna Ulu (our Mauna Ulu period Ia). Initial eruption rates associated with volatile-driven fountaining are always higher than the eruption rates after fountaining subsides (e.g. Kinoshita et al. 1969, table 2; Jackson et al. 1975, table 1). Thus, more magma reaches the surface per unit time compared with what the same plumbing could deliver in sustained activity. The early stages of the 1967–1968 eruption were marked by cyclic low fountaining (Kinoshita et al. 1969), which may have also contributed to a greater eruption rate than what could be sustained without fountaining. Lower rates for sustained eruption are favoured, compared with those previously published. The best estimate of what the Mauna Ulu plumbing could deliver on an ongoing basis is represented by Mauna Ulu stage IIa, with a calculated eruption rate of 0.084 km3 year21, contrasted with the eruption rate of 0.106 km3 year21, cited for Mauna Ulu stage Ia. Likewise, if one calculates the eruption rate for episodes 24 –31 (13 January to 14 July 1968) of the Halema‘uma‘u eruption, one gets a value of about 0.072 km3 year21, again lower than the value of 0.107 km3 year21 calculated for the entire eruption. The relative values are consistent with the difference in rates between the episodic

111

and continuous periods of the current eruption (Table 4). We suggest that the lower figures are a more accurate representation of what the plumbing system feeding an eruption can deliver during periods of sustained activity. What causes the end of sustained activity? Among many possible factors, increase in magma supply well beyond what the shallow magmatic plumbing can deliver as sustained eruption is an important effect during the 1967 –1976 period of study. For Mauna Ulu the increased magma supply redirected magma to the west side of the volcano. Another contributing factor, represented by the 1967 –1968 summit eruption, may be abrupt increases in spreading rate that facilitate redirection of magma away from the eruption site. Finally, the increased seismicity that marks every change of eruptive behaviour suggests that the shallow plumbing is subject to failure through unpredictable blockages, analogous to collapses of lava tube systems (Kauahikaua et al. 2003, p. 68).

Conclusions Volcanic activity at Kı¯lauea is governed by the relationship among rates of magma supply from the mantle, rates of rift dilation associated with seaward spreading, and rates at which the Kı¯lauea plumbing can deliver magma to the surface. During the time period from 1966 to the 29 November 1975 M7.2 south flank earthquake, magma delivery rates were stable while both magma supply and spreading rates were undergoing incremental increases. South flank swarm seismicity acts as a stress metre and, when manifested as slow intrusions, precedes increases in spreading rate. The occurrence of summit inflation or deflation and frequency of intrusion are related to the balance between spreading and magma supply rates. Times of sustained eruption without either intrusion or summit inflation represent times when spreading rate and magma supply rate are both in balance with the rate at which the Kı¯lauea plumbing can deliver magma to the surface. Our study is continuing past the 1975 earthquake, when more frequent deformation data acquisition and broader geographic coverage are available, including the period of Kı¯lauea’s current sustained eruption, where continuous deformation data can be accessed. This paper benefitted greatly from the thorough and incisive US Geological Survey reviews from Don Swanson and Mike Poland. Journal review by Bill Leeman and Corrado Cigolini, and editorial comments from Georg Zellmer have materially improved the presentation.

Cycle 1967 – 1968

1968

Event inf E ks def inf def

Begin date 29 December 1965 5 November 1967 5 November 1967 6 November 1967 8 November 1967

Dt (years)

5 November 1967 14 July 1968 6 November 1967 8 November 1967 8 June 1968

1.8617 0.6899

25 August 1968 21 August 1968 25 August 1968 9 October 1968 7 October 1968 9 October 1968 23 February 1969 21 February 1969 23 February 1969 25 May 1969 20 March 1969 22 March 1969 7 May 1969 9 May 1969 22 May 1969 25 May 1969 25 May 1969 28 May 1969 29 May 1969 11 June 1969 13 June 1969 25 June 1969 26 June 1969 2 July 1969 4 July 1969 14 July 1969 15 July 1969 2 August 1969 4 August 1969 5 August 1969 6 August 1969 21 August 1969

0.2136

Tmag*

Taz†

32.81 314.63

Tvol‡

Evol§

Eeff**

msr pre-eruption 0.0384 km3/ year21 eruption rate (5 November 1967 – 15 July 1968) ¼ 0.107 km3/year21 Eruption rate (13 January – 15 July 1968) ¼ 0.073 km3/year21†† August 1968 eruption/intrusion

0.0715 0.0744

22.77 125.39 2.38 282.99 3.55 162.47

Comment

20.0060 0.0052 20.0077

2.14 303.69 12.12 114.32

0.0047 0.0264 0.00008 0.003

8.83 282.82 12.61 116.25

0.0193 0.0275 0.0075

13.08 289.09 9.28 117.88

0.0285 0.0202 0.016

0.1232

October 1968 eruption/intrusion 0.2727

0.3751

February 1969 eruption/intrusion 0.7921

0.2491

Mauna Ulu precursory intrusions 4.14 0.62 5.68 0.77 1.64

285.83 101.34 317.97 104.34 316.47

0.0090 2 0.0014 0.0124 2 0.0017 0.0036

5.16 0.32 1.52 4.03 3.18 3.40 4.12 2.86 0.57 2.29 2.86 4.55 2.56 0.48 2.61 4.18

110.54 21.80 120.58 291.64 112.35 299.25 114.72 286.93 161.56 294.57 110.70 289.86 107.59 330.26 120.07 284.83

0.0112 0.0007 0.0033 0.0088 0.0069 0.0074 0.0090 0.0062 0.0012 0.0050 0.0062 0.0099 0.0056 0.0010 0.0057 0.0091

0.00424 0.3786 Mauna Ulu eruption: period Ia Episodic high fountaining 0.00329 0.9954 0.00376 0.5449 0.00424 0.4711

0.00376 0.6065 0.00329 0.5875 0.00376 0.6597

T. L. WRIGHT & F. W. KLEIN

8 June 1968 inf 8 June 1968 August 1968 E erz def 21 August 1968 1968 25 August 1968 inf 25 August 1968 October 1968 E erz def 7 October 1968 1968 – 1969 9 October 1968 inf 9 October 1968 February 1969 E erz def 21 February 1969 1969 23 February 1969 inf 23 February 1969 I erz def 20 March 1969 inf 22 March 1969 I erz def 7 May 1969 inf 9 May 1969 MU Ia E erz 24 May 1969 ep 1 def 22 May 1969 trans 25 May 1969 ep 2 def 28 May 1969 inf 29 May 1969 ep 3 def 11 June 1969 inf 13 June 1969 ep 4 def 25 June 1969 inf 26 June 1969 I erz def 2 July 1969 inf 4 July 1969 ep 5 def 14 July 1969 inf 15 July 1969 ep 6 def 2 August 1969 inf 4 August 1969 ep 7 def 5 August 1969 inf 6 August 1969

End date

112

Appendix Table A1. Tilt and volume changes over eruptive cycles 1967 – 1976

MU Ia

ep 8 def inf Ep 9 def inf I erz def inf I erz def inf ep 10 def ep 11 def ep 12 def inf I ks inf inf I erz def inf I erz def inf inf def

MU pause inf inf def E ks inf E ks/swr I ks/sswr def inf inf MU IIa

2.33 3.54 6.19 4.63 0.14 0.80 0.55 1.02 2.58 1.14 4.43 1.34 4.63 4.71

109.36 299.15 114.35 282.61 116.36 296.57 153.39 324.46 122.15 297.90 107.95 102.80 292.62 113.01

0.0051 0.0077 0.0135 0.0101 0.0003 0.0017 0.0012 0.0022 0.0056 0.0025 0.0097 0.0029 0.0101 0.0103

4.34 0.59 1.22 1.14 2.16 1.88 1.96 4.70 3.95

286.70 315.00 284.04 98.97 328.50 124.70 305.13 272.90 111.16

0.0095 0.0013 0.0027 0.0025 0.0047 0.0041 0.0043 0.0102 0.0086

5.08 313.58 2.78 340.02 3.14 112.20

0.0111 0.0061 0.0069

6.55 315.00

0.0143

1.4620

0.00329 0.6451 Mauna Ulu eruption: period Ia cont. Episodic high fountaining 0.01129 0.8363

0.00376 0.6714 0.00988 1.019 0.01035 1.005 0.0792

0.6352

21.92 4.84 7.09 5.00

353.79 116.25 306.49 322.72

Mauna Ulu pause Eruption/intrusion elsewhere 0.008

Summit eruption/intrusion

0.0064

Summit/SW rift zone eruptionIvol probably too high (see text)

0.0478 0.0105 0.0155 0.0109

1.2211

0.100‡‡ 1.28 0.83 2.84 0.84 3.71 2.35

103.31 94.09 105.78 315.00 257.97 132.95

0.0028 0.0018 0.0062 0.0018 0.0081 0.0051

Mauna Ulu eruption: period Ib continuous eruption

Mauna Ulu eruption: period IIa continuous eruption Eruption rate (5 February 1972 – 4 May 1973) ¼ 0.082 km3 year21 113

E erz def def I ? erz inf trans def

23 August 1969 6 September 1969 7 September 1969 1 October 1969 3 October 1969 5 October 1969 7 October 1969 8 October 1969 13 October 1969 19 October 1969 20 October 1969 4 November 1969 30 December 1969 31 December 1969 18 June 1971 23 January 1970 24 January 1970 7 February 1970 15 February 1970 14 May 1970 16 May 1970 21 August 1970 26 November 1970 18 June 1971 5 February 1972 9 August 1971 13 August 1971 16 August 1971 14 August 1971 24 September 1971 29 September 1971 25 September 1971 29 September 1971 26 November 1971 5 February 1972 26 April 1973 4 May 1973 7 February 1972 18 March 1972 22 March 1972 18 August 1972 23 November 1972 26 April 1973

KI¯LAUEA VOLCANO MAGMA SUPPLY

MU Ib

21 August 1969 23 August 1969 6 September 1969 7 September 1969 1 October 1969 3 October 1969 5 October 1969 7 October 1969 8 October 1969 13 October 1969 19 October 1969 20 October 1969 4 November 1969 30 December 1969 31 December 1969 31 December 1969 23 January 1970 24 January 1970 7 February 1970 15 February 1970 14 May 1970 16 May 1970 21 August 1970 26 November 1970 18 June 1971 21 June 1971 9 August 1971 13 August 1971 14 August 1971 16 August 1971 24 September 1971 24 September 1971 25 September 1971 29 September 1971 26 November 1971 5 February 1972 5 February 1972 5 February 1972 7 February 1972 18 March 1972 22 March 1972 18 August 1972 23 November 1972

(Continued)

Appendix Table A1. Continued

MU IIb

Begin date

26 April 1973 26 April 1973 27 April 1973 E erz 5 May 1973 E erz def 4 May 1973 E erz mu 6 May 1973 8 June 1973 9 June 1973 3 July 1973 7 September 1973 E erz 10 November 1973 E erz mu 9 December 1973 10 November 1973 13 November 1973 23 May 1974 18 June 1974 trans 18 June 1974 E ks erz 19 July 1974 def 19 July 1974 inf 21 July 1974 E ks 19 September 1974 I ks 18 September 1974 28 September 1974 19 November 1974 29 November 1974 E swr 31 December 1974 I sswr 30 December 1974 3 January 1975 11 May 1975 22 May 1975 eq 29 November 1975

inf §§

End date 18 June 1974 27 April 1973 4 May 1973 5 May 1973 6 May 1973

Dt (years)

Tmag*

Taz†

Tvol‡

Evol§

1.1444

Eeff** 0.026

4.43 305.18 0.38 141.34

0.0097 0.0008

5.44 111.80

0.0119

5.40 1.46 2.51 1.35 6.00

0.0118 0.0032 0.0055 0.0030 0.0131

Mauna Ulu eruption: period IIb Continuous eruption

0.0008 Pauahi eruption; Mauna Ulu quiet 0.002

8 June 1973 9 June 1973 3 July 1973 7 September 1973 10 November 1973 9 December 1973 30 June 1974 13 November 1973 23 May 1974 18 June 1974 27 November 1975 19 July 1974 22 July 1974 21 July 1974 18 September 1974 19 September 1974 28 September 1974 19 November 1974 29 November 1974 30 December 1974 31 December 1974 3 January 1975 11 May 1975 22 May 1975 27 November 1975 29 November 1975

Comment

290.63 101.77 301.43 142.13 271.13

7.04 114.68 3.63 288.14 2.18 304.99

0.0154 0.0079 0.0047

2.33

19.36

0.0051

3.11 113.63 6.17 324.01

0.0068 0.0134

7.17 319.71 5.80 315.00 1.55 85.60 4.20 331.26

0.0156 0.0126 0.0034 0.0092

0.0024 Pauahi eruption; Mauna Ulu quiet 0.024 Mauna Ulu period IIb: resume continuous eruption

1.4428 0.0052 0.0055

0.0047 32.26 8.18 2.92 16.38

130.82 281.31 123.37 306.75

Mauna Ulu period IIb: resume continuous eruption

Post-Mauna Ulu Eruption/intrusion elsewhere 0.7647 Summit/East rift zone eruption Summit eruption/intrusion

0.0669 SW rift zone eruption/intrusion

0.0703 0.0178 0.0064 0.0357 M7.2 south flank earthquake

*Tilt magnitude in seconds Tilt azimuth Tilt volume in cubic kilometers § Eruption volume in cubic kilometers after correction of reported lava volumes for vesiculation (see text) **Eruption efficiency calculated as eruption volume (column 9) divided by volume of deflation (column 8) †† Episodes 24 –30 (Kinoshita et al. 1969, table 2 and fig. 2). ‡‡ The Honomu earthquake disrupted the Mauna Ulu plumbing, reducing the amount of lava erupted. This figure represents the amount of lava erupted before Mauna Ulu temporarily shut down when activity shifted to Pauahi in May. §§ Honumu earthquake changed the character of the eruption, beginning with inflation of Kilauea’s summit. Abbreviation: msr, magma supply rate in km3 year21 ¼ [Ev + Tiltv]/time in years. † ‡

T. L. WRIGHT & F. W. KLEIN

Post-MU

Event

114

Cycle

KI¯LAUEA VOLCANO MAGMA SUPPLY

References A KI , K. & K OYANAGI , R. 1981. Deep volcanic tremor and magma ascent mechanism under Kı¯lauea, Hawai‘i. Journal of Geophysical Research, 86, 7095–7109. A NDO , M. 1979. The Hawai‘i earthquake of November 29, l975: low dip angle faulting due to forceful injection of magma. Journal of Geophysical Research, 84, 7616– 7626. B ORGIA , A. 1994. Dynamic basis of volcanic spreading. Journal of Geophysical Research, 99, 17,791– 17,804. C AYOL , V., D IETERICH , J. H., O KAMURA , A. & M IKLIUS , A. 2000. High magma storage rates before the 1983 eruption of Kı¯lauea, Hawai‘i. Science, 288, 2343–2346. C ERVELLI , P. F. & M IKLIUS , A. 2003. The Shallow Magmatic System of Kı¯lauea Volcano. In: H ELIKER , C. C., S WANSON , D. A. & T AKAHASHI , T. J. (eds) The ¯ ‘o¯-Kupaianaha eruption of Kı¯lauea Volcano, Pu‘u ‘O Hawai‘i: the first 20 years. US Geological Survey Professional Papers, 1676, 149– 163. C LAGUE , D. A. & D ENLINGER , R. 1994. Role of olivine cumulates in destabilizing the flanks of Hawai‘ian volcanoes. Bulletin of Volcanology, 56, 425– 434. D ELANEY , P. T., D ENLINGER , R. P. ET AL . 1998. Volcanic spreading at Kı¯lauea, 1976– 1996. Journal of Geophysical Research, 103, 18,003– 18,023. D ELANEY , P. T., F ISKE , R. S., M IKLIUS , A., O KAMURA , A. T. & S AKO , M. K. 1990. Deep magma body beneath the summit and rift zones of Kı¯lauea Volcano, Hawai‘i. Science, 247, 1311– 1316. D ELANEY , P. T., M IKLIUS , A., A RNADOTTIR , T., O KAMURA , A. T. & S AKO , M. K. 1993. Motion of Kı¯lauea Volcano during sustained eruption from the ¯ ‘o¯ and Kupaianaha vents, 1983–1991. Pu‘u ‘O Journal of Geophysical Research, 98, 17,801– 17,820. D ENLINGER , R. 1997. A dynamical balance between magma supply and eruption rate at Kı¯lauea Volcano, Hawai‘i. Journal of Geophysical Research, 102, 18,091– 18,100. D IETERICH , J. 1994. A constitutive law for rate of earthquake production and its application to earthquake clustering. Journal of Geophysical Research, 99, 2601–2618. D IETERICH , J., C AYOL , V. & O KUBO , P. 2000. The use of earthquake rate changes as a stress meter at Kı¯lauea Volcano. Nature, 408, 457–460. D UFFIELD , W. A., C HRISTIANSEN , R. L., K OYANAGI , R. Y. & P ETERSON , D. W. 1982. Storage, migration, and eruption of magma at Kı¯lauea Volcano, Hawai‘i, 1971–1972. Journal of Volcanology and Geothermal Research, 13, 273–307. D VORAK , J. J. & D ZURISIN , D. 1993. Variations in magma supply rate at Kı¯lauea Volcano, Hawai‘i. Journal of Geophysical Research, 98, 22,225– 222,268. D VORAK , J. J. & O KAMURA , A. T. 1985. Variations in tilt rate and harmonic tremor amplitude during the January–August 1983 east rift eruptions of Kı¯lauea Volcano, Hawai‘i. Journal of Volcanology and Geothermal Research, 25, 249–258. D VORAK , J. J., O KAMURA , A. T. & D IETERICH , J. H. 1983. Analysis of surface deformation data, Kı¯lauea

115

Volcano, Hawai‘i, October 1966 to September 1970. Journal of Geophysical Research, 88, 9295–9304. D ZURISIN , D., K OYANAGI , R. Y. & E NGLISH , T. T. 1984. Magma supply and storage at Kı¯lauea Volcano, Hawai‘i, 1956–1983. Journal of Volcanology and Geothermal Research, 21, 177 –206. E ATON , J. 1959. A portable water-tube tiltmeter. Bulletin of the Seismological Society of America, 49, 301–316. E ATON , J. 1962, Crustal Structure and Volcanism in Hawai‘i. American Geophysical Union Geophysical Monographs, 6, 13–29. F ISKE , R. S. & K INOSHITA , W. T. 1969. Inflation of Kı¯lauea volcano prior to its l967– l968 eruption. Science, 165, 341– 349. F ISKE , R. S. & K OYANAGI , R. Y. 1968. The December l965 Eruption of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 607. G ERLACH , T. M. & G RAEBER , E. J. 1985. Volatile budget of Kı¯lauea Volcano. Nature, 313, 273–277. G ERLACH , T. M., M C G EE , K. A., E LIAS , T., S UTTON , A. J. & D OUKAS , M. 2002. Carbon dioxide emission rate of Kı¯lauea Volcano: implications for primary magma and the summit reservoir. Journal of Geophysical Research, 107, ECV 3 –1 to 3– 15, DOI:10.1029/ 2001JB000407. G ILLARD , D., R UBIN , A. M. & O KUBO , P. 1996. Highly concentrated seismicity caused by deformation of Kı¯lauea’s deep magma system. Nature, 384, 343– 346. G REENLAND , P., R OSE , W. I. & S TOKES , J. B. 1985. An estimate of gas emissions and magmatic gas content from Kı¯lauea Volcano. Geochimica et Cosmochimica Acta, 49, 125– 129. H ELIKER , C. C., S WANSON , D. A. & T AKAHASHI , T. J. ¯ ‘o¯-Kupaianaha eruption of (eds). 2003. The Pu‘u ‘O Kı¯lauea Volcano, Hawai‘i: the first 20 years. US Geological Survey Professional Papers, 1676. H ILL , D. & Z UCCA , J. J. 1987, Geophysical Constraints on the Structure of Kı¯lauea and Mauna Loa Volcanoes and some Implications for Seismomagmatic Processes. In: D ECKER , R. W., W RIGHT , T. L. & S TAUFFER , H. (eds) Volcanism in Hawai‘i. US Geological Survey Professional Papers, 1350, 903–917. J ACKSON , D. B., S WANSON , D. A., K OYANAGI , R. Y. & W RIGHT , T. L. 1975. The August and October 1968 East Rift Eruptions of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 890. J OHNSON , D. J. 1995. Molten core model for Hawai‘ian rift zones. Journal of Volcanology and Geothermal Research, 66, 27–35. K AUAHIKAUA , J., S HERROD , D. R. ET AL . 2003, Hawai‘ian ¯ ‘o¯-Kupaianaha Lava-flow Dynamics during the Pu‘u ‘O Eruption: a Tale of Two Decades. In: H ELIKER , C. C., S WANSON , D. A. & T AKAHASHI , T. J. (eds), ¯ ‘o¯-Kupaianaha eruption of Kı¯lauea The Pu‘u ‘O Volcano, Hawai‘i: the first 20 years. US Geological Survey Professional Papers, 1676, 63–87. K INOSHITA , W. T., K OYANAGI , R. Y., W RIGHT , T. L. & F ISKE , R. S. 1969. Kı¯lauea volcano: the l967– l968 summit eruption. Science, 166, 459–468. K LEIN , F. W. & K OYANAGI , R. Y. 1980. Hawai‘ian Volcano Observatory Seismic Network History 1950– 1979. US Geological Survey Open-File Reports, 80– 302.

116

T. L. WRIGHT & F. W. KLEIN

K LEIN , F. W., K OYANAGI , R. Y., N AKATA , J. S. & T ANIGAWA , W. R. 1987. The Seismicity of Kı¯lauea’s Magma System. In: D ECKER , R. W., W RIGHT , T. L. & S TAUFFER , H. (eds) Volcanism in Hawai‘i. US Geological Survey Professional Papers, 1350, 1019–1185. K LEIN , F. W., W RIGHT , T. & N AKATA , J. 2006. Aftershock decay, productivity, and stress rates in Hawai‘i: indicators of temperature and stress from magma sources. Journal of Geophysical Research, 111, B07307. K OYANAGI , R. Y., C HOUET , B. & A KI , K. 1987. Origin of Volcanic Tremor in Hawai‘i, part I. Data from the Hawai‘ian Volcano Observatory, 1969– 1985. In: D ECKER , R. W., W RIGHT , T. L. & S TAUFFER , H. (eds) Volcanism in Hawai‘i. US Geological Survey Professional Papers, 1350, 1221– 1257. K OYANAGI , R. Y., S WANSON , D. A. & E NDO , E. T. 1972. Distribution of Earthquakes Related to Mobility of the South Flank of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 800-D, D98– D97. K OYANAGI , R. Y., U NGER , J. D., E NDO , E. T. & O KAMURA , A. T. 1974. Shallow Earthquakes Associated with Inflation Episodes at the Summit of Kı¯lauea Volcano, Hawai‘i. International Association of Volcanology and Chemistry of the Earth’s Interior, 9– 14 September, 1974. L OCKWOOD , J. P., T ILLING , R. I. ET AL . 1999. Magma Migration and Resupply During the 1974 summit eruptions of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 1613. M ACDONALD , G. A. & E ATON , J. 1964. Hawai‘ian Volcanoes during l955. US Geological Survey Bulletins, 1171, 1–170; plates 171– 175 in pocket. M ORGAN , J. K. & M C G OVERN , P. J. 2005. Discrete element simulations of gravitational volcanic deformation: 1. Deformation structures and geometries. Journal of Geophysical Research, 110, DOI: 10.1029/2004JB003252. N IELSEN , N. N., F URUMOTO , A. S., L UM , W. & M ORRILL , B. J. 1977. Honomu, Hawai‘i earthquake of April 26, 1973. World Conference on Earthquake Engineering, 6th, Proceedings, 10–14 January 1977. O WEN , S., S EGALL , P. ET AL . 1995. Rapid deformation of the south flank of Kı¯lauea volcano, Hawai‘i. Science, 267, 1328– 1332. O WEN , S., S EGALL , P. ET AL . 2000. Rapid deformation of Kı¯lauea Volcano: global positioning system measurements between 1990 and 1996. Journal of Geophysical Research, 105, 18,983–18,998. P ARFITT , E. A. & W ILSON , L. 1994. The 1983– 86 ¯ ‘o¯ eruption of Kı¯lauea Volcano, Hawai‘i: a Pu‘u ‘O study of dike geometry and eruption mechanisms for

a long-lived eruption. Journal of Volcanology and Geothermal Research, 59, 179–205. R UBIN , A. M. & G ILLARD , D. 1998. Dike-induced earthquakes: theoretical considerations. Journal of Geophysical Research, 103, DOI: 10,017– 010,010,030. R YAN , M. 1988. The mechanics and three-dimensional internal structure of active magmatic systems: Kı¯lauea Volcano, Hawai‘i. Journal of Geophysical Research, 93, 4213– 4248. S HAW , H. R. 1980. The Fracture Mechanisms Transport from the Mantle to the Surface. In: H ARGRAVES , R. B. (ed.) Physics of Magmatic Processes. Princeton University Press, Princeton, NJ, 201 –264. S WANSON , D. A. 1972. Magma supply rate at Kı¯lauea Volcano, l952–1971. Science, 175, 169– 170. S WANSON , D. A., D UFFIELD , W. A. & F ISKE , R. S. 1976a. Displacement of the South Flank of Kı¯lauea Volcano: the Result of Forceful Intrusion of Magma into the Rift Zones. US Geological Survey Professional Papers, 963. S WANSON , D. A., J ACKSON , D. B., K OYANAGI , R.Y & W RIGHT , T. L. 1976b. The February l969 East Rift Eruption of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 891, 1 –30. S WANSON , D. A., D UFFIELD , W. A., J ACKSON , D. B. & P ETERSON , D. W. 1979. Chronological Narrative of the 1969–71 Mauna Ulu eruption of Kı¯lauea Volcano, Hawai‘i. US Geological Survey Professional Papers, 1056. T ILLING , R. I. & D VORAK , J. J. 1993. Anatomy of a basaltic volcano. Nature, 363, 125– 133. T ILLING , R. I., C HRISTIANSEN , R. L. ET AL . 1987. The 1972– 1974 Mauna Ulu Eruption, Kı¯lauea Volcano: an Example of Quasi-steady-state Magma Transfer. In: D ECKER , R. W., W RIGHT , T. L. & S TAUFFER , H. (eds) Volcanism in Hawai‘i. US Geological Survey Professional Papers, 1350, 405 –469. W RIGHT , T. L. & F ISKE , R. S. 1996. A supply-side view of Kı¯lauea Volcano, 1950–1975 [abstract]. EOS, Transactions of the AGU, 77 (suppl.), F798. W RIGHT , T. L. & K LEIN , F. W. 2006, Deep magma transport at Kı¯lauea volcano, Hawai‘i. In: E DWARDS , B. R. & R USSELL , J. K. (eds) Symposium on Mantle to Magma – Lithospheric and Volcanic Processes in Western North Australia. Lithos, 50–79. W YSS , M., K LEIN , F., N AGAMINE , K. & W IEMER , S. 2001. Anomalously high b-values in the south flank of Kı¯lauea Volcano, Hawai‘i: evidence for the distribution of magma below Kı¯lauea’s east rift zone. Journal of Volcanology and Geothermal Research, 106, 23–37.

Magnetic signatures associated with magma ascent and stagnation at Popocatepetl volcano, Mexico, during 2006 ANA LILLIAN MARTIN-DEL POZZO1, GERARDO CIFUENTES1, ´ LEZ1, ALICIA MARTINEZ2 & FABIOLA MENDIOLA1 EDUARDO GONZA 1

Instituto de Geofisica, Universidad Nacional Auto´noma de Me´xico, Circuito Institutos, Me´xico D.F., 04510, Mexico (e-mail: [email protected]) 2

Centro Nacional de Prevencio´n de Desastres, Delfı´n Madrigal 665, Me´xico D.F. 04360, Me´xico

Abstract: Monitoring of real-time magnetic signals at Popocatepetl during 2006 has allowed discrimination of magma injection and dome growth. Magnetic signals correlated with seismic, volcanotectonic events and harmonic tremor, as well as number of small emissions, spring water pH, ash components and dome evolution helped define upward magma transport and yield a better understanding of the volcanic plumbing system. Magma ascent occurs mostly in periods of 7 + 3 days associated with harmonic tremor and decreasing magnetic signals between 21.1 and 215 nT, followed by increasing signals linked to cooling of the domes and increased seismicity over periods of 1 to more than 3 months. The dome clogs the vent after the negative magnetic anomaly– harmonic tremor period associated with magma ascent and forces an explosive crater-reopening explosion. Larger negative changes in the magnetic signals occurred in April (26 nT), August (23 to 26 nT) and October to December (25 to 215 nT), associated with dome formation and growth. Negative magnetic anomalies preceded eruptions by 3 days in 2006.

Popocatepetl volcano, 5452 m a.s.l., is located in central Mexico between Mexico City and Puebla, which have populations of more than 20 million and 2 million, respectively. An additional population of several hundred thousand people lives in smaller towns and cities at the base of the volcano (Fig. 1). Popocateptetl has large Plinian events about every 1000 years and smaller eruptions that occur approximately each century. The volcano began erupting in December 1994 with small ash eruptions, which turned magmatic in 1996. Since then, intermittent dome formation and explosive clearing has filled most of the 0.5 km2 crater, which used to be about 300 m deep and now is only 50 m deep in some parts. Different magma batches (magnesian andesites to dacites) rise to the crater forming domes, which are later ejected explosively. More than 20 successive domes have been formed in the crater and explosively destroyed since 1996. Several relatively large eruptions (VEI 2 and 3) occurred on 30 April 1996, 11 May 1997, 30 June 1997, 25 November 1998, 17 December 1998 and 22 January 2001. The largest of these recent eruptions, in January 2001, was subplinian with an eruptive plume that reached 18 km and produced pumice and ash fall as well as pyroclastic flows and mudflows. Since then, only the small eruptions in 2002 and 2003 have produced scoraceous ejecta, while most of

the ongoing activity has been associated with dome growth and destruction. Even though the effects of the recent eruptions have been relatively small, the potential eruption hazard is large. Ash fall has caused the airport in Mexico City to be shut several times, and the towns on the volcano have been evacuated on two occasions, when the activity increased in 1994 and 2000. Recent studies at Popocatepetl volcano indicate that magnetic anomalies may be used to evaluate magmatic activity when correlated with other monitoring parameters such as seismicity. Continuous magnetic monitoring since 1997 has shown good correlation with magma processes governing the eruptive activity at Popocatepetl (Martin-Del Pozzo et al. 2002a, 2003). In this paper, we investigate the timing and magma emplacement mechanisms of the Popocatepetl magma in 2006, through the study of the magnetic field anomalies correlated with data from ejecta and spring water, crater dome morphology and seismic activity.

Methods Eruptive activity Activity at Popocatepetl is monitored continuously through direct observation and with a camera set

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 117–131. DOI: 10.1144/SP304.6 0305-8719/08/$15.00 # The Geological Society of London 2008.

118

A. L. MARTIN-DEL POZZO ET AL.

Fig. 1. Popocatepetl is located in central Mexico, 60 km from Mexico City and 40 km from Puebla, two densely populated cities. Many small towns are also located on the volcano. The magnetic station, springs and seismic stations are indicated.

up on the north side of the volcano. Plume height and distribution for the larger events are estimated. The Centro Nacional de Prevencion de Desastres (CENAPRED) reports the general activity on a daily basis. Periodic air photographs are also taken of the volcano by the Secretaria de Comunicaciones y Transportes to provide images of the crater area. Dome growth and destruction are monitored through the photographs.

Magnetism Continuous magnetic monitoring at Popocatepetl was set up in 1997 and, since then, the data have been correlated with other monitoring parameters as well as with the volcanic activity. The Popocatepetl Chipiquixtle magnetic station (CHX) is installed 3 km SW of the volcano’s crater on andesitic lava. It houses a Geometrics G-56 precession proton magnetometer that records the magnetic total field at a 60 s sampling rate with a resolution of 0.1 nT and an accuracy of 0.4 nT. The magnetometer is linked to the Universidad Nacional Autonoma de Mexico (UNAM) in Mexico City through a Free Wave 900 MHz radio-modem for remote unattended data download and real-time graphic display (Fig. 1). For magnetic reference we use the Overhauser POS1 magnetometer installed in the Teoloyucan Magnetic Observatory (TEO), which is sampled every 30 s and has a resolution of 1 pT and an accuracy of 17 pT. The

reference station is located on dacitic bedrock, 91 km to the NW of Popocatepetl. Ionospheric conditions are similar at stations near each other (Wienert 1970), as is the case for CHX and TEO, since they are relatively close. We also assume that the volcanomagnetic effects are restricted to the proximal area of the volcano. In order to investigate the magnetic behaviour of Popocatepel, the CHX and TEO magnetic records were analysed as described by Martin-Del Pozzo et al. (2002a). Spikes were eliminated routinely. The hourly mean values from each station were used to calculate a 24 h (UTC) period. Days with less than 6 h of reliable signals during the night were eliminated. Results were then processed by the weighted differences method (Rikitake 1968) to minimize the effects of non-volcanogenic external sources. In this method the value of the reference station is multiplied by a, the weighted factor, and subtracted from the value at CHX to obtain the weighted difference in the field. The weighted differences using hourly mean values and daily mean values show the same linear behaviour (Fig. 2a). Note that magnetic signals from both stations present similar morphology and, when subtracted, diurnal variations are practically eliminated (Fig. 2b). The weighted difference is not an absolute value since it is relative to the previous values and should be considered in terms of its peak-to-peak amplitude and shape.

MAGNETIC SIGNATURES AT POPOCATEPETL

119

Fig. 2. (a) Continuous plot of the weighted differences using hourly main values and daily mean values (triangles). (b) Total magnetic field intensity in Teoloyucan (thick line) and Chiquipixtle (thin line). The scale on the left is for Chiquipixtle and the one on the right is for Teoloyucan.

Seismicity The seismic network operated by CENAPRED jointly with the Instituto de Geofisica consists of Eight Mark L-4 short-period 1 Hz stations, which operate on the flanks of the volcano between 3200 and 4200 m a.s.l., 2–10 km from the crater (Fig. 1). In order to better understand the magnetic behaviour as well as the evolution of the volcanic activity, seismic and magnetic data were analysed together.

Spring water and ash Spring water monitoring at Popocatepetl is carried out each month since 1991 (Martin-Del Pozzo et al. 2002b, c). Several springs show varying pH, HCO3, SO4, F and B concentrations related to the magmatic activity. Data (pH) from one of the representative springs on the southeastern flank of the volcano was also correlated with the other monitoring data to give additional insight into the magmatic processes at Popocatepetl. Ash has also been sampled after each eruption since 1994 for component, textural and chemical analyses (Martin-Del Pozzo et al. 1995, 2007).

A description of the ash sampled during 2006 is included here to correlate the geophysical parameters with the presence of juvenile magma as well as with products of the destruction of the domes.

Observations Volcanic activity Eruptions discussed in this paper began in December 2005 and extended through 2006. There were 13 small explosions from January to July 2006, which produced ash plumes mostly between 1 and 3.5 km above the crater (6.5–9 km a.s.l.). The eruption dates and types of deposits are reported in Table 1, as well as the plume height and direction. For the 4 April and 12, 13, 14 and 16 June explosions, the plume height could not be estimated because of low visibility. Besides these eruptions, small gas emissions were detected each day on the seismograms and confirmed visually. December and January were marked by a series of small eruptions, which produced ash and ballistics and destroyed the 2005 dome. Small eruptions that occurred on 19 and 26 January and 3 February

120

A. L. MARTIN-DEL POZZO ET AL.

Table 1. Eruptive Activity at Popocatepetl volcano Eruption date 1 December 2005 4 December 2005 13 December 2005 18 December 2005 25 December 2005 6 January 2006 19 January 2006 25 January 2006 26 January 2006 3 February 2006 5 February 2006 10 February 2006 22 March 2006 4 April 2006 23 May 2006 12 June 2006 13 June 2007 14 June 2008 16 June 2008 25 July 2006 16 August 2006 1 –5 November 2006 24 November 2006

Column height (km a.s.l.)

Plume direction

Dome growth

10.5, 8.0 8.5 8.0, 6.5 8.0 8.5 7.5 6.5

NE NE SE NE NE SE SE



 

8.5 6.5 6.7

NE SE SE

7.0 — 7.5 — — — — 9.0

NE NE SE SW SW SW — NW

Ejecta Ash Ash and ballistics Ash and ballistics Ash Ash and ballistics Ash and ballistics Incandescent scoria Ash Ash Ash and ballistics



were associated with the growth of a new dome, which was later partially cleared away by the 22 March eruption. Two other larger eruptions on 23 May and 25 July produced plumes which reached 7.5 and 9 km a.s.l. Another new dome was confirmed on 16 August and during the rest of the year only minor gas emissions occurred. Samples collected near the volcano after the February and July eruptions showed that ejecta were made up mostly of black andesitic dome lithics, some plagioclase, pyroxene and a few hornblende crystals covered with a fine white silicic dust (which accompanies many of the dome building eruptions). Ejecta are andesitic with an average composition of approximately 61% SiO2. Seventyfive per cent of the ash particles were between 0.12 and 0.06 mm in size.

Seismicity Seismic activity was characterized by low-intensity long-period events (LP) accompanied by gas emissions (steam and SO2 –CO2), occasionally with some ash. These low-frequency earthquakes show no distinct S wave arrival since they are shallow, generally very small, events. They have been associated with bubbling magma, gas separating from the magma and movement of magma and magmatic fluids (Malone 1983; Lahr et al. 1994). About 15 LPs per day were recorded at Popocatepetl during 2006, except for several days in May

Ash Ash Ash Ash Ash Ash Ash Ash   

and July when over 55 events were detected. Episodes of low-amplitude harmonic tremor lasting tens of minutes occurred sporadically throughout 2006. The occurrence of harmonic tremor is thought to be produced by internal movements of fluids, in this case related to ascending magma. Volcano-tectonic earthquakes (VT) are characterized by high-frequency impulsive arrivals and originate between 1 and 10 km depth. Their seismic motion is similar to that of the shallow rock-fracturing tectonic earthquakes. VT earthquakes may reflect the stress conditions in the deeper part of the volcano because they are generated by rock fracturing in response to the intrusion and migration of magma or the expansion of geothermal fluids at high pressures (Minakami 1960, 1974). At Popocatepetl, VTs have been associated with stress redistribution accompanying magma movement through the Popocatepetl plumbing system. During 2006, 140 VTs with magnitudes between 1.9 and 3.0 and depths of 2–10 km were recorded. The largest of these VTs occurred on 3 and 12 April. The magnitude 2.8 VT was 5.5 km deep while the magnitude 3 VTs on 8 June and 12 December were 4.7 and 5.5 km deep, respectively. The 8 June event displayed compression components at all seismic stations, which could reflect magma pushing up under the crater to a depth of 4.7 km. VTs more than 4.5 km deep have predominantly reverse mechanisms, probably due to an ascending towing traction caused by intruding

MAGNETIC SIGNATURES AT POPOCATEPETL

121

Fig. 3. Magnetic signals, periods of harmonic tremor, number of small emissions, number of volcano tectonic events, spring water pH, eruptions and domes for (a) the first semester of 2006 (including December of the previous year) and (b) the second semester of 2006. Daily cumulative K index (geomagnetic 3 h activity index) is also included. Decreasing magnetic signals are indicated in the lower part of the anomaly while increasing changes are indicated in the upper part. Bigger arrows show rising magma batches that are larger. Direct observation of the domes is indicated in black, while the rest is inferred from monitoring.

122

A. L. MARTIN-DEL POZZO ET AL.

magma through the conduit, as well as compression in the vicinity of the feeders (Ara´mbula 2006). Two deeper VTs (8.6 and 9.3 km deep) with magnitudes of 2.4 and 2.5 were located 8 km to the SW of the crater on 28 and 31 May. In this area, mechanisms correspond to a SE –NW trending normal fault with a transcurrent component, which may damp compression when the volume under the volcano increases due to rising magma (Ara´mbula 2006). This type of event is infrequent and had not occurred between the end of 2002 and May 2006. Seismicity increased from July onwards (Fig. 3b). Two important episodes of large-amplitude harmonic tremor were recorded in August and November and lasted 16 and 17 h, respectively. Seismicity increased slightly again in October, with the monthly average changing from 9 VTs to 20.

Spring water Small decreases in the pH at several of the cold springs monitored on the flanks of the volcano were detected in April, May, August and November (Fig. 3). Since the volcano has a large volume of water and rising magma batches are small, only small changes in the chemical concentration of magma-related components (F2, SO22 4 , pCO2, B) occur (Martin-Del Pozzo et al. 2002b, c). When the pH at the spring shown in Figure 3a & b decreases to below 6, new magma and hot gases are rising.

Magnetic signals Weighted differences are plotted in Figure 3, where they are compared with seismicity and spring water pH. Magnetic anomalies represent the difference in magnetic signal of the volcano between different measuring times since they are relative changes in the weighted differences. They are numbered sequentially in Table 2 and Figure 3 for easy text reference. Changes in the magnetic signal at active volcanoes may be due to thermomagnetic (demagnetization or remagnetization), piezomagnetic (stress-induced) or electrokinetic effects. Martin-Del Pozzo et al. (2002a) classified the magnetic signals at Popocatepetl volcano based on their morphology. Steeply increasing signals were closely correlated with seismicity and were stress-induced, piezomagnetic effects. Small positive and negative magnetic anomalies were associated with the ascent of small magma batches depending on the Curie temperature and stress induced by magma ascent. These magnetic signals (0.5– 7 nT) preceded eruptions in 1997–1999 by 1– 8 days. A stepwise increase in the signal which did not return to baseline was attributed to cooling and oxidation of the new lava dome in the crater. Magnetic signals (23 to 25 nT) also decreased before the

2000–2001 eruptions at Popocatepetl and were attributed to ascent of several magma batches (Martin-Del Pozzo et al. 2003). These changes in magnetic signal preceded increases in seismicity and dome growth by several days. The December 2005 and January 2006 eruptions were accompanied by 3 days of harmonic tremor and several negative magnetic anomalies. Between 13 January and 2 February 2006, a 26.5 nT anomaly (9), was detected before observation of a new crater dome, which preceded the harmonic tremor and decreasing pH in local springs (Tables 1 and 2, Fig. 3a). This activity was associated with magma ascent and dome emplacement, which was confirmed by aerial photographs on 10 February. Small eruptions occurred on 19 and 26 January and 3 February, associated with the growth of the new dome. Negative magnetic anomalies (arrows) were precursors to these eruptions (Fig. 3a). An increase in magnetic signal (18) followed the small eruption on 22 March and the associated seismicity, and destroyed part of the dome observed in February. Continued increase (anomalies 18, 20, 21 and 22) lasted 1 month, although the individual peaks correlate with the small emissions (Table 2, Fig. 3a). Two periods of harmonic tremor and magnitude 2.8 VTs at 5.5 km depth, as well as a slight decrease in the spring water pH, suggest further movement of magma. A decrease in the magnetic signal (23), 25.8 nT, from 15 to 26 April, as well as harmonic tremor and decreasing spring water pH, were related to another ascending magma batch (Fig. 3a). The anomaly lasted 10 days (feeding). There was a period of harmonic tremor during the first week of May and an increase in LPs, which generally produced opposite magnetic signals and could have resulted in a superimposed damped magnetic signal. Nevertheless, after the 23 May eruption, a decrease in the magnetic signal was observed (29). Deep VTs, harmonic tremor, and decreasing pH in the spring water were associated with a period of ascending magma and possible opening of cracks, which released the confining pressure. After magma injection, cooling, partial sealing and gas pressure build-up produced a month-long increase in the magnetic signal. The 25 July eruption (3.5 km high ash plume) cleared away most of the dome. Harmonic tremor occurred again in July and on 3 August. This large-amplitude tremor, which lasted 16 h, as well as the decreasing spring water pH, correlate with the 29.6 nT (39) decrease in the magnetic signal associated with magma ascent (Table 2, Fig. 3b). A close-up of anomalies 38 and 39 using hourly mean values shows the same tendency as the daily mean values (Fig. 4).

MAGNETIC SIGNATURES AT POPOCATEPETL

123

Table 2. Magnetic anomalies (MA) indicating weighted differences (WD) and their changes relative to each other and the previous WD (DWD) MA 1

2 3

4

5

6

7

8

9

Date

WD (nT)

04 December 2005 05 December 2005 06 December 2005 07 December 2005 08 December 2005 09 December 2005 09 December 2005 10 December 2005 11 December 2005 11 December 2005 12 December 2005 13 December 2005 14 December 2005 14 December 2005 15 December 2005 16 December 2005 17 December 2005 18 December 2005 19 December 2005 20 December 2005 21 December 2005 22 December 2005 22 December 2005 23 December 2005 24 December 2005 25 December 2005 26 December 2005 27 December 2005 28 December 2005 29 December 2005 30 December 2005 31 December 2005 01 January 2006 02 January 2006 03 January 2006 04 January 2006 05 January 2006 06 January 2006 06 January 2006 07 January 2006 08 January 2006 09 January 2006 09 January 2006 10 January 2006 11 January 2006 12 January 2006 13 January 2006 13 January 2006 14 January 2006 15 January 2006 16 January 2006 17 January 2006 18 January 2006 19 January 2006 20 January 2006 21 January 2006

0.3 20.4 0.1 20.5 20.7 20.8 20.8 20.1 0.5 0.5 0.1 20.5 21.1 21.1 20.9 20.4 20.2 20.3 20.1 20.4 1.9 2.3 2.3 1.5 1.2 0.6 0.4 0.4 1.0 0.8 0.3

DWD (nT)

21.1 1.2

21.6

10

11 3.4

12

21.9

3.5 3.5 2.1

3.2

1.4 1.4 1.9

22.1

4.4 4.4

3.0

0.9

MA

13

14

15

Date 22 January 2006 23 January 2006 24 January 2006 25 January 2006 26 January 2006 27 January 2006 28 January 2006 29 January 2006 30 January 2006 31 January 2006 01 February 2006 02 February 2006 02 February 2006 03 February 2006 04 February 2006 05 February 2006 06 February 2006 07 February 2006 08 February 2006 09 February 2006 09 February 2006 10 February 2006 11 February 2006 12 February 2006 13 February 2006 13 February 2006 14 February 2006 15 February 2006 16 February 2006 17 February 2006 18 February 2006 18 February 2006 19 February 2006 20 February 2006 21 February 2006 22 February 2006 23 February 2006 23 February 2006 24 February 2006 25 February 2006 26 February 2006 27 February 2006 28 February 2006 01 March 2006 02 March 2006 03 March 2006 03 March 2006 04 March 2006 05 March 2006 06 March 2006 07 March 2006 08 March 2006 09 March 2006 10 March 2006 11 March 2006

WD (nT)

DWD (nT)

21.0 22.1 22.1

26.5

20.5 1.4 20.1 0.6 2.7 2.7 1.9 0.1 22.1 22.4 22.4 0.0 0.4 1.9 0.9 2.5 2.5 0.8 0.1 21.7 21.7 20.4 0.0 20.6 20.1 21.1 0.7 0.3 0.9 0.9 0.1 0.5 20.1 0.1 20.8 20.8 0.5 21.3

4.8

25.1

4.9

24.1

2.5

22.2

(Continued)

124

A. L. MARTIN-DEL POZZO ET AL.

Table 2. Continued MA 16

17 18

19

20 21

22

23

24 25

11 12 13 14 15 15 16 17 17 18 19 20 20 21 22 23 24 25 26 27 27 28 29 30 31 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 15 16 17 18 19 20 21 22 23 24 25 26 26 27 27 28 29 30

Date

WD (nT)

March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 March 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006 April 2006

21.3 20.7 21.1 20.1 0.4 0.4 21.3 21.4 21.4 21.0 1.0 3.3 3.3 0.5 2.4 0.8 1.0 0.8 2.7 20.1 20.1 1.0 3.8 1.2 1.4 1.3 1.9 1.5 2.9 20.7 2.6 2.9 0.8 2.0 2.0 2.2 3.7 3.3 3.1 4.5 4.5 3.3 1.0 0.5 0.5 0.8 0.3 0.4 0.2 0.8 20.4 21.3 21.3 0.8 0.8 0.4 20.1 20.1

DWD (nT)

1.7 21.7

MA

26

27

4.7

28 23.4 3.9

29

30 4.1 31

32 5.4

33

25.8 2.1

Date 01 02 03 04 05 06 07 08 08 09 10 11 12 13 14 15 16 16 17 18 19 20 20 21 22 23 24 25 26 26 27 28 29 30 30 31 01 02 02 03 04 05 06 07 08 09 09 10 11 12 13 14 15 16 17 18 19

May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 May 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006 June 2006

WD (nT) 0.4 20.3 0.2 20.6 0.5 20.6 20.3 1.4 1.4 0.7 20.5 0.4 0.5 0.1 0.6 20.6 21.7 21.7 20.6 0.9 1.5 1.9 1.9 20.8 1.0 20.5 0.2 21.2 21.2 21.2 0.3 0.3 0.4 1.3 1.3 0.7 0.1 20.3 20.3 1.0 0.9 0.0 1.1 1.1 1.4 2.4 2.4 1.9 1.9

DWD (nT)

21.4

1.9

23.0

3.6

23.1

2.5

21.6

2.7

(Continued)

MAGNETIC SIGNATURES AT POPOCATEPETL

125

Table 2. Continued MA

34

35

36

37

38

39

40

Date 20 June 2006 21 June 2006 22 June 2006 23 June 2006 24 June 2006 25 June 2006 26 June 2006 27 June 2006 28 June 2006 29 June 2006 30 June 2006 01 July 2006 02 July 2006 03 July 2006 04 July 2006 05 July 2006 06 July 2006 06 July 2006 07 July 2006 08 July 2006 09 July 2006 10 July 2006 11 July 2006 11 July 2006 12 July 2006 13 July 2006 14 July 2006 15 July 2006 16 July 2006 17 July 2006 18 July 2006 19 July 2006 20 July 2006 21 July 2006 22 July 2006 23 July 2006 24 July 2006 24 July 2006 25 July 2006 26 July 2006 27 July 2006 27 July 2006 28 July 2006 29 July 2006 30 July 2006 31 July 2006 01 August 2006 01 August 2006 02 August 2006 03 August 2006 04 August 2006 04 August 2006 05 August 2006 06 August 2006 07 August 2006

WD (nT)

DWD (nT)

MA

Date

WD (nT)

41

07 August 2006 08 August 2006 09 August 2006 10 August 2006 11 August 2006 11 August 2006 12 August 2006 13 August 2006 14 August 2006 14 August 2006 15 August 2006 16 August 2006 17 August 2006 17 August 2006 18 August 2006 19 August 2006 20 August 2006 20 August 2006 21 August 2006 22 August 2006 23 August 2006 24 August 2006 25 August 2006 26 August 2006 26 August 2006 27 August 2006 28 August 2006 28 August 2006 29 August 2006 30 August 2006 30 August 2006 31 August 2006 01 September 2006 02 September 2006 03 September 2006 04 September 2006 05 September 2006 06 September 2006 06 September 2006 07 September 2006 08 September 2006 09 September 2006 10 September 2006 10 September 2006 11 September 2006 12 September 2006 13 September 2006 13 September 2006 14 September 2006 15 September 2006 16 September 2006 16 September 2006 17 September 2006 18 September 2006

22.6 24.1 23.8 26.2 28.3 28.3

42

43

44 21.3 21.3 0.6 0.5 20.4 20.1 0.5 0.5 0.2 0.0 0.5 20.7 1.1 20.4 0.2 0.0 20.3 0.5 1.9 1.9 1.8 1.4 20.7 20.7 20.5 1.5 1.3 0.7 2.2 2.2 2.2 0.1 27.4 27.4 26.8 26.0 22.6

23.7

45

1.9 46 21.2

47 48

2.2 49 22.6 50 2.9 51 29.6

52

24.8 24.4 24.4 27.3 25.3 27.5 27.5 25.4 25.1 20.2 20.2 21.9 23.3 22.6 23.1 22.9 26.4 26.4 23.9 22.1 22.1 24.3 25.0 25.0 24.2 22.3 22.5 24.0 20.7 21.4 1.8 1.8 21.9 22.5 22.5 23.7 23.7 23.0 23.3 20.3 20.3 22.3 23.4 24.2 24.2 22.1 1.6

DWD (nT)

25.7

3.9

23.2

7.3

26.2 4.3 22.9

6.8

25.5

3.4

23.9 5.8

4.8 (Continued)

126

A. L. MARTIN-DEL POZZO ET AL.

Table 2. Continued MA 53

54 55

56 57

58 59 60

61

62 63

64 65

Date

WD (nT)

18 September 2006 19 September 2006 20 September 2006 21 September 2006 22 September 2006 23 September 2006 23 September 2006 24 September 2006 24 September 2006 25 September 2006 26 September 2006 27 September 2006 28 September 2006 29 September 2006 29 September 2006 30 September 2006 01 October 2006 01 October 2006 02 October 2006 03 October 2006 04 October 2006 05 October 2006 06 October 2006 07 October 2006 07 October 2006 08 October 2006 09 October 2006 09 October 2006 10 October 2006 11 October 2006 11 October 2006 12 October 2006 13 October 2006 14 October 2006 15 October 2006 16 October 2006 16 October 2006 16 October 2006 17 October 2006 18 October 2006 19 October 2006 20 October 2006 20 October 2006 21 October 2006 22 October 2006 22 October 2006 23 October 2006 24 October 2006 25 October 2006 26 October 2006 27 October 2006 27 October 2006 28 October 2006 29 October 2006 29 October 2006 30 October 2006 31 October 2006

1.6 0.7 24.3 0.1 22.2 25.3 25.3 2.9 2.9 0.8 1.1 1.6 20.9 21.7 21.7 21.5 6.9 6.9 0.8 2.9 0.7 2.5 20.9 22.2 22.2 2.1 3.6 3.6 0.3 21.3 21.3 20.7 21.0 2.2 4.9 6.5 6.5 6.5 4.8 2.4 1.3 21.5 21.5 6.1 7.5 7.5 4.8 2.5 4.2 3.2 20.5 20.5 3.9 12.1 12.1 7.3 8.0

DWD (nT)

MA

66 26.9 8.2 67 68 24.6 8.6

69 70

29.1

71

5.8 72 25.0

73 74

7.8

28.0

75

9.0 76

27.9 12.6

Date

WD (nT)

01 November 2006 02 November 2006 03 November 2006 04 November 2006 04 November 2006 05 November 2006 06 November 2006 07 November 2006 07 November 2006 08 November 2006 09 November 2006 09 November 2006 10 November 2006 11 November 2006 12 November 2006 13 November 2006 13 November 2006 14 November 2006 15 November 2006 15 November 2006 16 November 2006 17 November 2006 18 November 2006 18 November 2006 19 November 2006 20 November 2006 21 November 2006 21 November 2006 22 November 2006 23 November 2006 23 November 2006 24 November 2006 25 November 2006 25 November 2006 26 November 2006 27 November 2006 28 November 2006 29 November 2006 30 November 2006 01 December 2006 02 December 2006 02 December 2006 03 December 2006 04 December 2006 05 December 2006 05 December 2006 06 December 2006 07 December 2006 08 December 2006 09 December 2006 10 December 2006 11 December 2006 12 December 2006 13 December 2006 14 December 2006

4.2 1.9 4.4 20.3 20.3 5.7 5.2 8.3 8.3 0.8 20.7 20.7 6.1 9.4 12.1 14.4 14.4 7.3 2.9 2.9 9.1 7.7 10.1 10.1 4.7 5.1 2.4 2.4 3.2 9.8 9.8 0.3 22.2 22.2 3.1 0.9 0.3 1.8 10.8 10.8 1.3 3.2 24.2 24.2 0.6

DWD (nT)

212.4

8.5 29.0

15.1 211.6

7.3

27.7 7.3 212.0

13.0

215.0

5.8 5.0 6.6 5.8 13.6 7.8

17.9

MAGNETIC SIGNATURES AT POPOCATEPETL

127

Fig. 4. Close-up of anomalies 38 and 39. Hourly mean magnetic intensity values in nT with the daily mean magnetic intensity marked as dots.

The new dome was confirmed visually on 16 August. Besides this anomaly, other large changes in the magnetic signal occurred in August (25.7, 23.2, 26.2 and 22.9 nT), September (25.5, 23.9, 26.9 and 24.6 nT) and grew larger from October (29.1, 25, 28 and 27.9 nT) through to December (212.4, 29.0, 211.6, 27.7, 212 and 215 nT). This dome seems to be cooling (dashed outlined area with positive slope), although it is fed by new small magma batches (Fig. 3b).

Discussion Magnetic anomalies Our interpretation of the 2006 magnetic results, besides the correlation with eruptive activity and other geophysical and geochemical studies, is based on previous multiparameter monitoring at Popocatepetl. Several magnetic studies on active volcanoes also registered relative changes in the signal associated with magmatic activity. Small short-term precursory signals, 21.3 nT, occurred at Merapi Volcano before the 1992 eruption, where rapid signals were associated with changes in the stress field (Zlotnicki & Bof 1998; Zlotnicki et al. 2000). A 20 nT change at Unzen Volcano

preceded the major volcanic activity in 1990– 1992, but increased to more than 80 nT during dome extrusion (Tanaka 1995). Magnetic anomalies at another dacitic dome, Mount St Helens, were mainly thermomagnetic, associated with a hot non-magnetized core surrounded by a cool magnetized carapace (Dzurisin et al. 1990). The authors also found that precipitation had a significant effect on the heat lost from the dacitic dome during quiet periods. Long-term variations in the magnetic signal at Piton de la Fournaise Volcano in Reunion are related to a major fracture zone, while rapid magnetic signals appear before pre-eruptive seismic swarms and are caused by electrokinetic effects (Zlotnicki et al. 1993). The largest anomalies were recorded during intrusive crises and opening of eruptive fissures. Rapid changes of 4–5 nT were also related to seismic swarms at Etna Volcano, and step-like variations (9–10 nT) were caused by the opening of an eruptive fissure (Del Negro et al. 2004; Currenti et al. 2005). The authors link them to stress redistribution due to magmatic intrusion. Magnetic signals associated with dyke intrusion at Miyakejima in 2000 exceeded 200 nT, where crustal deformation data suggest that the effect was mainly piezomagnetic due to propagation of the dyke (Ueda et al. 2006).

128

A. L. MARTIN-DEL POZZO ET AL.

At Popocatepetl, small negative (21.1 to 2 2.1 nT) and positive (1.2 to 3.4 nT) anomalies occurred from the end of the previous year to January 2006, but increased in size after midJanuary (24.1 to 26.5 nT). Small increases in the magnetic signal (anomalies 2, 4, 6 and 8) mimic the emissions and are stress-related, piezomagnetic effects (Fig. 3a). Larger negative anomalies also occurred in April associated with prolonged harmonic tremor. This increase in activity caused the hazard level designated by Civil Protection for the volcano to increase. Large magnetic anomalies from August onwards (39, 61, 65, 69, 73 and 75) were associated with harmonic tremor and decreasing spring water pH, with visible confirmation of dome growth on 10 August and 24 November (Fig. 3b). Longer magnetic signals such as this increasing magnetic trend (40 –68) are related to cooling and contraction of the crater dome, which is a thermomagnetic effect as the dome cools under the Curie point. This trend is disrupted by rising small batches of magma and newly formed cracks in the dome, which also released pressure and produced the sharper piezomagnetic peaks. The larger negative magnetic anomaly, which occurred in the last week of November, divided the ascending trend in 2, confirming the cooling and cracking of the same dome. Negative magnetic anomalies occur with rising magma batches (above the Curie Point) accompanied by hot gases, which may fracture the pre-existing dome, forming a new one which begins to cool, producing positive trends by magnetization and pressure build-up. Short, sharp positive anomalies correlate with the number of emissions and are mostly stress-related, piezomagnetic anomalies. Magma moves up by several small magma injections, producing the negative anomalies, followed by an increase in seismicity and magnetic signal associated with lava extrusion forming the new dome. The negative anomalies in January, April, May, August, October and November 2006 show ascent of magma batches followed by cooling of the dome and ascent of other magma batches (thermomagnetic and piezomagnetic effects).

Popocatepetl magmatic plumbing Depth constrained by mineral geobarometry indicates magma rises from depths of 4–13 km, which correlates with the depth of the VTs, although most of them occur around 5 + 0.5 km. Larger eruptions are associated with influx of deeper magma. High-Mg# clinopyroxene geobarometry indicates crystallization occurred between 10 and 40 km, while low-Mg# clinopyroxene

crystallized between 9 and 13 km depth and plagioclase at about 4 km depth and possibly shallower during ascent (Straub & Martin-Del Pozzo 2001). Different batches of ascending mafic magma mix with the more evolved dacitic magma at shallow depths, 4 –13 km under the volcano. Arrows in Figure 3 denote incoming magma. The stagnation level appears to be a series of interconnected dykes, not a large magma chamber, since during the 10 years of monitoring, no timeprogressive compositional trend in the ejecta has been observed, implying a steady influx and outflux of melts through the Popocatepetl plumbing system in a short timescale, coincident with the decreasing magnetic signals. The deep (10 km) earthquakes located SE of the crater are associated with the normal fault recently proposed by Lermo-Samaniego et al. (2006) and are perpendicular to the vent lineation mapped by Espinasa and Martin-Del Pozzo (2005). These earthquakes may be related to magma movement through a dyke along this SE – NW trending axis, part of a more complex feeder dyke system. Negative magnetic anomalies are related to magma movement, which produces both thermomagnetic and piezomagnetic effects.

Dome evolution Cumulative seismic energy increased in July (related to the eruption on 23 July and an increased number of small emissions), August (16 h of harmonic tremor on 3 August and another increase in number of small emissions), and more steeply in October to December (Fig. 5). This last increase associated with the higher number of VTs corresponds with the increase in magnetic signal and periods of harmonic tremor linked to magma movement and dome growth, as well as dome contraction observed at the end of the year. This change in behaviour was probably due to a more viscous nature of the new dome. The evolution of the domes is shown in Figure 6. The March, May and July eruptions destroyed part of the dome that was observed in February (Fig. 6a), so the August photograph is of a different dome (Fig. 6b), which grew to fill the inner pit again (Fig. 6c). No dome-clearing eruptions occurred after July, so the dome grew until at least November, as suggested by periods of harmonic tremor, increasing spring-water pH and decreasing magnetic anomalies, and then contracted (Fig. 6d). Outlined areas in Figure 3 showing positive anomalies are related to cooling of the domes augmented by piezomagnetic effects caused by increasing seismicity. Larger anomalies in the magnetic signal probably result from this summed effect.

MAGNETIC SIGNATURES AT POPOCATEPETL

129

Fig. 5. Cumulative seismic energy of real-time seismic amplitude measurements (RSAM) for 2006.

Fig. 6. Aerial photographs of the crater showing dome evolution during 2006. The triangular feature on the left is the Pico Mayor, the highest part of the volcano. Because of its height, Popocatepetl is generally covered by snow. (a) February 10, small dome growing in the central pit left by the 2001 and 2003 eruptions in the previous crater domes. Note there are two higher ledges in the earlier domes, indicating their larger volume. (b) August 16, dome growth. Note steep front of the dome. (c) October 10, dome grew, filling the inner pit. (d) Dome seen on 24 January 2007. Note contraction in the central part. Aerial photographs were taken by the Secretaria de Comunicaciones y Transportes by request for the Scientific Committee.

130

A. L. MARTIN-DEL POZZO ET AL.

Conclusions At Popocateptl several magnetic signatures are superimposed. Piezomagnetic and thermal magnetization as well as crystallization of the ascending magma in the new dome occurs concurrently. These processes can be distinguished on basis of signal morphology, and corroborated when correlated with other monitoring techniques. The magnetic behaviour at Popocatepetl is complex since it reflects a series of summed magnetic effects. Piezomagnetic changes are associated with stress redistribution, which accompanies magma ascent as well as thermomagnetic effects. Short, sharp magnetic increases correlate with seismicity, especially with number of long period events, and are of piezomagnetic origin, while longer increases lasting more than 3 weeks are attributed to thermomagnetic cooling effects in the domes. Magnetic anomalies show repeated periods of magma injection lasting 5–7 days, associated with decreasing magnetic signals between 21.1 and 215 nT. Magma injection appears to be related to a NW-trending dyke (10 km deep) that reaches stagnation level at 5 + 0.5 km under the crater and then ascends to form a dome. The dome clogs the vent after the negative magnetic anomaly–harmonic tremor period associated with magma ascent, and forces an explosive crater-reopening explosion. The periods of harmonic tremor accompanied by higher emission of SO2, as well as the magnetic behaviour, suggested that an eruption would follow at the end of 2006. However only minor ash emission and extrusive dome growth has occurred, suggesting that part of the dome may be emplaced as a cryptodome. The steep fronts seen in the lava dome morphology suggest that this crater dome is more viscous than some of the previous, more fluid domes. It had a slower effusion rate than the more fluid magma that formed the January 1997 and December 2000 domes. Temperature conditions in the remnants of the partially ejected dome seem to be reset by ascent of new magma batches. Solidification rate is perturbed by increased heat from the new incoming batch. DGAPA-PAPIIT and the Instituto de Geofisica (Universidad Nacional Auto´noma de Me´xico) funded the research. The authors also wish to thank J. Urrutia, H. Rymer and T. Wright for their helpful suggestions.

References A RA´ MBULA , M. R. 2006. Estado de esfuerzos en el volca´n Popocate´petl obtenidos con mecanismos focales, en el periodo de actividad de 1996 a 2003. Thesis. Posgrado en Ciencias de la Tierra. Universidad Nacional Auto´noma de Me´xico.

C URRENTI , G., D EL N EGRO , C., L APENNA , V. & T ELESCA , L. 2005. Fluctuation analysis of the hourly time variability of volcano-magnetic signals recorded at Mt. Etna Volcano, Sicily (Italy). Chaos, Solutions and Fractals, 23, 1921–1929. D EL N EGRO , C., C URRENTI , G., N APOLI , R. & V ICARI , A. 2004. Volcanomagnetic changes accompanying the onset of the 2002–2003 eruption of Mt. Etna, Italy. Earth and Planetary Science Letters, 229, 1–14. D ZURISIN , D., D ENLINGER , R. & R OSENBAUM , J. G. 1990. Cooling rate and thermal structure determined from progressive magnetization of Dacite dome at Mount St. Helens, Washington. Journal of Geophysical Research, 95, 2763–2780. E SPINASA , R. & M ARTIN -D EL P OZZO , A. L. 2005. Morphostratigrafic Evolution of Popocatepetl Volcano. Geological Society of America Special Papers, 402, 115–137. L AHR , J. C., C HOUET , B. A., S TEPHENS , C. D., P OWER , J. A. & P AGE , R. A. 1994. Earthquake classification, location, and error analysis in a volcanic environment: implications for the magmatic system of the 1989– 1990 eruptions of Redbout Volcano, Alaska. Journal of Volcanology and Geothermal Research, 62, 137–151. L ERMO -S AMANIEGO , J., A NTAYHUA -V ERA , Y. & C HAVACA´ N -A VILA , M. 2006. Ana´lisis de la actividad sı´smica del Popocate´petl (Me´xico) durante el periodo 1994– 1997. Boletı´n Sociedad Geolo´gica Mexicana, 58, 253–257. M ALONE , S. D. 1983. Volcanic earthquakes: examples from Mount St. Helens. In: K ANAMORI , H. & B OSCHI , E. (eds) Earthquakes: Observation, Theory and Interpretation. Societa´ Italiana di Fisica, Bologna, Italia, 436– 455. M ARTIN -D EL P OZZO , A. L., E SPINASA -P EREN˜ A , R. ´ n de cenizas y variaciones ET AL . 1995. La emisio geoquı´micas durante diciembre-marzo en el Volca´n Popocate´petl; Estudios Realizados durante la Crisis de 1994–1995. Sistema Nacional de Proteccio´n Civil-CENAPRED-UNAM, Me´xico, 285–294. M ARTIN -D EL P OZZO , A. L., C IFUENTES -N AVA , G. ET AL . 2002a. Volcanomagnetics signals during the recent Popocate´petl (Me´xico) eruptions and their relation to eruptive activity. Journal of Volcanology and Geothermal Research, 113, 415–428. M ARTIN -D EL P OZZO , A. L., A CEVES , F. ET AL . 2002b. Influence of volcanic activity on spring water at Popocatepetl volcano, Mexico. Chemical Geology, 190, 207–229. M ARTIN -D EL P OZZO , A. L., A CEVES , F., I NGUAGGIATO , S., S AENZ , H. & A GUAYO , A. 2002c. Spring water and CO2 interaction at Popocatepetl volcano, Mexico. Geofı´sica Internacional, 41, 345–351. M ARTIN -D EL P OZZO , A. L., C IFUENTES , G., C ABRAL C ANO , E., B ONIFAZ , R., C ORREA , F. & M ENDIOLA , I. F. 2003. Timing magma ascent al Popocate´petl Volcano, Me´xico, 2000–2001. Journal of Volcanology and Geothermal Research, 125, 107–120. M ARTIN -D EL P OZZO , A. L., G ONZA´ LEZ -M ORAN , T., E SPINASA -P EREN˜ A , R. & B UTRO´ N , M. A. 2007. Characterization of the recent ash emissions at Popocate´petl volcano, Mexico. Journal of Volcanology and Geothermal Research, 170, 61– 75.

MAGNETIC SIGNATURES AT POPOCATEPETL M INAKAMI , T. 1960. Fundamental research for predicting volcanic eruptions (Part 1). Earthquakes and crustal deformations originating from volcanic activities. Bulletin of Earthquake Research, 38, 497–544. M INAKAMI , T. 1974. Seismology of volcanoes in Japan. In: C IVETTA , L., G ASPARINI , P., L UONGO , G. & R APOLLA , A. (eds) Developments in Solid Earth Geophysics. Physical Volcanology, Elsevier, Amsterdam, 1 –27. R IKITAKE , T. 1968. Geomagnetism and earthquake prediction. Tectonophysics, 6, 59–68. S TRAUB , S. M. & M ARTIN -D EL P OZZO , A. L. 2001. The significance of phenocryst diversity in the tephra from recent eruptions at Popocatepetl Volcano (Central Mexico). Contributions to Mineralogy and Petrology, 140, 487– 510. T ANAKA , Y. 1995. Volcanomagnetic effects on the Unzen Volcano (1990– 1992). Journal of Geomagnetism and Geoelectricity, 47, 325–336. U EDA , H., M ATSUMOTO , T., F UJITA , E., U KAWA , M., Y AMAMOTO , E., S ASAI , Y., I RWAN , M. &

131

K IMATA , F. 2006. Geomagnetic changes associated with the dike intrusion during the 2000 Miyakejima eruptive activity, Japan. Earth and Planetary Science Letters, 245, 416– 426. W IENERT , K. A. 1970. Notes on Geomagnetic Observatory and Survey Practice. UNESCO, Belgium, 217. Z LOTNICKI , J. & B OF , M. 1998. Volcanomagnetic signals associated with the quasi-continuous activity of the andesitic Merapi Volcano, Indonesia: 1990– 1995. Physics of the Earth and Planetary Interiors, 105, 119– 130. Z LOTNICKI , J., B OF , M. ET AL . 2000. Magnetic monitoring at Merapi Volcano, Indonesia. Journal of Volcanology and Geothermal Research, 100, 321– 336. Z LOTNICKI , J., L E -M OUE¨ L , J., D ELMOND , J., P AMBRUM , C. & D ELORME , H. 1993. Magnetic variations on Piton de la Fournaise Volcano Volcanomagnetic signals associated with the November 6 and 30, 1987, eruptions. Journal of Volcanology and Geothermal Research, 56, 281–296.

Understanding crystal populations and their significance through the magma plumbing system DOUGAL A. JERRAM & VICTORIA M. MARTIN Department of Earth Sciences, University of Durham Science Laboratories, South Road, Durham DH1 3LE, UK (e-mail: [email protected]) Abstract: Crystals are rarely composed of a single crystal population that have grown solely from the batch of magma in which they are resident on emplacement, either by eruption or shallow intrusion. Close investigation of the majority of crystal populations reveal that they comprise up to four main components: phenocrysts, crystals co-genetic with their magmatic host; xenocrysts, crystals wholly, or in part, foreign to the magmatic host and magma system; antecrysts, crystals which are recycled one or several times before inclusion in the host magma but have an origin within the magmatic system; and microlites, which represent small co-genetic crystals which nucleate and grow rapidly on decompression and eruption. Textural analysis techniques are employed to quantify key aspects of the crystal population, including crystal shape, crystal size distributions, spatial distribution patterns and textural modification using dihedral angles. Santorini provides a case study of an active volcanic system where a combined textural analysis study has been developed, highlighting how the crystal population is being continuously modified by a series of replenishment and mixing events. Developing textural and microgeochemical techniques provides the next stage in the interrogation of crystal populations, linking textures to isotopic heterogeneities and providing fingerprints of where crystals are sourced and re-cycled.

Magmas rarely arrive at their destinations, either as eruptive products or shallow level intrusions, without a payload of crystals. In the few examples where they do, solidification–crystallization follows without the significant development of large crystals, as is seen in some sill complexes and lava sequences (e.g. Marsh 2004). It is more typical for magmas to contain large crystals entrained within the magma host on emplacement, leading to a variety of porphyritic textures, observed within a range of different magma compositions. Further examination of these ‘crystal cargos’ commonly identifies a variety of crystal populations, which are not formed from a single genetically related crystallization event (e.g. Wallace & Bergantz 2002; Turner et al. 2003, Davidson et al. 2007a), but are the product of a number of processes, such as crystal recycling, which contribute to the final crystal populations that are observed. Although recycled and assimilated crystals pose potential problems to standard whole-rock geochemical approaches, the information that is locked within such crystals provides a window into key aspects of the magma system such as timescales of magma recharge events, contamination histories, magma mixing and magma plumbing histories (e.g. Davidson et al. 2007a). Thus, by using a combination of textural and microgeochemical analysis, some of this information, and the contribution of the crystal population to the overall chemical budget of the magma, can be determined.

This article aims to summarize the nature of crystal populations in volcanic and shallow magmatic systems and the methods used to quantify their textures. For example, what are their components? How are they formed? How do we begin to interrogate them to maximize the information that can be obtained from the final crystal population? Firstly, the different components of the crystal population are introduced. Then a brief summary of the various textural analysis techniques available to quantify these textures is presented, and a case study volcanic system (Santorini, Greece) is used to highlight the benefits of a combined textural approach. Finally, we look forward to the emerging research into the joint application of textural and microgeochemical analysis.

What makes up a crystal population? Magmatic systems are typically very complex, and involve numerous processes such as cooling, nucleation and growth of crystals, magma and crystal recycling, decompression crystallization and storage. The resultant crystal populations that we find erupted in volcanic rocks or frozen in shallow intrusions are therefore often far more complex than might be immediately obvious (Jerram & Davidson 2007). Thus, a detailed examination and quantification of the crystal textures can yield important insights into the geochemical and time history of the magma system. There are four main components

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 133–148. DOI: 10.1144/SP304.7 0305-8719/08/$15.00 # The Geological Society of London 2008.

134

D. A. JERRAM & V. M. MARTIN

that may be part of a multiple crystal population within an igneous rock: phenocrysts, xenocrysts, antecrysts, and microlites (see Fig. 1).

Phenocrysts Phenocrysts are large, often euhedral, crystals in a groundmass matrix of finer grained crystals and/ or glass, and are presumed to represent nucleation and growth from the liquid that is now represented by the groundmass or glass. Thus, we define phenocrysts to be crystals which crystallized directly from the magma involved in the current eruption (e.g. Gill et al. 2006), and as such, to be in both chemical and isotopic equilibrium with the melt. The term primocrysts is often used in the plutonic literature to describe primary (euhedral) crystals which are formed early in the textural development of cumulate rocks, and are probably a mixture of both accumulated phenocrysts and antecrysts.

recycled through different magma replenishment events, or stored in crystal accumulations from the magma, to be reincorporated during the final stages of eruption or emplacement. Antecrysts can therefore be used to fingerprint key stages in the evolution of the magma system as a whole.

Microlites Microlites are typically less than a few hundred microns in size, and are formed during degassing of the magma on eruption (e.g. Couch et al. 2003). Conditions at this final interval during eruption favour increased nucleation and rapid crystal growth, typically resulting in microlites with an acicular morphology, which can be quantified quite readily when in low abundance in glassy samples (e.g. Castro et al. 2003). Microlites are typically in chemical equilibrium with the melt, the rims of phenocrysts and antecrysts, and possibly the xenocryst rims that are growing at the time of eruption.

Xenocrysts Xenocrysts can be defined as crystals which are foreign to the magma system as a whole but have been incorporated into the magma by some physical process. Xenocrysts often have new overgrowths from the magma that they are now resident in, and may be recycled during replenishing events, resulting in multiple additional zones. They are accidental crystals which may have a wide variety of sources and thus may not represent part of the active magma system. The variety of possible sources complicates their identification and interpretation. For example, xenocrysts which are sourced from very shallow wall rocks comprised of previous eruptive products, or from solidified plutonic rocks from the same system, may have similar geochemical signatures to the current eruptive products, and will thus be difficult to detect. In such cases they are often grouped together with antecrysts (e.g. Gill et al. 2006) during classification. In contrast, xenocrysts which are sourced from markedly different rock types such as wall rocks or basement, can be relatively easy to identify.

Antecrysts The term ‘antecrysts’ (after Wes Hildreth at the ‘Longevity and Dynamics of Rhyolitic Magma Systems’ Penrose Conference, 2001; from Gill et al. 2006; see also Charlier et al. 2005; Davidson et al. 2007a) is becoming the accepted term used to define crystals that have been ‘reincorporated’ into the final magma; they differ from xenocrysts, in that they have crystallized from progenitors of the final magma, and so are directly associated with the active magmatic system. They may have been

Mixing the components to make the final texture? In a volcanic system it is easy to see how the different crystal components described above may be combined to arrive at a final crystal population. For example, consider an open magmatic system with a series of linked chambers; the final mixed crystal population may contain xenocrysts scavenged from country or wall rock, antecrysts derived from various replenishing magma batches or remobilized cumulate piles, as well as phenocrysts and microlites that have crystallized directly from the erupted magma. Some textural modification of the xenocrysts and antecrysts may occur after incorporation into, and transport by, the host magma, but the different components are typically still readily identifiable. In contrast, when considering shallow-level intrusive rocks, additional processes may significantly modify the primary composition and textures. Thus, identification of the original crystal components is often difficult, and further terms are used to describe any modification. For example, cumulates essentially represent accumulations of crystals which are often significantly modified in both texture and composition by post-cumulus processes. The term adcumulate is used to reflect texturally equilibrated examples of accumulations of unzoned cumulus crystals that lack any intercumulus phases, and orthocumulate describes cumulus crystals that are typically normally zoned, and are poikilitically enclosed by new mineral phases which are nucleated from the intercumulus liquid. A mesocumulate texture is intermediate between that of an

UNDERSTANDING CRYSTAL POPULATIONS

135

Fig. 1. The magmatic crystal cargo: key components of crystal populations in volcanic systems (see text for details). ‘P’ indicates phenocrysts.

136

D. A. JERRAM & V. M. MARTIN

adcumulate and an orthocumulate, with some intercumulus phases present, and limited zoning of the cumulus phases.

Quantifying the textural parameters of the population: crystal size, shape, distribution and textural equilibration The first stage that should be employed when dealing with rocks that contain a discernible crystal population on eruption or emplacement is to quantify the key textural parameters such as size, shape and spatial distribution of the crystal population, and to ascertain a level of textural equilibration. This important first step, which is often overlooked, can provide an early indication as to whether xenocrysts and antecrysts may play a significant role in the population, and therefore need to be identified and accounted for when undertaking geochemical analysis (Jerram & Kent 2006; Jerram & Davidson 2007). They can also be used to model the development of the crystal population through time (e.g. Marsh 1988, 1998). Simple petrographic analysis may identify xenocrysts, and possibly antecrysts, if there are significant textural differences between them and the phenocryst population (e.g. sieve cores, clear resorbtion horizons, etc.); however, this is not always the case. The advent of computer analysis systems and the development of detailed textural analysis techniques means that the key parameters that define a crystal population can be readily determined (e.g. Higgins 2006 and references therein). The size, shape and spatial distribution of a crystal population can be ascertained from a statistically valid sample of a two-dimensional (2D) slice through the crystal population (Fig. 2). In order to analyse the rock, the texture is digitized so that individual crystals and different phases can be identified by an image analysis system. Either a true binary image (Higgins 2006) or a combination grey scale image (where multiple phases are present, e.g. Jerram et al. 2003), is used to provide the basis for the geometric determination. Individual phases can be selected, and the image analysis system is used to measure a variety of parameters, including the twodimensional crystal long-axis and short-axis, crystal roundness and x –y coordinates of the crystal centre. The data can then be used to determine the crystal shape, crystal size distribution and spatial distribution pattern of the texture, which are briefly described below.

Crystal shapes The crystal shape can be defined by the long, intermediate and short axis (often labelled L:I:S, Fig. 2b)

as a ratio where S ¼ 1. It is desirable to determine the habit or shape of the crystal population for a number of reasons. Firstly, the L:I:S ratios are used to convert 2D crystal size measurements into predicted 3D crystal size distributions (see below). Secondly, the crystal habit is known to be a function of the degree of undercooling; marked changes in undercooling and saturation control the resultant morphology of the crystals. If the shape can be determined accurately, we may then be able to determine the conditions under which the crystals grew. Studies by Lofgren (1974, 1980) have shown that, at different degrees of undercooling, plagioclase morphology ranges from acicular, to tabular, to equant shapes. Experiments in olivine systems (e.g. Donaldson 1976) also show a variety of morphologies within different cooling systems. It is also likely that the different population components, such as phenocrysts, xenocrysts and antecrysts, have different morphologies originating from differing growth conditions. It is possible to determine a 3D crystal shape from 2D crystal sections, providing the 3D shape of all the crystals is the same using the width length information from the 2D data (Higgins 1994; Garrido et al. 2001). Morgan & Jerram (2006) produced a spread sheet which can be used to statistically compare 2D shape measurements to the best fit 3D shape. Recently, investigation of 3D crystal populations has shown that shape can vary significantly and so further exploration of the role of crystal shape is required (Jerram & Higgins 2007).

Crystal size distributions One of the most commonly measured textural parameters is that of crystal size distribution (CSD analysis), as the size distribution of the crystal population can provide important information about nucleation and growth within the magma system (Marsh 1988; Cashman & Marsh 1988). A plot of population density v. size is used to characterize the distribution (Fig. 2c), and the shape of the CSD curve can be used to identify both simple crystallization and physical processes such as mixing of crystal populations, or Ostwald ripening within the crystal population (e.g. Higgins 1998; Marsh 1998; Turner et al. 2003). The slope of the CSD is a function of the growth rate and residence time of the crystals, thus where crystal growth rates are well constrained, timescales within magma systems can be investigated (or vice versa). Most CSD data are measured from 2D sections, and the final 3D CSD is achieved using stereological conversions (Higgins 2000). These conversions require information on crystal shape, roundness and texture fabric, which can be determined from the 2D statistics from thin sections. More recently,

UNDERSTANDING CRYSTAL POPULATIONS

137

Fig. 2. Detailed textural analysis techniques. (a) Digitized 2D sections through crystal populations are analysed using image analysis software to measure crystal parameters such as the long–short axis, grain-centre, roundness, etc. (b) Statistics of the long:short axis ratio in 2D are used to define the true 3D aspect ratio of the crystal (e.g. Morgan & Jerram 2006), S, I and L represent short, intermediate and long axes, respectively. (c) The crystal size distribution is calculated from crystal length, 3D shape, roundness and orientation information, to show the population density as a function of size in 3D (e.g. Higgins 2000). (d) From the grain centre coordinates the spatial distribution pattern (R) can be determined, and the packing arrangement in 3D can be quantified on a porosity (% melt) v. R plot (e.g. Jerram et al. 2003), where RSDL denotes the random sphere distribution line (see text for details). The SDPs of a number of known 3D textures provide reference points on the R v. porosity plot, and can then be used to interpret real rock textures (see Jerram et al. 2003 for details).

true 3D CSD measurements have been produced by using either serial sectioning/reconstruction techniques or X-ray CT analysis (e.g. Mock & Jerram 2005; Gualda 2006), and additional plots have been suggested to further examine the crystal population (e.g. Higgins 2002, 2006).

Spatial distribution patterns and crystal orientations The packing arrangement of crystals in a rock greatly influences its material properties such as porosity, permeability and strength (Allen 1985;

138

D. A. JERRAM & V. M. MARTIN

Rogers et al. 1994). In addition, a knowledge of the spatial organisation or spatial distribution pattern (SDP) of crystals in an igneous rock can provide important information on the nucleation distribution of the crystals and reveal much about how the texture formed (Jerram et al. 1996), particularly as clustered crystal frameworks are common in igneous rock systems (Jerram et al. 2003). The porosity (modal abundance) v. R value, a measure of spatial arrangement based on grain centre coordinates (after Jerram et al. 1996), provides a means to quantify the spatial distribution pattern of crystal population (e.g. Fig. 2d). The SDPs of a number of known 3D textures provide reference points on the R v. porosity plot, which can then be used to interpret real rock textures (e.g. Jerram et al. 1996, 2003; Mock et al. 2003). The SDP pattern of randomly packed distributions of equal sized spheres varies systematically with porosity, producing a line on the R v. porosity plot, termed the random sphere distribution line (RSDL). Textures which plot below the RSDL have a clustered texture where groups of individual particles are closer together than expected, and textures which plot above the RSDL have a more ordered distribution (Jerram et al. 1996). Thus, the crystal population can be clustered, random or ordered and it is possible to ascertain if the crystals are forming a touching framework in 3D, which can be very useful where it is not possible to image in 3D or samples are restricted as in lunar and meteorite samples (Day & Taylor 2007). The application of more detailed cluster analysis techniques (e.g. Jerram & Cheadle 2000) can further investigate the packing arrangement of the crystals, and the application of such techniques to 3D data sets will be the next stage. Where there is a preferred orientation of the crystals, techniques can be employed to quantify the degree of anisotropy in the texture. For example, anisotropy of magnetic susceptibility (AMS) of igneous rocks measures the preferred orientation of the magnetic minerals in a sample and is used to determine the shape, strength and orientation of the magnetic fabric in a sample (e.g. Bouchez 1997; O’Driscoll et al. 2007). Other methods used to quantify three-dimensional fabrics include traditional structural measurements and the collection of grain orientation data obtained from three orthogonal sections [specific methods include the intersection method of Launeau & Cruden (1998) and the use of cumulative distribution functions by Gee et al. (2004)].

the degree of textural equilibration that has occurred between the connecting crystals and the melt can be quantified. The 3D dihedral angle is the angle formed at the junction between any interstitial phase and/or melt, and any two crystal grains, under conditions of textural equilibrium, and can provide a wealth of information on the textural development of a magmatic system through time (Fig. 3). Crystal shapes controlled by the kinetics of growth tend to be bounded by planar surfaces; the apparent dihedral angles which develop at pore corners in an aggregate of randomly orientated crystals form an ‘impingement texture’, with a median angle of c. 608, and a standard deviation of 25– 308 (Elliot et al. 1997; Holness et al. 2005a). In contrast, a texturally equilibrated pore structure has a lower median angle of c. 308, with a standard deviation of 13–148 (e.g. Von Bargen & Waff 1988; Laporte 1994; Holness et al. 2005a), and is associated with more rounded crystal forms (see Fig. 3). Detailed analysis of dihedral angles has been used to investigate the late-stage textural evolution of large intrusive igneous bodies (e.g. Rum, Skaergaard; Holness et al. 2005a, 2007). Differences in the textural maturity of different phase assemblages have also been shown to reflect differences in cooling and solidification rates, demonstrating that the liquidus assemblage plays a direct role in determining the thermal history of plutons (Holness et al. 2007). The dihedral angle population of glomeroporphyritic olivine crystals in komatiities from Zimbabwe has also been investigated, revealing that the crystal mush source of the glomerocrysts was not fully texturally equilibrated before it was disaggregated and incorporated into the komatiite magma (Jerram et al. 2003). This suggests that either the crystal mush was heading to equilibration but did not reach it prior to incorporation, or that the crystal population is a complex mix of partially equilibrated glomerocrysts which have been incorporated at different stages by the magma. Modification of the dihedral angle population in some volcanic rocks has also been observed. Holness et al. (2005b) have shown that plagioclase – plagioclase – melt dihedral angles in mafic enclaves from Santorini, Greece, are modified when quench overgrowths develop as the enclaves are incorporated within a cooler host. The lowest dihedral angles occur in samples with the most quench overgrowth, where the growth rate exceeds the diffusion rate of material to the interface.

Textural equilibration – dihedral angles

A combined textural analysis approach

In examples where crystals form an interlocking framework or are found in clumps and clusters,

Textural analysis techniques offer quantitative insights into the crystal population, and although

UNDERSTANDING CRYSTAL POPULATIONS

139

Fig. 3. Relating textural equilibration to the crystal population. A crystal network forms from an initial population, with an impingement texture defined by the angles of the original interlocking crystals. During textural equilibration these angles are modified through time to a lower median dihedral angle (Holness et al. 2005b), but the CSD remains unchanged. However, CSDs may be modified by other processes of textural maturation, such as annealing and Ostwald ripening, where the surface energy of a phase is minimized by the simultaneous dissolution of smaller grains and growth of larger grains, thus modifying the original texture.

somewhat time-consuming, are an ideal and costeffective precursor to geochemical approaches. Some case study volcanoes are now emerging where a number of the techniques presented above have been tested (e.g. Soufriere, St Vincent, Zellmer et al. 2005; Santorini, Greece, Martin et al. 2006a, b). The application of textural analysis to active volcanic systems allows us to constrain vital timescale information on rates and cyclicity of volcanic processes. Below we will use Nea Kameni Volcano, Santorini, as a brief case study to highlight how a variety of detailed textural analysis techniques have been applied to obtain information on both long- and short-term processes occurring beneath the volcano.

Santorini Volcano, Greece The post-caldera Kameni Islands lie in the centre of the flooded caldera of Santorini (cf. Fig. 5i, ii). The

caldera is a composite structure resulting from at least four caldera forming events (Druitt & Francaviglia 1992), the most recent of which occurred after the Minoan eruption c. 3600 years ago (Druitt et al. 1999). Magmatic activity probably resumed soon after the Minoan eruption, producing the intracaldera volcanic islands of Palaea and Nea Kameni, which have been the focus of historic volcanism on Santorini. The Nea Kameni edifice broke the surface in 197 BC and subsequently there have been at least nine subaerial episodes of volcanic activity, the last of which occurred in AD 1950. The Kameni islands are formed from dacite lava flows, which contain abundant magmatic enclaves, interpreted to be the quenched fragments of replenishing magmas which periodically trigger eruptive events (Martin et al. 2006a). The timing of these events is very well constrained, with eyewitness accounts existing for many of the eruptions. Combined with the availability of textural information

140

D. A. JERRAM & V. M. MARTIN

from both the host lavas and the replenishing enclaves, this makes Santorini an ideal case study for the application of textural analysis to an active volcanic system.

What makes up the crystal population of the Kameni enclaves? The Kameni lava and enclave crystal populations display many of the characteristic crystal types described above. The enclaves typically contain a population of large crystals comprising a mixture of phenocrysts and antecrysts, and a population of smaller microlite crystals (Fig. 4). The phenocrysts and microlites are in both chemical and isotopic equilibrium with the melt, and are believed to have crystallized directly from the batch of replenishing magma in which they are resident. The antecryst population is formed from crystals which may be chemically and isotopically zoned, and display a sieve-textured core or evidence of resorbtion and overgrowth. These crystals are also typically in chemical and isotopic disequilibrium with the host melt, and are thought to represent crystals which have been recycled from isotopically distinct magma batches which reside at deeper levels within the current, active system. Martin et al. (2006a) recognized three replenishing magmatic enclave types in the flows of Nea Kameni, each with a characteristic CSD, SDP and dihedral angle signature (Martin et al. 2006a, b; Holness et al. 2005b). Figure 5 provides a schematic look at the magma system underneath Santorini, indicating where the key textural data provide quantitative information.

Crystal size distributions In order to develop a better understanding of the magmatic processes occurring at Santorini, crystal size distributions have been determined for both the host lavas (Higgins 1996) and the mafic enclaves (Fig. 5a; Martin et al. 2006a). The initial CSD study of the lavas (Higgins 1996) found the plagioclase crystal population to be the result of mixing of several different magmas, leading Higgins (1996) to propose a model involving successive periods of replenishment of the chamber with an aphyric dacite which then mixed completely with the older, phenocryst-bearing, host. Using arguments based on plausible plagioclase growth rates, he suggested that the aphyric magmas intruded into the magma chamber between 6 and 13 years prior to eruptions, and that the subsequent eruption did not empty the chamber, leaving behind dacite into which further aphyric magma would later intrude. These short timescales of storage and crystallization are consistent with observations of short-lived

radioactive disequilibria in Kameni lavas (Zellmer et al. 2000), and with the patterns of zonation in Kameni phenocrysts (e.g. Druitt et al. 1999; Zellmer et al. 1999). More recent studies have concentrated on the abundant magmatic enclaves, which are found in all the flows on the Kameni Islands. Textural analysis of groundmass plagioclase in the enclaves has been used to establish that the replenishing magmas were emplaced as a dense layer at the base of the dacite chamber, where crystallizationinduced vesiculation in the layer of replenishing magma was then sufficient to lower the density of the mafic magma below that of the host resulting in layer overturn and mixing (Holness et al. 2005). Groundmass CSDs from the various enclave types also suggest that there is a positive correlation between the volume of each flow and the volume of replenishing magma emplaced in the chamber prior to eruption, implying that the replenishment may act as an eruption trigger (Martin et al. 2006a). Disequilibrium phenocryst assemblages in the lavas and phenocryst-bearing enclaves provide evidence for the entrainment and recycling of older phenocryst populations into the lavas and enclave magmas (StamatelopoulouSeymour et al. 1990; Martin et al. 2006a).

Spatial distribution patterns SDP analysis and melting experiments have been used to understand the rheology and dispersal mechanisms of the replenishing magmas (Martin et al. 2006b), and demonstrate that the plagioclase crystallinity and crystal aspect ratios observed in the Kameni enclaves are sufficient to form a touching framework of crystals (Fig. 5b). Petrographic observations on the coherence of the framework during exsolution of volatiles and enclave transport suggest that the morphology of the dominant framework-forming phase places a significant control on the strength and viscosity of the crystal framework, which in turn determines the extent to which the replenishing magmas are able to fragment and mix with the dacite host (Martin et al. 2006b).

Dihedral angles Holness et al. (2005b) have shown that the angles subtended between pairs of framework-forming plagioclase grains in mafic enclaves from Santorini have been modified as a result of quench-related overgrowth (Fig. 5c). The various batches of Kameni replenishing magma were emplaced as layers at the base of the chamber, and the enclaves formed subsequently by overturn of this partially crystallized layer once it had become buoyant due to crystallization-induced vesiculation (Martin

UNDERSTANDING CRYSTAL POPULATIONS

141

Fig. 4. Photomicrographs to illustrate the crystal types which make up the crystal population in the Kameni enclaves. (a) The enclaves contain microlites and phenocrysts which have crystallized from the host melt, and a population of recycled antecrysts. (b) The antecrysts typically have sieve-textured cores and or not in isotopic equilibrium with the host. (c) The microlite population typically comprises plagioclase crystals with a high aspect-ratio, indicative of rapid growth on quenching or eruption.

et al. 2006a). Each enclave type has a distinct population of modified angles, and the lowest dihedral angles occur in samples with the most quench overgrowth. Holness et al. (2005b) suggest that this is due to temperature variations between the different batches of replenishing magmas and the host dacite when they mix, caused by varying H2O contents in the replenishing magmas. For a fixed host dacite H2O content, the density of the intruding layer reaches that of the host dacite magma at successively lower temperatures as the H2O content of the replenishing magma is

decreased (Holness et al. 2005b). The amount of quench-related growth, and modification of the original impingement angle is therefore a function of the temperature difference and H2O content of the two magmas (Holness et al. 2005b). Microgeochemical techniques are now being used on the enclave magmas to examine the recycled crystal population in detail, in order to determine a potential source (or sources) for the crystals, and the timescale over which the recycling and replenishment processes take place. This work will complement the existing textural data, enabling

142

D. A. JERRAM & V. M. MARTIN

Fig. 5. An integrated textural study of the Nea Kameni lavas and mafic enclaves, Santorini (see text for details). ‘P’ denotes phenocrysts which have formed within the upper chamber, and ‘A’ denotes antecrysts which are derived from the disaggregation of mafic enclaves.

UNDERSTANDING CRYSTAL POPULATIONS

us to develop a more complete understanding of the processes occurring in the magmatic plumbing system beneath Santorini.

Future developments Developments in textural analysis A new and exciting area of textural analysis is that of crystal growth modelling. This can be done in both 2D and 3D using kinetic growth models, and has the potential to incorporate geochemical constraints. In the field of materials science, work has focused on evolution of internal interfaces in polycrystalline materials (e.g. Holm & Battaile 2001). In igneous petrology, the texture development of the crystal population from nucleation to growth and the formation of crystal frameworks is the focus. Elliott et al. (1997) produced the first curves for randomly orientated dihedral angles using geometric growth models of crystal populations in 3D, and these

143

models were later used to help define percolation thresholds and permeability in crystallizing textures (Cheadle et al. 2004). More sophisticated models are being developed as computer technologies advance, and we are now realizing complex models of fluid flow in igneous rocks (e.g. Hersum et al. 2005) and the development of virtual 3D igneous textures (Hersum & Marsh 2006, 2007). Recent developments in 3D imaging techniques have provided an important new tool to aid in the quantification of igneous texture. Jerram & Higgins (2007) provide a summary of the state of the art in terms of 3D textural analysis, including techniques of serial sectioning – grinding (e.g. Mock & Jerram 2005), X-ray CT analysis (e.g. Ketcham & Carlson 2001; Gualda and Rivers 2006), and reconstruction of the true 3D crystal population within a rock. Figure 6 shows 3D reconstructions of crystal populations from some of these recent studies. As 3D imaging techniques become more accessible to the community, it will soon become commonplace to have a 3D virtual texture

Fig. 6. Examples of 3D reconstructions and models of igneous textures (see Jerram & Higgins 2007). (a) 3D crystal population constructed from serial sections; (b) X-ray CT image of crystals and vesicles from a lava on Teide Volcano, Tenerife; (c) 3D virtual texture modelling of crystal growth (e.g. Hersum & Marsh 2006).

144

D. A. JERRAM & V. M. MARTIN

Fig. 7. (a) Schematic to illustrate the relationships of CSD and micro-analysis to the crystal population. The CSD quantifies the overall population and the largest crystal can be micro-sampled to reveal the isotopic history through crystal growth. Both the shared and separate histories may be mapped out. (b) The combined CSD and isotope diagram (the ICSD plot after Morgan et al. 2007; different shades of grey represent data for three different crystals). Microsampling of Sr isotopes are linked to the CSD plot to show the development of the isotopes through the growth of the texture. In the example shown, a marked kink in the CSD is coincident with the change in isotopic signature (Morgan et al. 2007).

UNDERSTANDING CRYSTAL POPULATIONS

image of a sample to guide sampling strategies for microgeochemical analysis. In such a petrographic framework, the role of the different crystal types will be easier to determine and their influence on the geochemical budget of the rock can be accounted for within the true volume.

Linking textures with microgeochemical analysis With some communities developing techniques in textural analysis, and others pioneering new fields in micro-analysis, it is important that their paths cross as often as possible to see where synergies exist between the disciplines (Jerram & Kent 2006; Jerram & Davidson 2007). There are a number of ways in which the geochemical evolution of the crystal population can be interrogated. Major and trace element data can be obtained from core to rim probe traverses of key crystal phases and correlated with high-resolution SEM images (e.g. Ginibre et al. 2002), mineral separates can be used to look at specific elements (e.g. uranium series, Turner & Costa 2007), and laser ablation (Ramos et al. 2004) and micromilling (e.g. Charlier et al. 2006) can be used to investigate within-crystal isotopic heterogeneities. It is this latter development that we will focus on here. Isotopes have long been used to fingerprint the origin of magmatic components (Faure 1986), and until recently were only considered from whole rock analyses, with any isotopic heterogeneity being linked to source variations or contamination. With recent advances in technology it has become possible to measure isotopes in increasingly small sample sizes, allowing for the investigation of different components of the magmatic system, e.g. crystals, individual crystal zones and groundmass (see Davidson et al. 2007a, b, and references therein). This area of study within igneous petrology has provided a level of detail which has allowed the distinction between phenocrysts and antecrysts and the realization that there is a primary isotopic variation at the grain/sub-grain scale in many crystal-bearing igneous rocks. If isotopic heterogeneities can be mapped out within a texture, then potentially texture development can be linked to key processes such as magma recharge, mixing of crystal components, and contamination, at specific times in the evolution of the crystal population. Turner et al. (2003) combined the use of CSD data and isotopic analysis to resolve discordant age information from Tonga as the result of recycling of crystal cumulates. Zellmer et al. (2005) considered CSD data in conjunction with a number of different geochemical methods to

145

constrain magma evolution at volcanic arcs. With the CSD data enabling the population density at a given crystal size to be established, the potential exists for the combination of the core to rim isotope variation displayed in the largest crystal sizes to be mapped out through the evolution of the crystal population, where the final rim of the crystal is equivalent to the phenocryst-microlite stage of the magma system in equilibrium with the melt. This concept is highlighted in Figure 7a, where the CSD has been measured for the whole crystal population and the largest crystals examined for core to rim isotope variation. The first study to directly link the crystal isotope stratigraphy with the CSD plot is that of Morgan et al. (2007). Focussing on the relatively simple system of Stromboli, it was possible to show an isotopic evolution from core to rim which coincided with a clear kink in the CSD plot (Fig. 7b). The methodology requires the largest crystal to be analysed and the 2D geochemical variation to be corrected for section effects, so that each of the ‘shared’ isotopic zones of the crystal are aligned. The resultant plot shows the large to small CSD plot with the core to rim isotope variation superimposed. The significance of this data is that, not only is it possible to demonstrate that a core to rim isotopic heterogeneity exists, it is also possible to tie it directly to the evolution of the crystal population as inferred from the CSD data (Morgan et al. 2007).

Conclusion With careful examination of the crystals that are preserved in volcanic and shallow intrusive events, it is possible to quantify igneous textures with a high degree of certainty. The shape, size, spatial packing and interlocking angles of the crystal population provide key information on a variety of magmatic processes (growth rates, crystal recycling, textural equilibration, etc.) to enable us to unravel magmatic plumbing systems. Furthermore, by linking the textural analysis techniques with focused micro-geochemical techniques, we are now in a position to fully quantify the crystal component of magma systems, and its role and influence on the chemical budget of the rocks. We would like to thank A. Mock, D. Morgan, D. Chertkoff, J. Davidson, B. Marsh, M. Cheadle, V. Troll, M. Holness and many others for numerous discussions on magmatic textures. This contribution benefited from reviews by J. Roberge, B. O’Driscoll and M. Ban, and we thank them for their input. Bob Hunter is fondly remembered for his enthusiasm of textural analysis, and for getting me (D.A.J.) into this mess in the first place.

146

D. A. JERRAM & V. M. MARTIN

References A LLEN , J. R. L. 1985. Principles of Physical Sedimentology. George Allen & Unwin, Woking, 21–38. B OUCHEZ , J. L. 1997. Granite is never isotropic: an Introduction to AMS studies of granitic rocks. In: B OUCHEZ , J. L., H UTTON , D. H. W. & S TEPHENS , W. E. (eds) Granite: From Segregation of Melt to Emplacement Fabrics. Kluwer Academic, Dordrecht, 95–112. C ASHMAN , K. V. & M ARSH , B. D. 1988. Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization II: Makaopuhi Lava Lake. Contributions to Mineralogy and Petrology, 99, 292 –305. C ASTRO , J. M., C ASHMAN , K. V. & M ANGA , M. 2003. A technique for measuring 3D crystal-size distributions of prismatic microlites in obsidian. American Mineralogist, 88, 1230–1240. C HARLIER , B. L. A., G INIBRE , C., M ORGAN , D. J., P EARSON , D. G., D AVIDSON , J. P. & O TTLEY , C. J. 2006. Methods for the microsampling and highprecision analysis of strontium and rubidium isotopes at single crystal scale for petrological and geochronological applications. Chemical Geology, 232, 114– 133. C HARLIER , B. L. A., W ILSON , C. J. N., L OWENSTERN , J. B., B LAKE , S., V AN C ALSTEREN , P. W. & D AVIDSON , J. P. 2005. Magma generation at a large, hyperactive silicic volcano (Taupo, New Zealand) revealed by U– Th and U–Pb systematics in zircons. Journal of Petrology, 46, 3– 32. C HEADLE , M. J., E LLIOTT , M. T. & M C K ENZIE , D. 2004. Percolation threshold and permeability of crystallizing igneous rocks: the importance of textural equilibrium. Geology, 32, 757 –760. C OUCH , S., H ARFORD , C. L., S PARKS , R. S. J. & C ARROLL , M. R. 2003. Experimental constraints on the conditions of formation of highly calcic plagioclase microlites at the Soufrire Hills volcano, Montserrat. Journal of Petrology, 44, 1455–1475. D AVIDSON , J. P., M ORGAN , D. J., C HARLIER , B. L. A., H ARLOU , R. & H ORA , J. 2007a. Tracing magmatic processes and timescales through mineral-scale isotopic data. Annual Reviews of Earth and Planetary Sciences, 35, 273 –311. D AVIDSON , J. P., M ORGAN , D. & C HARLIER , B. 2007b. Isotopic microsampling of magmatic rocks. Elements, 3, 253– 260. D AY , J. M. D. & T AYLOR , L. A. 2007. On the structure of mare basalt lava flows from textural analysis of the LaPaz Icefield and Northwest Africa 032 lunar meteorites. Meteoritics and Planetary Science, 42, 3 –17. D ONALDSON , C. H. 1976. An experimental investigation of olivine morphology. Contributions to Mineralogy and Petrology, 57, 187– 213. D RUITT , T. H. & F RANCAVIGLIA , V. 1992. Caldera formation on Santorini and the physiography of the islands in the late Bronze Age. Bulletin of Volcanology, 54, 484– 493. D RUITT , T. H., E DWARDS , L. ET AL . 1999. Santorini Volcano. Geological Society, London, Memoirs, 19. E LLIOTT , M. T., C HEADLE , M. J. & J ERRAM , D. A. 1997. On the identification of textural equilibrium in rocks

using dihedral angle measurements. Geology, 25, 355–358. F AURE , G. 1986. Principles of Isotope Geology. J. Wiley, New York. G ARRIDO , C. J., K ELEMEN , P. B. & H IRTH , G. 2001. Variation of cooling rate with depth in lower crust formed at an oceanic spreading ridge: plagioclase crystal size distributions in gabbros from the Oman ophiolite. Geochemistry, Geophysics, and Geosystems, 2, Paper number 2000GC000136. G ILL , J., R EAGAN , M., T EPLEY , F. & M ALAVASSI , E. 2006. Arenal volcano, Costa Rica. Magma Genesis and Volcanological Processes. 157, 15 September, 1– 8. G EE , J. S., M EURER , W. P., S ELKIN , P. A. & C HEADLE , M. J. 2004. Quantifying three-dimensional silicate fabrics in cumulates using cumulative distribution functions. Journal of Petrology, 45, 1983– 2009. G INIBRE , C., K RONZ , A. & W ORNER , G. 2002. Highresolution quantitative imaging of plagioclase composition using accumulated backscattered electron images: new constraints on oscillatory zoning. Contributions to Mineralogy and Petrology, 142, 436– 448. G UALDA , G. A. R. 2006. Crystal size distributions derived from 3D datasets: sample size versus uncertainties. Journal of Petrology, 47, 1245– 1254. G UALDA , G. A. R. & R IVERS , M. 2006. Quantitative 3D petrography using X-ray tomography: application to Bishop Tuff pumice clasts. Journal of Volcanology and Geothermal Research, 154, 48– 62. H ERSUM , T. G. & M ARSH , B. D. 2006. Igneous microstructures from kinetic models of crystallization. Journal of Volcanology and Geothermal Research, 154, 34–47. H ERSUM , T. & M ARSH , B. 2007. Igneous textures: on the kinetics behind the words. Elements, 3, 247– 252. H ERSUM , T. G., H ILPERT , M. & M ARSH , B. 2005. Permeability and melt flow in simulated and natural partially molten basaltic magmas. Earth and Planetary Science Letters, 237, 798–814. H IGGINS , M. D. 1994. Determination of crystal morphology and size from bulk measurements on thin sections: numerical modelling. American Mineralogist, 79, 113–119. H IGGINS , M. D. 1996. Magma dynamics beneath Kameni volcano, Thera, Greece, as revealed by crystal size and shape measurements. Journal of Volcanology and Geothermal Research, 70, 37–48. H IGGINS , M. D. 1998. Origin of anorthosite by textural coarsening: quantitative measurements of a natural sequence of textural development. Journal of Petrology, 39, 1307– 1323. H IGGINS , M. D. 2000. Measurement of crystal size distributions, American Mineralogist, 85, 1105–1116. H IGGINS , M. D. 2002. Closure in crystal size distribution (CSD), verification of CSD calculations and the significance of CSD fans. American Mineralogist, 87, 160–164. H IGGINS , M. D. 2006. Quantitative Textural Measurements in Igneous and Metamorphic Petrology. Cambridge University Press, Cambridge. H OLM , E. A. & B ATTAILE , C. B. 2001. The computer simulation of microstructural evolution. JOMJournal of the Minerals, Metals and Materials Society, 53, 20–23.

UNDERSTANDING CRYSTAL POPULATIONS H OLNESS , M. B., C HEADLE , M. J. & M C K ENZIE , D. 2005a. On the use of changes in dihedral angle to decode late-stage textural evolution in cumulates. Journal of Petrology, 46, 1565– 1583. H OLNESS , M. B., M ARTIN , V. M. & P YLE , D. M. 2005b. Information about open-system magma chambers derived from textures in magmatic enclaves: the Kameni Islands, Santorini, Greece. Geological Magazine, 142, 1– 13. H OLNESS , M. B., N IELSEN , T. F. D. & T EGNER , C. 2007. Textural maturity of cumulates: a record of chamber filling, liquidus assemblage, cooling rate and large-scale convection in mafic layered intrusions. Journal of Petrology, 48, 141–157. J ERRAM , D. A. & C HEADLE , M. J. 2000. On the cluster analysis of grains and crystals in rocks. American Mineralogist, 85, 47– 67. J ERRAM , D. A. & D AVIDSON , J. P. 2007. Frontiers in textural and microgeochemical analysis. Elements, 3, 235–238. J ERRAM , D. A. & H IGGINS , M. D. 2007. 3D analysis of rock textures: quantifying igneous microstructures. Elements, 3, 239– 245. J ERRAM , D. A. & K ENT , A. 2006. An overview of modern trends in petrography: textural and microanalysis of igneous rocks. Journal of Volcanology and Geothermal Research, 154, vii–ix. J ERRAM , D. A., C HEADLE , M. J., H UNTER , R. H. & E LLIOTT , M. T. 1996. The spatial distribution of grains and crystals in rocks. Contributions to Mineralogy and Petrology, 125, 60–74. J ERRAM , D. A., C HEADLE , M. C. & P HILPOTTS , A. R. 2003. Quantifying the building blocks of igneous rocks: are clustered crystal frameworks the foundation? Journal of Petrology, 44, 2033–2051. K ETCHAM , R. A. & C ARLSON , W. D. 2001. Acquisition, optimization and interpretation of X-ray computed tomographic imagery: applications to the geosciences. Computers & Geoscience, 27, 381– 400. L APORTE , D. 1994. Wetting behavior of partial melts during crustal anatexis – the distribution of hydrous silicic melts in polycrystalline aggregates of quartz. Contributions to Mineralogy and Petrology, 116, 486–499. L AUNEAU , P. & C RUDEN , A. R. 1998. Magmatic fabric acquisition mechanisms in a syenite: results of a combined anisotropy of magnetic susceptibility and image analysis study. Journal of Geophysical Research, 103, 5067–5089. L OFGREN , G. 1974. An experimental study of plagioclase crystal morphology: isothermal crystallization. American Journal of Science, 274, 243–273. L OFGREN , G. E. 1980. Experimental studies on the dynamic crystallization of silicate melts. In: H ARGRAVES , R. B. (ed.) Physics of Magmatic Processes. Princeton University Press, Princeton, NJ, 487–551. M ARSH , B. D. 1988. Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization I. Theory. Contributions to Mineralogy and Petrology, 99, 277– 291. M ARSH , B. D. 1998. On the interpretation of crystal size distributions in magmatic systems. Journal of Petrology, 39, 553 –599.

147

M ARSH , B. D. 2004. A magmatic mush column Rosetta Stone: the McMurdo DryValleys of Antarctica. Eos Transactions of the AGU, 85, 497– 502. M ARTIN , V. M., H OLNESS , M. B. & P YLE , D. M. 2006a. Textural analysis of magmatic enclaves from the Kameni Islands, Santorini, Greece. Journal of Volcanology and Geothermal Research, 154, 89–102. M ARTIN , V. M., P YLE , D. M. & H OLNESS , M. B. 2006b. The role of crystal frameworks in the preservation of enclaves during magma mixing. Earth and Planetary Science Letters, 248, 787–799. M OCK , A. & J ERRAM , D. A. 2005. Crystal size distributions (CSD) in three dimensions: insights from the 3D reconstruction of a highly porphyritic rhyolite. Journal of Petrology, 46, 1525–1541. M OCK , A., J ERRAM , D. A. & B REITKREUZ , C. 2003. Using quantitative textural analysis to understand the emplacement of shallow-level rhyolitic laccoliths – a case study from the Halle Volcanic Complex, Germany. Journal of Petrology, 44, 833– 849. M ORGAN , D. J. & J ERRAM , D. A. 2006 On estimating crystal shape for crystal size distribution analysis. Journal of Volcanology and Geothermal Research, 154, 1–7. M ORGAN , D. J., J ERRAM , D. A. ET AL . 2007. Combining CSD and isotopic microanalysis: magma supply and mixing processes at Stromboli volcano, Aeolian Islands, Italy. Earth and Planetary Science Letters, 260, 419–431. O’D RISCOLL , B., H ARGRAVES , R. B., E MELEUS , C. H., T ROLL , V. R., D ONALDSON , C. H. & L EAVY , R. J. 2007. Magmatic lineations inferred from anisotropy of magnetic susceptibility fabrics in Units 8, 9, and 10 of the Rum Eastern Layered Series, NW Scotland. Lithos, 98, 27– 44. R AMOS , F. C., W OLFF , J. A. & T OLLSTRUP , D. L. 2004. Measuring 87Sr/86Sr variations in minerals and groundmass from basalts using LAMC-ICPMS. Chemical Geology, 211, 135–158. R OGERS , C. D. F., D IJKSTRA , T. A. & S MALLEY , I. J. 1994. Particle packing from an earth science viewpoint. Earth Science Reviews, 36, 59–82. S TAMATELOPOULOU -S EYMOUR , K., V LASSOPOULOS , D., P EARCE , T. H. & R ICE , C. 1990. The record of magma chamber processes in plagioclase phenocrysts at Thera volcano, Aegean Volcanic Arc, Greece. Contributions to Mineralogy and Petrology, 104, 73–84. T URNER , S. & C OSTA , F. 2007. Measuring time scales of magmatic evolution. Elements, 3, 267– 273. T URNER , S. P., G EORGE , R., J ERRAM , D. A., C ARPENTER , N. & H AWKESWORTH , C. J. 2003. Case studies of plagioclase growth and residence times in island arc lavas from Tonga and the Lesser Antilles, and a model to reconcile discordant age information. Earth and Planetary Science Letters, 214, 279– 294. V ON B ARGEN , N. & W AFF , H. S. 1988. Wetting of enstatite by basaltic melt at 1350 8C and 1.0–2.5 GPa pressure. Journal of Geophysical Research, 93, 1153– 1158. W ALLACE , G. S. & B ERGANTZ , G. W. 2002. Waveletbased correlation (WBC) of zoned crystal populations and magma mixing. Earth and Planetary Science Letters, 202, 133– 145.

148

D. A. JERRAM & V. M. MARTIN

Z ELLMER , G. F., A NNEN , C., C HARLIER , B. L. A., G EORGE , R. M. M., T URNER , S. P. & H AWKESWORTH , C. J. 2005. Magma evolution and ascent at volcanic arcs: constraining petrogenetic processes through rates and chronologies. Journal of Volcanology and Geothermal Research, 140, 171– 191. Z ELLMER , G. F., B LAKE , S., V ANCE , D., H AWKESWORTH , C. & T URNER , S. 1999. Plagioclase residence

times at two island arc volcanoes (Kameni Islands, Santorini, and Soufriere, St. Vincent) determined by Sr diffusion systematics. Contributions to Mineralogy and Petrology, 136, 345–357. Z ELLMER , G. F., T URNER , S. P. & H AWKESWORTH , C. J. 2000. Timescales of destructive plate margin magmatism: new insights from Santorini, Aegean volcanic arc. Earth and Planetary Science Letters, 174, 265–281.

Evidence for a short-lived stratified magma chamber: petrology of the Z-To5 tephra layer (c. 5.8 ka) at Zao volcano, NE Japan MASAO BAN, HIYORI SAGAWA, KOTARO MIURA & SHIHO HIROTANI Department of Earth and Environmental Sciences, Faculty of Science, Yamagata University, 1-4-12, Kojirakawa-machi, Yamagata 990-8560, Japan (e-mail: [email protected]) Abstract: Volcanic rocks from the Z-To5 tephra layer of Zao volcano, NE Japan, preserve petrological information that reflects the magmatic processes under the volcano. The Z-To5 rocks were formed by the mixing of three magmas that differed in composition and phenocryst assemblage: basalt (1150–1200 8C), with high Mg (Foc. 81) olivine; basaltic andesite (1020– 1100 8C), with Mg-rich orthopyroxene (Mg# ¼ c. 78) and clinopyroxene (Mg# ¼ c. 78), lower Mg olivine (Mg# ¼ c. 78), and calcic (Anc. 85) plagioclase; and andesite (900–1000 8C) with Mg-poor orthopyroxene (Mg# ¼ 61– 66) and clinopyroxene (Mg# ¼ 64– 68), and An-poor plagioclase. The basaltic magma was formed through fractionation of Foc. 85 olivine from a less differentiated basaltic magma during its fast ascent from the depths. The andesitic magma, which occupied a shallow magma chamber, was heated by underplating of the basaltic magma, resulting in dissolution of some minerals. Subsequently, the basaltic andesite magma was formed by mixing of the basaltic and andesitic magmas in the chamber. Petrological evidence for the rapid growth of phenocrysts in the basaltic andesite magma suggests that the magma residence time was short. The basaltic andesite magma, mixing with a small portion of the andesitic magma, was withdrawn upon eruption. The rates of these processes are inferred to be rapid based on petrological considerations.

In general, mature island arc volcanism is characterized by calc-alkaline andesitic magma (e.g. Gill 1981), and in many cases it can be shown that andesitic magma is formed by mixing between mafic and felsic end-member magmas at a shallow level (e.g. Alaska-Aleutian: Myers et al. 2002; George et al. 2004; Costa Rica: Hannah et al. 2002; Peru: Gerbe & Thoret 2004; New Zealand: Schmitz & Smith 2004; Indonesia: Reubi & Nicholls 2005; Philippines: Vogel et al. 2006). The buoyant felsic magmas are stored in shallow magma chambers, where they may be modified by multiple replenishments of mafic magmas (e.g. Wiebe 1994; Browne et al. 2006). Using this general model, recent petrological studies have aimed to reveal the detailed processes operating in magma reservoirs, such as the timing and nature of magma interactions (e.g. Chertkoff & Gardner 2004). However, not all mixed rocks provide useful information about these processes, and hence it is necessary to focus on those rocks that possess petrological information for shallow level magma processes. NE Japan is a well-known mature volcanic arc. Many petrological studies (e.g. Hayashi 1986; Ban & Yamamoto 2002; Toya et al. 2005; Ban et al. 2007) have shown that most of the calc-alkaline andesites from stratovolcanoes in NE Japan were formed through magma mixing/mingling processes involving two contrasting end-member magma compositions. With a few noteworthy exceptions

(e.g. Sakuyama & Koyaguchi 1984; Hirotani & Ban 2006), these studies did not deal with detailed shallow-level magmatic processes. We have identified several rock samples, which have petrological and mineral chemical characteristics that allow us to examine shallow-level magmatic processes in Zao volcano, and lead us to suggest the past existence of a short-lived stratified magma chamber.

Geological outline of Zao volcano Zao volcano is situated in the central part of the Quaternary volcanic front of the NE Japan arc (Fig. 1). Volcanic activity commenced at about 1 Ma (Takaoka et al. 1989), and has continued to the present day. Geological and petrological studies of the complete volcanic sequence were performed by Tiba (1961), Oba & Konta (1989) and Sakayori (1991, 1992). A simplified geological map of Zao volcano based on Sakayori (1992), who recognized four stages during the entire activity of Zao volcano, is shown in Figure 2a. The youngest stage of Zao volcano began at about 30 ka, and numerous small to medium-sized eruptions have occurred since then. At about 30 ka, the horseshoe-shaped Umanose caldera (1.7 km in diameter) was formed. A small cone, Goshikidake, formed subsequently in the inner part of the caldera. The Crater Lake Okama is located in the cone.

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 149–168. DOI: 10.1144/SP304.8 0305-8719/08/$15.00 # The Geological Society of London 2008.

150

M. BAN ET AL.

Fig. 1. (a) Simplified tectonic map of Japan showing the locality of NE Japan. The lines indicate tectonic boundaries. (b) Distribution of volcanic regions in NE Japan (after Umeda et al. 1999), and location of Zao and Sannomegata volcanoes.

The Goshikidake pyroclastics are the main products of the youngest stage, and can be further divided into the Komakusadaira agglutinate (c. 31–18 ka), Umanose agglutinate (c. 7.5–4.1 ka) and Goshikidake pyroclastic rock (c. 2.0 ka to present) (Ban et al. 2005; see Fig. 2b). The

Komakusadaira and Umanose agglutinates consist of more than 14 and four layers, respectively. The Goshikidake pyroclastic unit is mainly made up of pyroclastic surge deposits and can be divided into five units. The detailed geological descriptions of the past c. 30 ka eruptive products will be presented

SHORT-LIVED STRATIFIED MAGMA CHAMBER

151

Fig. 2. (a) Geological sketch map and stratigraphy of Zao volcano after Sakayori (1992), partly modified. GS, Goshikidake; NG, Nigorigawa; YK, Yokokurayama; HP, Happozawa; KN, Kumanodake; IM, Ichimaiishizawa; SK, Sainokawara; NM, Nakamaruyama; SN, Senninzawa; HM, Hiyamizuyama; TK, Torikabutoyama; ZO, Zaozawa; OW, Oiwake; MY, Mayuyamazawa; RO, Robanomimiiwa. (b) Geological sketch map of the summit area of Zao volcano. The outcrop in Figure 3 is at Point A.

elsewhere. Imura (1999) recognized 10 tephra layers for the youngest stage of Zao, and Ban et al. (2005) re-examined the tephra stratigraphy and recognized 16 layers. Six of them (Z-To5a, 5b, 5, 6, 7 and 8) were formed during c. 7.5– 4.1 ka eruptions. The main subject of this study is the products of the Z-To5 tephra layer, whose age is estimated to be c. 5.8 ka (Ban et al. 2005). We will compare these data with those for samples from Z-To6 and 7, the ages of which are estimated to be c. 5.3 and 4.5 ka respectively.

Geological features of the Z-To5, 6 and 7 tephra layers The Z-To5 tephra deposits are mainly distributed in an area southward to eastward from the Crater Lake Okama. They reach a maximum thickness approximately 1.5 km southeastward from Okama (Fig. 3a), where a lower pyroclastic surge deposit, 30 cm in thickness, is covered by black and red scoria fall deposits, which are respectively 45 and 40 cm in thickness. These scoria fall deposits consist mostly of lapilli-sized scoriae, with or without andesitic lithic fragments. Alternation of thin scoria fall and coarse ash layers can be found between the two scoria fall deposits. We collected scoria samples from the lower surge layer and upper scoria fall deposits. The scoriae from the red fall deposit are altered by oxidation and are not suitable for bulk rock chemical analyses. Even

the black scoriae usually have weathered crusts, thus we only used the least-altered inner parts of the black scoriae for bulk rock chemical analyses. For the upper, middle and lower part samples of the black fall deposits, component and grain size distribution analyses were carried out (Fig. 3a). The lithic fragments were observed only in the samples from the upper part, where the percentage was c. 5%. The grain analyses show the uni-modal distribution for all three samples, and the grain size peak position increases slightly toward the upper part. The Z-To6 and 7 tephra layers, which are composed of coarse ash, successively cover the Z-To5 tephra, intercalating loam and palaeosol layers in the distribution area of the Z-To5. In the case of the Z-To6 and 7, we were able to identify a proximal agglutinate facies around the western rim of the Umanose caldera (Fig. 2b). Samples for chemical analyses were collected from this facies.

Analytical methods Whole-rock major element and trace elements (Rb, Sr, Ba, Zr, Y, Nb and V) concentrations were determined by X-ray fluorescence analysis with a Rigaku RIX2000 spectrometer in Yamagata University. Operating conditions were 50 kV accelerating voltage and 50 mA current. The preparation method of the glass disks and the calibration method for major elements followed Yamada

152

M. BAN ET AL.

Fig. 3. (a) A photograph of the outcrop at point A in Figure 2b, showing the different facies of Z-To5 tephra. Z-To6 and 7 are also indicated. Z-To7 is scattered within the deposit. (b) Results of the component analysis and frequency histograms showing grain size distributions of black-coloured fall deposits from the Z-To5 tephra.

et al. (1995). The calibration method for the trace elements followed the background method of Yamada et al. (1995). The standards used in the analyses are the GSJ (Geological Survey of Japan) igneous rocks series. Analytical uncertainties for XRF trace elements are ,5% for Nb, Zr, Y, Sr, Rb and Ni; ,10% for V and Cr; and 5–15% for Ba. The range of uncertainties for a single element is based on the concentration range observed in standards. Chemical analyses of minerals and glasses from representative rocks were performed with a JEOL JXA8600M wavelength-dispersive type electron probe microanalyser in Yamagata University, using natural and synthetic minerals as standards. Operating conditions were a 15 kV accelerating voltage, a beam current of 10 nA (plagioclase and glass) or 20 nA (olivine, pyroxene and oxide minerals), and an 8– 20 s counting time for each element, except Ni and Ca in olivine, and MgO contents in plagioclase. In these two cases, the accelerating voltage and counting time were 25 kV and 90 s, and 15 kV and 90 s, respectively. All analyses were corrected using the oxide ZAF method.

in Table 1 and photomicroscopic images of the Z-To5 and 7 are shown in Figure 4. These samples show porphyritic textures, with phenocryst contents ranging from 25 to 30%. Scoriae from Z-To5 are basaltic andesites with olivine, clinopyroxene, orthopyroxene, plagioclase and Fe –Ti oxide as phenocrysts, while those from Z-To6 and 7 are basaltic andesites with +olivine, clinopyroxene, orthopyroxene, plagioclase and Fe –Ti oxide. The groundmasses are mainly composed of glass with plagioclase, orthopyroxene and clinopyroxene microlites, and various amounts of vesicles.

Petrography of the scoria samples from the Z-To5, 6 and 7 tephra layers

Z-To5 (n ¼ 11) c. 20 c. 4 c. 1 c. 2 ,1 Z-To6 and 7 (n ¼ 2) c. 25 ,2 ,0.2 ,0.1 ,1

The phenocrystic assemblage of the samples from the Z-To5, 6 and 7 tephra layers are summarized

The Z-To5 tephra layer The most abundant type of plagioclase phenocrysts in the Z-To5 tephra is the oscillatory zoning type with or without a dusty zone (Tsuchiyama 1985) Table 1. Phenocryst assemblage of Z-To5, 6 and 7 Phenocrysts (vol%) plg

opx

cpx

olv

opq

.2000 points were counted for analyses. n, number of samples; plg, plagioclase; opx, orthopyroxene; cpx, clinopyroxene; olv, olivine; opq, Fe –Ti oxides.

SHORT-LIVED STRATIFIED MAGMA CHAMBER

153

Fig. 4. Photomicrographs of thin sections from the Z-To5 (a) and Z-To7 (b) tephra layers. Orthopyroxene phenocryst in the Z-To5 sample have clearly defined cores and mantles. plg, plagioclase; olv, olivine; opx, orthopyroxene; cpx, clinopyroxene.

in the outer part of the crystals (Fig. 5a & b). The degree of the oscillation varies. This type usually has moderate amounts of glass inclusions in the core part, which sometimes constitutes patchy zoning (Vance 1965). In more extreme cases the crystals have a honeycomb texture (Fig. 5c).

Overgrowths of clear crystals are common in this phenocryst type. Clear-type phenocrysts are rarely observed and show normal zoning, occasionally with a dusty zone in the outer part (Fig. 5d). The modal amount of plagioclase phenocrysts is c. 20%.

154

M. BAN ET AL.

Fig. 5. Back scattered electron images of plagioclase phenocrysts in thin sections from Z-To5 samples showing: oscillatory zoned (a, b), honeycomb textured (c) and clear (d) crystals. Numbers indicate anorthite contents.

Orthopyroxene phenocrysts in Z-To5 can be divided into two types: one has a thick mantle (Fig. 4a), while the other does not. The core of the first type is made up of Mg-poorer orthopyroxene with a minor amount of glass inclusions, while the mantle consists of Mg-richer orthopyroxene with abundant small glass inclusions and many fractures perpendicular to the crystal surface. The boundary between the core and mantle is resorbed. The second type (up to 1.5 mm) is anhedral to subhedral, and some grains have an embayed form. Glass inclusions (up to 70 mm) are sometimes observed. The modal amount of orthopyroxene phenocrysts is c. 4%, with the two types being about equally abundant. Clinopyroxene phenocrysts (up to 3.2 mm) are usually anhedral to subhedral, and some grains have an embayed form. Glass inclusions (up to 100 mm) are sometimes observed. The modal amount of clinopyroxene phenocrysts is c. 1%. Olivine phenocrysts (up to 0.9 mm) in the Z-To5 samples are subhedral to anhedral in shape, and some show dendritic texture. Tiny Cr-spinel crystals are included in some olivine phenocrysts.

Glass inclusions (less than 35 mm) are sometimes observed. The modal amount of olivine phenocrysts is c. 2%. Fe–Ti oxides are mostly equant-shaped titanomagnetite. The size is usually less than 0.1 mm; however 0.3 mm grains are rarely observed. Ilmenite grains (up to 0.2 mm) are very rarely found.

The Z-To6 and 7 tephra layers Oscillatory zoned plagioclase is the dominant type of feldspar phenocrysts in the Z-To6 and 7 tephra layers. Glass inclusions are usually observed in their cores, which, as noted in Z-To5, can exhibit patchy zoning, while dusty zones are occasionally observed in the outer part. Honeycomb texture crystals are rarely found, and there are few crystals of the clear type. The modal amount of plagioclase phenocrysts is c. 25%. Orthopyroxene phenocrysts in Z-To6 and 7 are subhedral to euhedral (up to 2.5 mm). Glass inclusions (up to 40 mm) are sometimes observed. The modal amount of the orthopyroxene phenocryst is less than 2%.

SHORT-LIVED STRATIFIED MAGMA CHAMBER

Clinopyroxene phenocrysts (up to 1 mm) are usually anhedral to subhedral, and some grains have embayed forms. Glass inclusions (up to 30 mm) are sometimes observed. The modal amount of clinopyroxene phenocryst is less than 0.2%. Olivine phenocrysts (up to 0.2 mm) are rarely found in Z-To6 and 7; these are subhedral to anhedral in shape. Fe– Ti oxides are mostly equantshaped titano-magnetite. The size is up to 0.2 mm.

155

type plagioclase phenocrysts show a core composition of An78 – 86. Rarely observed clear type phenocrysts have a core composition higher than An90, and show compositional normal zoning. The groundmass core compositions fall in the range of An69 – 72. The Mg# of the core of mantled orthopyroxene phenocrysts is 61 –65 with the mantle parts having an Mg# of 76– 80. The Mg# of the core of unmantled orthopyroxene is 61 –66 and a reverse zoning (up to Mg# ¼ 80) is occasionally observed around c. 20 mm inside the rim. The Mg# of the groundmass core is between 74 and 77. The Mg# of the core of most of the clinopyroxene phenocrysts is 64 –68. Reverse zoning (up to Mg# ¼ 80) is usually observed around c. 20 mm inside the rim. Smaller sized, Mg-rich (Mg# ¼ 76) clinopyroxene phenocrysts are very rarely found. The Mg# of the groundmass clinopyroxene is between 74 and 77. The core composition of olivine phenocrysts has two peaks in Mg# (around 78 and 84–85). The high-Mg phenocrysts show normal zoning and are relatively large in size, whereas the low-Mg phenocrysts do not show compositional zoning and are smaller in size (,150 mm). The Mg# of the groundmass olivine is between 75 and 77.

Mineral chemistry of the scoria samples from the Z-To5, 6 and 7 tephra layers Representative compositional data for plagioclase, orthopyroxene, clinopyroxene, olivine and Fe– Ti oxides are given in Tables 2 and 3. Distribution histograms of the anorthite content or the Mg# of phenocrysts of those minerals are presented in Figure 6.

The Z-To5 tephra layer The oscillatory zoned plagioclase phenocrysts show compositional variation of their core within the range of An63 – 81, with peak positions around An64 and An69 – 76. In contrast, the honeycomb texture

Table 2. Representative chemical compositions of plagioclase phenocrysts Plagioclase Unit Type SiO2 Al2O3 Fe2O3 MgO CaO Na2O K2O Total An mol%

Z-To5

SiO2 Al2O3 Fe2O3 MgO CaO Na2O K2O Total An mol%

Z-To7

os

os

os

os

os

os

os

os

os

os

os

os

52.41 29.02 0.82 0.09 12.69 3.82 0.18 99.04 65

52.74 29.81 0.78 0.10 12.92 3.78 0.16 100.29 65

51.84 29.10 1.03 0.10 12.87 3.57 0.18 99.27 67

51.11 29.46 0.82 0.11 13.37 3.39 0.16 98.56 69

50.58 29.90 0.94 0.09 13.86 3.10 0.14 98.59 71

50.24 30.38 0.98 0.10 14.10 2.98 0.12 98.89 72

49.52 30.45 0.95 0.09 14.66 2.96 0.08 98.71 73

49.84 31.28 0.92 0.08 15.17 2.65 0.10 100.02 76

51.66 30.11 0.89 0.07 13.89 3.66 0.14 99.59 67

49.03 31.78 1.03 0.08 15.82 2.61 0.07 98.69 77

54.02 29.24 1.03 0.11 12.90 4.10 0.20 99.47 63

49.58 30.46 1.10 0.10 15.10 2.92 0.10 98.95 74

Z-To6

Z-To7

cl

cl

Unit Type

Z-To6

Z-To5

Z-To6

hc

hc

hc

48.43 31.80 0.94 0.13 15.73 2.02 0.16 99.23 81

47.63 32.68 0.69 0.12 16.62 1.90 0.06 99.70 83

47.80 31.64 1.04 0.14 16.52 1.81 0.06 99.01 83

hc

hc

46.56 48.17 33.29 31.95 0.81 0.95 0.12 0.12 17.28 16.69 1.56 2.07 0.06 0.07 99.68 100.03 86 81

Z-To7 hc

46.78 32.02 0.80 0.10 17.02 1.62 0.04 98.38 85

hc

hc

47.31 48.35 32.29 32.78 1.03 1.08 0.13 0.09 16.64 17.18 1.96 1.90 0.06 0.03 99.42 101.41 82 83

An, anorthite; os, oscillatory zoning type; hc, honeycomb texture type; cl, clear type.

Z-To5 cl 45.01 33.98 0.80 0.04 17.72 0.90 0.03 98.48 91

cl

44.97 46.09 45.37 34.32 34.75 34.80 0.67 0.85 0.68 0.05 0.04 0.07 18.36 18.59 19.04 0.76 1.01 0.94 0.02 0.01 0.02 99.16 101.35 100.91 92 91 92

156

Table 3. Representative chemical compositions of orthopyroxene, clinopyroxene, olivine and Fe – Ti oxides phenocrysts and Cr-spinel inclusions in olivine Unit

Orthopyroxene

Clinopyroxene

Z-To5 C

M

53.12 53.02 56.25 0.29 0.36 0.18 1.03 1.03 0.94 20.64 22.07 14.12 0.41 0.44 0.34 22.66 21.57 27.76 1.71 2.08 2.00 — — — 99.85 100.57 101.59 66 64 78

M

R

54.89 0.13 1.10 12.65 0.34 28.28 1.74 — 99.13 80

54.76 0.22 1.13 14.53 0.37 26.68 1.88 — 99.56 77

Unit

R

C

R

C

Z-To5 R

Olivine

C 38.51 20.61 0.32 41.14 0.19 0.12 100.89 78

C

C

38.57 39.00 19.55 17.52 0.31 0.26 41.12 43.03 0.21 0.19 0.10 0.15 99.85 100.14 78 81

Z-To7 C

C

C

C

Z-To6 R

54.32 53.11 53.79 53.37 54.74 51.84 51.92 51.86 0.18 0.35 0.28 0.34 0.23 0.66 0.55 0.54 1.46 1.14 1.49 1.03 1.14 2.05 2.18 3.60 13.86 21.36 16.48 22.52 17.63 12.75 12.33 9.04 0.32 0.41 0.36 0.41 0.45 0.35 0.38 0.32 27.76 21.93 26.11 21.09 25.28 14.32 14.39 16.55 1.91 1.83 1.71 1.84 1.93 18.28 18.29 19.05 — — — — — 0.21 0.28 0.21 99.81 100.12 100.22 100.60 101.40 100.46 100.32 101.17 78 65 74 63 72 67 68 77

Z-To5

SiO2 FeO MnO MgO CaO NiO Total Fo

Z-To7

C

C

C

39.18 39.19 39.61 38.73 38.68 13.82 14.29 13.80 20.12 19.97 0.20 0.20 0.20 0.32 0.31 45.45 46.84 46.06 40.64 41.47 0.16 0.16 0.15 0.18 0.18 0.22 0.25 0.23 0.14 0.16 99.03 100.93 100.04 100.13 100.77 85 85 86 78 79

Mg#, 100  Mg/(Mg þ Fe); Fo, forsterite; C, core; R, rim; M, mantle. Fe2O3, Xusp and Xilm for Fe– Ti oxides are calculated after Stormer (1983).

C TiO2 Al2O3 Fe2O3 Cr2O3 V2O3 FeO MnO MgO Total Xusp

R

C

51.97 53.06 0.47 0.62 3.06 1.83 8.15 12.31 0.19 0.28 16.16 14.62 19.02 18.82 0.25 0.23 99.27 101.76 78 68

Z-To7 R

C

R

52.75 51.85 52.98 0.46 0.63 0.45 2.42 2.04 1.81 10.39 13.89 11.23 0.37 0.46 0.34 14.32 14.19 15.19 18.39 18.10 18.92 0.18 0.26 0.21 99.28 101.43 101.13 71 65 71

Titanomagnetite

Ilmenite

Spinel

Z-To5

Z-To5

Z-To5

C

C

14.43 15.65 17.97 TiO2 2.67 2.48 1.99 Al2O3 1.24 1.23 1.18 Fe2O3 0.40 0.21 0.16 Cr2O3 38.15 36.25 32.53 V2O3 40.50 42.14 43.05 FeO 0.37 0.43 0.40 MnO 2.67 2.37 3.04 MgO 100.42 100.76 100.31 Total 0.43 0.47 0.52 Xilm

C

C

45.68 46.48 TiO2 0.06 0.09 Cr2O3 12.99 12.81 Al2O3 0.03 0.07 FeO 0.48 0.56 Fe2O3 36.54 36.83 MnO 1.73 1.65 NiO 1.52 1.80 MgO 99.02 100.29 Total 0.87 0.87 Cr#

C

C

0.46 3.23 29.69 22.89 26.43 12.18 16.14 22.78 13.60 32.81 0.24 0.31 0.19 0.15 12.73 7.06 99.49 101.42 0.43 0.56

M. BAN ET AL.

SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O Total Mg#

C

Z-To6

SHORT-LIVED STRATIFIED MAGMA CHAMBER

157

Fig. 6. Histograms of anorthite content of plagioclase, and of Mg# of orthopyroxene, clinopyroxene and olivine phenocrysts for the Z-To5, 6 and 7 samples. An, anorthite; Mg#, 100  [Mg/(Mg þ Fe)]; opx, orthopyroxene; cpx, clinopyroxene; olv, olivine; plg, plagioclase.

The Cr# of the Cr-spinel inclusions in high-Mg olivine phenocrysts is 0.41–0.46, whereas that in low-Mg olivine phenocrysts is 0.47– 0.58. The proportion of ulvo¨spinel molecules (Xusp) in most of the titano-magnetite phenocrysts falls in the range 0.35– 0.53, and the proportion of ilmenite molecules (Xilm) in ilmenites is around 0.87.

overgrowth can be seen just inside the rim in most grains. The Mg# of groundmass orthopyroxenes is around 68. The Mg# of the core of most of the clinopyroxene phenocrysts is 64 –69, and that in the groundmass is 65 –70. The Mg# of the olivine phenocrysts is 78 –80. The Xusp in most of the titano-magnetite phenocrysts is 0.35 –0.41.

The Z-To6 and 7 tephra layers

The compositions of glass inclusions in phenocrysts from the Z-To5 tephra layer

The oscillatory zoned plagioclase phenocrysts display a wide compositional variation in the range of An62 – 82, with peak positions around An66 – 67 and An74 – 75. An-rich phenocrysts tend to have larger amounts of glass inclusions. The honeycomb textured plagioclase phenocrysts have a core composition of An76 – 85. Clear phenocrysts have a composition of An91 – 92. The groundmass core composition is An65 – 72. The Mg# of the core of orthopyroxene phenocrysts is 62 –69 with reverse zoning (up to Mg# ¼ 74) at c. 20 mm inside the rim. Fe-richer

The compositions of glass inclusions (.20 mm) in phenocrysts were analysed by electron microprobe using a 10 mm defocused beam. Representative data are presented in Table 4 and a plot of K2O v. SiO2 contents is shown in Figure 7. The compositional range of glass inclusions from Mg-poor pyroxenes phenocrysts is different from that observed in Mg-rich pyroxenes and low-Mg olivine phenocrysts. Glass inclusions in An64 – 75 plagioclase

158

M. BAN ET AL.

Table 4. Representative chemical compositions of glass inclusions in phenocrysts in Z-To5 Glass in plagioclase g-C

g-C

g-A

Glass in orthopyroxene g-A

g-C

g-C

g-A

Glass in clinopyroxene g-A

wt% 55.88 57.33 65.92 66.37 55.51 57.88 65.14 66.80 SiO2 TiO2 1.01 0.87 1.20 1.07 1.18 1.30 1.06 1.07 Al2O3 15.87 14.58 14.22 14.36 16.73 15.96 13.99 16.22 FeOT 9.26 9.99 6.93 7.09 11.11 10.43 7.68 5.11 MnO 0.24 0.14 0.31 0.15 0.26 0.08 0.15 0.23 MgO 5.01 4.94 1.62 1.47 2.33 1.85 1.21 0.78 CaO 9.80 8.42 5.25 5.13 9.74 8.78 5.44 4.43 Na2O 2.00 2.56 2.36 2.17 2.40 2.42 3.24 3.20 K2O 0.82 1.04 2.03 2.02 0.63 1.07 1.92 2.01 P2OS 0.10 0.12 0.17 0.18 0.12 0.24 0.18 0.15 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

g-C

57.51 1.18 17.30 10.33 0.17 3.04 7.05 2.48 0.85 0.09 100.00

g-C

g-A

Glass in olivine g-A

58.03 65.42 68.46 1.42 1.42 1.24 16.61 13.46 13.08 11.31 8.82 6.00 0.18 0.12 0.15 1.97 1.13 1.75 6.86 5.85 4.15 2.14 1.95 2.30 1.27 1.65 2.63 0.21 0.16 0.23 100.00 100.00 100.00

g-C*

g-C*

59.74 60.29 1.81 1.47 13.12 16.07 10.27 7.47 0.19 0.13 2.14 1.81 9.33 9.00 1.70 2.12 1.56 1.40 0.15 0.24 100.00 100.00

g-A, glass in An-poor plagioclase or Mg-poor pyroxene phenocryst; g-C, glass in honeycomb texture-type plagioclase or Mg-rich pyroxene phenocryst; g-C*, glass in Mg-low olivine phenocryst; FeOT, total iron calculated as FeO; Total, normalized value to 100% volatile free.

phenocrysts of the oscillatory zoning type are compositionally similar to those of the Mg-poor pyroxenes, whereas those in the honeycomb texture type are similar to those found in Mg-rich pyroxenes and low-Mg olivines.

Whole rock compositions of the scoria samples from the Z-To5, 6 and 7 tephra layers Whole rock analyses of Z-To5, 6 and 7 samples as well as Komakusadaira agglutinate and

Goshikidake pyroclastic rock samples are given in Table 5. In this paper, all analyses have been normalized to 100% volatile free with total iron (FeO*) calculated as FeO. The Z-To5 –7 sequence belongs to the medium-K calc-alkaline series according to the classification scheme of Gill (1981) (Fig. 8). Major oxide and trace element variation diagrams are shown in Figure 8. Data from stages 1–3 of Zao volcano (Sakayori 1992) and data from the youngest stage units (the Komakusadaira agglutinate and Goshikidake pyroclastic rock) are included in Figure 8 for comparative purposes.

Fig. 7. K2O v. SiO2 variation diagram for glass inclusions in phenocrysts from the Z-To5 tephra layer. opx, orthopyroxene; cpx, clinopyroxene; olv, olivine; plg, plagioclase.

Table 5. Whole rock compositions of Z-To5, 6 and 7 as well as Goshikidake pyroclastic rock and Komakusadaira agglutinate Sample no.

Z-To5

Z-To6

2

3

4

5

wt% SiO2 TiO2 Al2O3 Fe2OT3 MnO MgO CaO Na2O K2O P2O5 Total

55.44 0.90 16.59 10.03 0.15 5.58 8.39 2.35 0.80 0.11 100.34

55.19 0.88 16.46 10.10 0.15 5.61 8.47 2.37 0.82 0.10 100.15

55.38 0.89 16.41 9.98 0.15 5.79 8.20 2.36 0.82 0.10 100.08

55.57 0.90 16.41 9.99 0.15 5.79 8.22 2.40 0.84 0.10 100.37

ppm Ba Rb Sr Zr Nb Y V Cr Ni

193 19 209 78 2.3 20 183 155 46

232 24 254 95 2.9 24 229 187 55

224 23 259 93 2.6 24 208 182 53

242 24 246 93 3.0 25 220 187 57

Sample no.

b 2 a*

a*

1

2

3

1

2

3

55.59 0.90 16.21 10.06 0.16 5.88 8.18 2.35 0.84 0.10 100.27

50.00 0.90 17.20 13.11 0.20 8.20 9.80 1.50 0.20 0.09 101.20

56.00 0.89 17.00 9.89 0.16 4.90 8.60 2.45 0.90 0.10 100.89

62.00 0.87 16.80 6.67 0.11 1.60 7.40 3.40 1.60 0.11 100.56

57.90 0.89 16.93 9.18 0.15 4.57 8.23 2.74 1.00 0.11 101.70

57.96 0.88 17.21 8.88 0.14 4.30 8.52 2.79 1.00 0.11 101.79

58.16 0.89 17.25 8.99 0.14 4.31 8.21 2.78 1.02 0.11 101.86

58.29 0.89 16.72 9.05 0.14 3.95 7.96 2.75 1.14 0.11 101.00

58.03 0.87 17.06 8.74 0.14 3.94 8.21 2.78 1.09 0.11 100.97

58.76 0.88 17.27 8.66 0.13 3.92 8.12 2.82 1.11 0.11 101.78

240 24 250 94 2.9 24 224 186 58

30 5 255 45 2.3 14 270 170 66

233 24 278 95 3.0 25 220 88 35

435 43 300 145 3.7 36 170 5 4

289 26 280 106 3.0 27 215 66 30

293 28 288 106 3.4 26 204 65 28

286 28 290 107 2.9 27 211 66 27

313 32 277 114 3.4 28 197 56 25

297 31 283 110 3.4 27 198 57 24

318 30 285 114 3.4 29 206 54 25

Goshikidake pyroclastic rock 1

wt% SiO2 TiO2 Al2O3 Fe2OT3 MnO

b*

55.86 0.92 16.66 9.83 0.15

2 55.83 0.89 16.82 9.60 0.15

3 56.13 0.91 17.01 9.50 0.15

4 55.86 0.91 16.31 9.93 0.16

5 56.96 0.95 17.08 9.23 0.14

Komakusadaira agglutinate 6 56.88 0.94 16.94 9.29 0.14

7 56.68 0.94 17.22 9.18 0.15

8 57.12 0.95 17.04 9.18 0.14

1 55.82 0.87 17.22 9.46 0.15

2 55.68 0.86 17.51 9.33 0.15

3 55.74 0.87 17.18 9.54 0.16

4 55.23 0.88 17.70 9.70 0.16

5

SHORT-LIVED STRATIFIED MAGMA CHAMBER

1

Z-To7

6

55.46 0.87 17.24 9.73 0.16

55.42 0.88 17.42 9.69 0.16 159

(Continued)

160

Table 5. Continued Sample no.

Goshikidake pyroclastic rock

Komakusadaira agglutinate

2

3

4

5

6

7

8

1

2

3

MgO CaO Na2O K2O P2O5 Total

4.83 8.48 2.50 0.89 0.13 100.25

4.79 8.80 2.45 0.85 0.13 100.31

4.49 8.54 2.60 0.92 0.13 100.38

4.99 8.71 2.52 0.89 0.13 100.41

4.00 8.14 2.81 1.12 0.14 100.57

4.01 7.99 2.70 1.06 0.13 100.08

4.08 7.91 2.69 1.08 0.14 100.07

3.89 7.88 2.66 1.02 0.14 100.02

4.72 8.60 2.48 0.85 0.12 100.29

4.62 8.59 2.63 0.86 0.12 100.35

4.72 8.57 2.64 0.86 0.12 100.40

ppm Ba Rb Sr Zr Nb Y V Cr Ni

276 25 299 96 2.8 26 217 57 33

264 24 301 93 2.9 25 208 55 31

265 26 304 98 3.0 25 229 49 30

272 26 293 94 3.0 25 216 55 32

339 32 318 113 3.1 28 205 35 23

319 30 304 112 3.3 28 197 36 23

331 31 308 112 3.6 27 210 35 23

326 29 307 111 3.5 27 205 38 23

271 24 303 93 3.2 24 222 61 28

265 24 306 93 3.2 23 216 59 28

284 23 312 92 2.9 25 219 61 28

Fe2OT3 , total iron calculated as Fe2O3.

4 4.73 8.24 2.47 0.75 0.12 99.98 266 23 314 89 3.2 24 224 57 29

5

6

4.84 8.63 2.62 0.80 0.12 100.47

4.90 8.43 2.51 0.78 0.12 100.31

256 21 311 89 2.7 24 223 59 27

274 22 317 89 2.8 23 219 58 28

M. BAN ET AL.

1

SHORT-LIVED STRATIFIED MAGMA CHAMBER 161

Fig. 8. SiO2 variation diagrams showing abundances of major oxides and trace elements in whole-rock samples from the Z-To5, 6 and 7 tephra layers. See details in the text for the estimated compositions of the andesitic end-member and the olivine accumulation-corrected Z-To5, and for the eruptive stages of Sakayori (1992). Thin lines are linear trends defined by the Z-To6 and 7 samples. The boundaries defining the low-K and medium-K fields in the K2O v. SiO2 diagram are from Gill (1981), and that between TH (tholeiitic) and CA (calc-alkalic) fields in the FeOT/MgO v. SiO2 diagram is from Miyashiro (1974). FeOT is total iron calculated as FeO.

162

M. BAN ET AL.

Bulk silica contents of the youngest units are 55.1– 58.1%, while those of the older stages show a wider range (52–65%). Looking at the compositions of the youngest stage more closely, the bulk SiO2 contents of samples from the Goshikidake pyroclastic rock are 56.2– 58.1%, which is slightly higher than those of the Komakusadaira agglutinate (55.1–56.2%). The bulk SiO2 contents of Z-To5– 7 samples are c. 55.8, 57.2 and 58%, respectively, and are not distinguishable from those of the Komakusadaira agglutinate and the Goshikidake pyroclastic rock, but samples from Z-To5– 7 have higher MgO, Cr and Ni and lower in Al2O3, Ba and Sr abundances than those from the above two units (Fig. 8). It should be noted that the MgO, Cr and Ni contents of the Z-To5 samples are higher than any data reported from Zao volcano so far.

Discussion As in many cases of calc-alkaline andesites in island arc settings, the Z-To5 samples contain petrological evidence for magma mixing, such as the dissolution textures of An-poor plagioclase, reverse zoning of Mg-poor pyroxene phenocrysts, and a wide range of plagioclase, orthopyroxene, clinopyroxene and olivine phenocryst compositions. The Z-To6 and 7 samples also show some of these features. In the following discussion, we will characterize the magmas involved in the formation of Z-To5, and will then examine relationships among the magmas.

Three magmas for the Z-To5, 6 and 7 tephra layers As described above, phenocrystic mafic minerals in the Z-To5 samples show compositional variations. Based on the iron–magnesium exchange equilibrium between olivine and calcium-free pyroxene (Matsui & Nishizawa 1974), and between olivine and calcium-rich pyroxene (Obata et al. 1974), phenocrystic mafic minerals can be divided into the following three coexisting equilibrium groups. Group A includes Mg-poor orthopyroxene (Mg# ¼ 61–66) and clinopyroxene (Mg# ¼ 64–68), group B consists of high-Mg olivine (Mg# ¼ 81–86), and group C has Mg-rich orthopyroxene (Mg# ¼ c. 80), clinopyroxene (Mg# ¼ c. 80) and low-Mg olivine (Mg# ¼ c. 78). Based on the glass inclusion data, the oscillatory zoned and honeycomb textured plagioclase phenocrysts can be assigned to groups A and C, respectively. The clear plagioclase phenocrysts would belong to group B. The Fe –Ti oxide phenocrysts would belong to group A. The Mg# of groundmass mafic minerals is 74 (75)–77, which is slightly lower than that of group C pyroxenes.

In some cases, variations in phenocryst mineral chemistry are interpreted to reflect changes in liquidus phases. In this case, however, mineral disequilibrium textures provide evidence for magma mixing, which suggests that minerals in each group precipitated from compositionally distinct melts. The chemical compositions of phenocrysts are such that A, B and C were derived from andesitic, basaltic and basaltic andesite magmas, respectively. These three magmas would be involved in the mixing process to produce the Z-To5 tephra. Intermediate chemical composition of group C phenocrysts suggests that the basaltic andesite magma was formed by the mixing of basaltic and andesitic end-member magmas. With respect to the Z-To6 and 7 samples, the phenocrysts can be grouped in the same way, although the Z-To6 and 7 samples lack Mg-rich orthopyroxene, clinopyroxene and high-Mg olivine phenocrysts. Thus the three magmas would also have been involved in mixing to form the Z-To6 and 7 products.

Petrological features of the andesitic end-member magmas in the Z-To5, 6 and 7 tephra layers The Z-To6 and 7 samples define a linear trend on most of the SiO2 variation diagrams (Fig. 8), and the group A phenocrysts have similar compositional features in Z-To6 and 7. Thus the andesitic endmember magmas inferred for Z-To6 and 7 would be similar. In terms of Z-To5, the compositional features of group A phenocrysts are similar to those in Z-To6 and 7, thus the andesitic endmember magma of the Z-To5 samples would also have a similar whole rock composition to that of Z-To6 and 7. Each of the group A phenocrysts shows some degree of compositional variation, which we interpreted to have resulted from the repeated injections of basaltic magma into the andesitic magma chamber (e.g. Landi et al. 2004; Koyaguchi & Kaneko 2000), so that the chamber would have been somewhat heterogeneous (Kuritani et al. 2003). We have calculated a representative bulk rock composition (see Table 5) of the andesitic endmember magma for Z-To5 to be on the extension the linear trends, which satisfies the criteria of iron-magnesium exchange equilibrium between clinopyroxene and liquid (Sisson & Grove 1993), assuming Mg# ¼ 68 for the clinopyroxene and an NNO buffer condition. Using a two-pyroxene thermometer (Brey & Ko¨hler 1990), the deduced magmatic temperature from core compositions of group A orthopyroxene and clinopyroxene phenocrysts is 900–1000 8C.

SHORT-LIVED STRATIFIED MAGMA CHAMBER

Using the iron–titanium oxide thermobarometer (Stormer 1983), the magmatic temperature and oxygen fugacity indicated by the coexisting magnetite-ulvo¨spinel and ilmenite-haematite solid solutions in the Z-To5 samples are calculated to be 900 –970 8C, and the fo2 is slightly lower than the nickel–nickel oxide (NNO) buffer. The temperature range obtained is thus consistent with that calculated by the two-pyroxene thermometer applied to group A phenocrysts.

Petrological features of basaltic end-member magmas for the Z-To5, 6 and 7 tephra layers In MgO, Al2O3, CaO, Ni, Cr and Sr variation diagrams (Fig. 8), the Z-To 5 samples do not plot on the extension of the linear trends defined by the Z-To6 and 7 samples. This is believed to reflect the effect of the olivine accumulation. The modal amount of olivine phenocrysts in the Z-To5 samples is c. 2%, and about half of this is high-Mg olivine. The Mg# of the high-Mg olivine shows some variation of 81–86 with a peak in 84– 85 (Fig. 6). The olivine with Mg# higher than 82 is believed to have crystallized earlier and would have an accumulative origin. Consequently, we have calculated the accumulation-corrected compositions for Z-To5. The calculated compositions do plot on the extension of the linear trends observed in Figure 9, except in the case of the Cr diagram. The difference observed in the Cr diagram could arise from the effect of Cr-spinel inclusions in the olivine. When 0.02% Cr-spinel is taken into account, the calculated Cr content comes close to the extension of the linear trend. Thus, it is likely that the compositions of basaltic end-member magma for Z-To5 lie on the silica poor extension of the linear trends defined by the Z-To6 and 7 samples. We have calculated the bulk rock compositions of the basaltic end-member magma for Z-To5 to be on the extension of the linear trends and also to satisfy the criteria of the iron-magnesium exchange equilibrium between olivine and liquid (e.g. Roeder & Emslie 1970; Sugawara 2000). For this calculation, compositions of the Mg-poorest olivine (Fo81) among those in group B were used, and the ferric –ferrous ratios were calculated using the method of Sack et al. (1980), assuming an NNO buffer condition. The obtained composition (see Table 5) is that of a low-alkali basalt. Further we estimated the basaltic magma compositions which crystallized the Fo85 olivine, adding the olivine (Fo85) compositions to the estimated basaltic end-member compositions until the melt compositions satisfied the criteria of the iron– magnesium exchange equilibrium between olivine

163

and liquid. The result shows a less differentiated basaltic character (MgO ¼ 10.5%, Ni ¼ 170 ppm). Most of the basalts in NE Japan are differentiated to some degree, because the c. 30 km thickness of crust under compressional stress prevents the primary magmas from ascending to the surface. Thus, the MgO and Ni contents of these basalts are usually less than 9.5% and 100 ppm, respectively, although recently a primitive basalt (MgO ¼ 10.37%, Ni ¼ 184 ppm) was reported from the Sannome-gata volcano (Yoshinaga & Nakagawa 1999), with similar to MgO and Ni contents to those calculated using the data for Z-To5. In Figure 9, NiO and MnO v. Fo diagrams for olivine phenocrysts, and the Cr# v. Fo diagram for coexisting Cr-spinel and olivine phenocrysts from Z-To5 are presented. The mantle olivine arrays are from Takahashi (1986) and Arai (1987). It is clear from these diagrams that, although the basaltic magma in equilibrium with the olivine (Fo85) is less differentiated, it is not primitive. Precise compositional data for forsterite rich olivines have not been reported before from Quaternary volcanoes in NE Japan, with the exception of Sannome-gata volcano situated in the rear arc. We note that the source mantle for the basaltic magma of Z-To5 is more fertile than that of the Sannome-gata volcano, which would support the notion of Yamamoto & Abe (1991) that the upper mantle under the volcanic front is more fertile than that under the rear arc. Takagi et al. (2005) showed experimentally that An c. 90 plagioclase crystallizes from low-alkali basaltic melt at an H2O saturated pressure of 2 –3 kbar, while An c. 85 plagioclase forms under pressures of ,4 kbar and 2.5–4.5% H2O. Taking into account these experimental results, we assumed that the H2O content in the basaltic endmember magma is 2.5% in order to estimate its temperature and pressure conditions using the MELTS algorithm of Ghiorso and Sack (1995). The calculations show that olivine of Fo82 is the only liquidus phase in the range 1160 –1180 8C at less than 3 kbar. An-rich plagioclase is on the liquidus when the temperature is 70–90 8C lower. It is therefore probable that the An-rich plagioclase would have crystallized slightly later than the olivine phenocrysts.

Petrological features of the basaltic andesite magmas for the Z-To5, 6 and 7 tephra layers Honeycomb textures observed in plagioclase phenocrysts in group C can be explained by growth in a rapidly cooling melt (e.g. Kawamoto 1992), as can the textures observed in the orthopyroxene mantles. The smaller grain size of the group C olivine and clinopyroxene phenocrysts may also

164

M. BAN ET AL.

Fig. 10. (a) MgO v. An content diagram of plagioclase phenocrysts, and (b) CaO v. Fo diagram of olivine phenocrysts from Z-To5. plg, plagioclase; olv, olivine.

Fig. 9. (a, b) NiO and MnO v. Fo diagrams for group B olivine phenocrysts from Z-To5. (c) Cr# [Cr/(Cr þ Al)] v. Fo diagram for spinel in group B olivine phenocrysts from Z-To5. Mantle olivine arrays in (a, b) are from Takahashi (1986). The compositional ranges of Sannome-gata volcano are after Yoshinaga & Nakagawa (1999). OSMA (olivine-spinel mantle array) and the broken lines with percentages in (c) are from Arai (1987). The percentages indicate the degree of partial melting based on experimental data (Jaques & Green 1980), with Tinaquillo peridotite minus 40% olivine as starting material. The arrows indicate Fo-Cr# fractionation trends. Open circles are from the Sannome-gata volcano after Yoshinaga & Nakagawa (1999).

indicate a shorter timescale for their growth. Plagioclase and olivine phenocrysts in group C have higher MgO and CaO contents than in the other groups (Fig. 10), which would be due to

disequilibrium partitioning caused by rapid growth of these phenocrysts. Evidence for rapid growth coupled with the intermediate chemical compositions of group C phenocrysts suggests that the basaltic andesite was formed by the mixing of basaltic and andesitic end-member magmas shortly prior to being withdrawn to the conduit. We have calculated the bulk rock compositions of the basaltic andesite magma, using the same method applied to the calculation of the basaltic end-member magma. For this calculation, we used Fo78 as olivine composition. The obtained composition is listed in Table 5, which is similar to that of the accumulationcorrected compositions for Z-To5 with 56% in SiO2 content. Using a two-pyroxene thermometer (Brey & Ko¨hler 1990), the magmatic temperature calculated from compositions of group C orthopyroxene and clinopyroxene phenocrysts is 1020 –1100 8C.

SHORT-LIVED STRATIFIED MAGMA CHAMBER

Inferred magmatic processes for the c. 5.8 ka activity It has been inferred that mixing in arc magmas with large compositional contrasts is promoted by forced convection. The proposed classical model is based on vertically arranged double chamber systems (e.g. McBirney 1980; Sakuyama 1984). When mafic magma from the deeper chamber is injected into the shallower felsic one, forced mixing would have taken place and would have continued during the ascent of these magmas through the vent to the surface. Since this model was first proposed, there have been advances in our understanding of upper crustal magmatic processes. It has, for example, been proposed (Koyaguchi & Kaneko 2000) that the felsic chamber is composed of crystal–melt mixtures with variable melt fractions, i.e. that only a part of it behaves as a liquid while the rest behaves as a solid crystal mush. The volume of the liquid part would be increased by heat derived from mafic magma that ascends from depth and underplates the felsic chamber. Repeated replenishment of the mafic magma into the felsic chamber results in the eruption of mixed magma. Two principle processes were proposed to operate upon replenishment of mafic magma. One possibility is that felsic and mafic magmas mixed during the

165

injection of the mafic magma into the chamber, and that the mixed magma subsequently erupted (e.g. Nakamura 1995; Browne et al. 2006). In this case, an intermediate magma is likely to have formed before the eruption (e.g. Reagan et al. 1987; Izbekov et al. 2004; Chertkoff & Gardner 2004). The other possibility is that the mafic magma acts as a heat source to increase the melt proportion of the shallower felsic chamber (e.g. Murphy et al. 2000; Tamura & Tatsumi 2002; Devine et al. 2003). A cartoon of the magmatic processes envisaged to have taken place during the evolution and eruption of Z-To5 is presented in Figure 11. In the case of Z-To5, andesitic magma would have been stored in a shallow-level magma chamber, which had experienced replenishments of mafic magmas that resulted in the compositional diversity observed in group A phenocrysts. We infer that the embayed forms of some Mg-poor pyroxenes, the resorbed textures of the cores of mantled orthopyroxenes and the patchyzoning of An-poor plagioclases indicate that dissolution of these minerals has occurred in the andesitic magma when it was heated by the underplating of basaltic end-member magma. The basaltic endmember magma had already been differentiated and it precipitated Fo81 olivine crystals at less than 3 kb; however, it must have risen rapidly

Fig. 11. Schematic representation of the magmatic processes involved in the generation of Z-To5. An, anorthite; Mg#, 100  [Mg/(Mg þ Fe)]; opx, orthopyroxene; cpx, clinopyroxene; olv, olivine; plg, plagioclase.

166

M. BAN ET AL.

from depth because it entrained early fractionated Foc. 85 olivine as an accumulated phase. At this stage, the chamber would have consisted of upper andesitic (heated to unheated parts) and lower basaltic layers (Fig. 11a). Subsequently, the injected basaltic magma mixed with the andesitic magma, and formed the basaltic andesite magma, in which the Mg-rich rims and mantles of the orthopyroxene crystals grew (Fig. 11b). The textures and chemical compositions of the group C minerals suggest that the basaltic andesite magma was short-lived. Subsequently, the basaltic andesite magma mixed with the andesitic magma during withdrawal from the chamber to the conduit. Consequently, the embayed Mg-poor orthopyroxene and the mantled orthopyroxene coexisted in the mixed magma (Fig. 11c). Judging from the similarity of estimated SiO2 contents for the basaltic andesite magma and the accumulation-corrected Z-To5, the portion of the andesitic magma involved in this mixing should have been small. During the ascent of the mixed magma through the conduit and after the eruption at the surface, the groundmass minerals, which show slightly differentiated chemical compositions than those of the group C phenocrysts, precipitated from the mixed magma. A similar scenario can be inferred for the magmatic processes that led to the formation of Z-To6 and 7, although the opportunity for examining the processes is much better in the case Z-To5. Diffusion profiles of some elements in the phenocrystic minerals are useful for estimating the residence time of crystals in a magma chamber following a mixing event. Recent studies (e.g. Nakamura 1995; Zellmer et al. 1999, 2003; Gioncada et al. 2005), which have applied this tool on arc volcanoes, indicate that crystal residence times must be short, ranging from months to centuries. Our results are consistent with these studies. At Zao volcano, the intermediate magma appears to have been generated contemporaneously with the replenishment of the mafic magma.

Conclusions 1.

Petrological examination shows that volcanic rocks from the Z-To5 tephra layer (c. 5.8 ka) at Zao volcano, NE Japan, were formed by the mixing of three magmas. These are: basalt with high-Mg olivine (Mg# ¼ 81); basaltic andesite with Mg-rich orthopyroxene (Mg# ¼ c. 80), clinopyroxene (Mg# ¼ c. 80), low-Mg olivine (Mg# ¼ c. 78) and Anc. 85 plagioclase; and andesite with Mg-poor orthopyroxene (Mg# ¼ 61 –66) and clinopyroxene (Mg# ¼ 64 –68) and An-poor plagioclase as phenocrysts.

2.

3.

4.

Estimated temperatures for the basaltic, basaltic andesite and andesitic magmas are 1150– 1200, 1020–1100 and 900–1000 8C, respectively. The fo2 of the andesitic endmember magma was slightly lower than that of the nickel– nickel oxide buffer. The depth at which Fo81 olivine precipitated from the basaltic end-member magma is estimated to be shallow (less than 3 kb). The basaltic end-member magma differentiated from a more primitive magma through crystal fractionation of Foc. 85 olivine. Foc. 85 olivine is included in the basaltic end-member magma as an accumulated phase, suggesting that the less differentiated magma rose from depth very fast, differentiating on the way and forming the basaltic end-member magma, which subsequently underplated the shallow andesitic chamber. Underplated basaltic magma heated the lower part of the andesitic magma chamber. Subsequently, mixing of the basaltic and a proportion of the andesitic end-member magmas formed the basaltic andesite magma. Evidence for rapid crystal growth is recorded in the textures and chemical compositions of phenocrysts of the basaltic andesite magma, suggesting that this magma was short-lived. When the basaltic andesite magma was withdrawn to the conduit, it was mixed with a small portion of the original andesitic magma.

We express our hearty thanks to G. Zellmer and C. Annen for providing us with an opportunity to present the results of our study and for their strong editing. We are grateful to R. Price and D. Pyle for many constructive comments and suggestions on the manuscript and to R. W. Jordan for correcting the English in this paper.

References A RAI , S. 1987. An estimation of the least depleted spinel peridotite on the basis of olivine–spinel mantle array. Neues Jahrbuch fuer Mineralogie, Monatshefte, 8, 347–354. B AN , M. & Y AMAMOTO , T. 2002. Petrological study of Nasu-Chausudake Volcano (ca. 16 ka to present), northeastern Japan. Bulletin of Volcanology, 64, 100–116. B AN , M., H IROTANI , S. ET AL . 2007. Origin of silicic magmas in a large-caldera-related stratovolcano in the central part of NE Japan – petrogenesis of the Takamatsu volcano. Journal of Volcanology and Geothermal Research, 167, 100– 118. B AN , M., S AGAWA , H., M IURA , K. & T ANAKA , Y. 2005. Volcanic hazard map of Mt. Zao. Earth Monthly, 27, 317–320 (in Japanese). B REY , G. & K O¨ HLER , T. 1990. Geobarimetry in four phase lherzolites II. New thermobarometers, and

SHORT-LIVED STRATIFIED MAGMA CHAMBER practical assessment of existing thermobarometers. Journal of Petrology, 31, 1353– 1378. B ROWNE , B. L., E ICHELBERGER , J. C., P ATINO , L. C., V OGEL , T. A., D EHN , J., U TO , K. & H OSHIZUMI , H. 2006. Generation of porphyritic and equigranular mafic enclaves during magma recharge events at Unzen volcano, Japan. Journal of Petrology, 47, 301–328. C HERTKOFF , D. G. & G ARDNER , J. E. 2004. Nature and timing of magma interactions before, during, and after the caldera-forming eruption of Volcan Ceboruco, Mexico. Contributions to Mineralogy and Petrology, 146, 715– 735. D EVINE , J. D., R UTHERFORD , M. J., N ORTON , G. E. & Y OUNG , S. R. 2003. Magma storage region processes inferred from geochemistry of Fe–Ti oxides in andesitic magma, Soufriere Hills volcano, Montserrat, W.I. Journal of Petrology, 44, 1375– 1400. G EORGE , R., T URNER , S., H AWKESWORTH , C., B ACON , C. R., N YE , C., S TELLING , P. & D REHER , S. 2004. Chemical versus temporal controls on the evolution of tholeiitic and calc-alkaline magmas at two volcanoes in the Alaska– Aleutian arc. Journal of Petrology, 45, 203–219. G ERBE , M. C. & T HOURET , J. C. 2004. Role of magma mixing in the petrogenesis of tephra erupted during the 1990–98 explosive activity of Nevado Sabancaya, southern Peru. Bulletin of Volcanology, 66, 541–561. G HIORSO , M. S. & S ACK , R. O. 1995. Chemical mass transfer in magmatic processes. IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid– solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology, 119, 197– 212. G ILL , J. B. 1981. Orogenic Andesites and Plate Tectonics. New York, Springer. G IONCADA , T. A., M AZZUOLIA , R. & M ILTONB , A. J. 2005. Magma mixing at Lipari (Aeolian Islands, Italy): insights from textural and compositional features of phenocrysts. Journal of Volcanology and Geothermal Research, 145, 197– 118. H ANNAH , R. S., V OGEL , T. A., P ATINO , L. C., A LVARADO , G. E., P E´ REZ , W. & S MITH , D. R. 2002. Origin of silicic volcanic rocks in Central Costa Rica: a study of a chemically variable ash-flow sheet in the Tiribı´ Tuff. Bulletin of Volcanology, 64, 117–133. H AYASHI , S. 1986. Petrology of Chokai volcano. Part III – Trace elements and petrogenesis. Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists, 81, 370– 383 (in Japanese with English abstract). H IROTANI , S. & B AN , M. 2006. Origin of silicic magma and magma feeding system of Shirataka volcano, NE Japan. Journal of Volcanology and Geothermal Research, 156, 229– 251. I MURA , R. 1999. Zao volcano. In: T AKAHASHI , M. & K OBAYASHI , T. (eds) Volcanoes in the Tohoku Region: Field Guide of Japanese Volcano 4. Tsukijishokan, Tokyo, 70–88 (in Japanese). I ZBEKOV , P. E., E ICHELBERGER , J. C. & I VANOV , B. V. 2004. The 1996 eruption of Karymsky volcano,

167

Kamchatka: Historical record of basaltic replenishment of an andesite reservoir. Journal of Petrology, 45, 2325–2345. J AQUES , A. L. & G REEN , D. H. 1980. Anhydrous melting of peridotite at 0–15 kb pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology, 73, 287–310. K AWAMOTO , T. 1992. Dusty and honeycomb plagioclase: indicators of processes in the Uchino stratified magma chamber, Izu Peninsula, Japan. Journal of Volcanology and Geothermal Research, 29, 413 –450. K OYAGUCHI , T. & K ANEKO , K. 2000. Thermal evolution of silicic magma chambers after basalt replenishments. Transactions of the Royal Society of Edinburgh: Earth Sciences, 91, 47–60. K URITANI , T., Y OKOYAMA , T., K OBAYASHI , K. & N AKAMURA , E. 2003. Shift and rotation of composition trends by magma mixing: 1983 eruption at Miyake-jima Volcano, Japan. Journal of Petrology, 44, 1895–1916. L ANDI , P., M ETRICH , A. N., B ERTAGNINI , A. & R OSI , A. M. 2004. Dynamics of magma mixing and degassing recorded in plagioclase at Stromboli (Aeolian Archipelago, Italy). Contributions to Mineralogy and Petrology, 147, 213–227. M ATSUI , Y. & N ISHIZAWA , O. 1974. Iron (II)– magnesium exchange equilibrium between olivine and calcium-free pyroxene over a temperature range 800 to 1300 8C. Bulletin de la Societe Francaise de Mineralogie et de Cristallographie, 97, 122– 130. M C B IRNEY , A. R. 1980. Mixing and unmixing of magmas. Journal of Volcanology and Geothermal Research, 7, 357–371. M IYASHIRO , A. 1974. Volcanic rock series in island arcs and active continental margins. American Journal of Science, 274, 321– 355. M URPHY , M. D., S PARKS , R. S. J., B ARCLAY , J., C ARROLL , M. R. & B REWER , T. S. 2000. Remobilization of andesitic magma by intrusion of mafic magma at the Soufriere Hills volcano, Montserrat, West Indies. Journal of Petrology, 41, 21–42 M YERS , J. D., M ARSH , B. D., F ROST , C. D. & L INTON , J. A. 2002. Petrologic constraints on the spatial distribution of crustal magma chambers, Atka Volcanic Center, central Aleutian arc. Contributions to Mineralogy and Petrology, 143, 567– 586. N AKAMURA , M. 1995. Continuous mixing of crystal mush and replenished magma in the ongoing Unzen eruption. Geology, 23, 807– 810. O BA , Y. & K ONTA , T. 1989. Geology and petrology of central Zao volcano, Yamagata prefecture. Bulletin of Yamagata University, Natural Sciences, 12, 199– 210 (in Japanese with English abstract). O BATA , M., B ANNO , S. & M ORI , T. 1974. The ironmagnesium partitioning between naturally occurring coexisting olivine and Ca-rich clinopyroxene: an application of the simple mixture model to olivine solid solution. Bulletin de la Societe Francaise de Mineralogie et de Cristallographie, 97, 101– 107. R EAGAN , M. K., G ILL , J. B., M ALAVASSI , E. & G ARCIA , M. 1987. Changes in magma composition at Arenal volcano, Costa Rica, 1968– 1985: real-time monitoring of open-system differentiation. Bulletin of Volcanology, 49, 415–434.

168

M. BAN ET AL.

R EUBI , O. & N ICHOLLS , I. A. 2005. Structure and dynamics of a silicic magmatic system associated with caldera-forming eruption at Batur Volcanic Field, Bali, Indonesia. Journal of Petrology, 46, 1367– 1391. R OEDER , P. L. & E MSLIE , R. F. 1970. Olivine– liquid equilibrium. Contributions to Mineralogy and Petrology, 29, 275–289. S ACK , R. O., C ARMICHAEL , I. S. E., R IVERS , M. & G HIORSO , M. S. 1980. Ferric–ferrous equilibria in natural silicate liquids at 1 bar. Contributions to Mineralogy and Petrology, 75, 369– 376. S AKAYORI , A. 1991. Magmatic evolution of Zao volcano, Northeast Japan. Bulletin of the Volcanological Society of Japan, 36, 79–92 (in Japanese with English abstract). S AKAYORI , A. 1992. Geology and petrology of Zao volcano. Journal of Mineralogy, Petrology and Economic Geology, 87, 433–444 (in Japanese with English abstract). S AKUYAMA , M. 1984. Magma mixing and magma plumbing systems in island arcs. Bulletin of Volcanology, 47, 685–703. S AKUYAMA , M. & K OYAGUCHI , T. 1984. Magma mixing in mantle xenolith-bearing calc-alkalic ejecta, Ichinomegata volcano, northeastern Japan. Journal of Volcanology and Geothermal Research, 22, 199– 224. S CHMITZ , M. D. & S MITH , I. E. M. 2004. The petrology of the Rotoiti eruption sequence, Taupo Volcanic Zone: an example of fractionation and mixing in a rhyolitic system. Journal of Petrology, 45, 2045–2066. S ISSON , T. W. & G ROVE , T. L. 1993. Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology, 113, 143– 166. S TORMER , J. C. 1983. The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multicomponent iron– titanium oxides. American Mineralogist, 66, 1189–1201. S UGAWARA , T. 2000. Empirical relationships between temperature, pressure, and MgO content in olivines and pyroxene saturated liquid. Journal of Geophysical Research, 105, 8457– 8472. T AKAGI , D., S ATO , H. & N AKAGAWA , M. 2005. Experimental study of a low-alkali tholeiite at 1– 5 kbar: optimal condition for the crystallization of high-An plagioclase in hydrous arc tholeiite. Contributions to Mineralogy and Petrology, 149, 527–540. T AKAHASHI , E. 1986. Genesis of calc-alkali andesite magma in a hydrous mantle–crust boundary: petrology of lherzolite xenoliths from the Ichinomegata crater, Oga peninsula, northeast Japan, part II. Journal of Volcanology and Geothermal Research, 29, 355 –395. T AKAOKA , N., K ONNO , K., O BA , Y. & K ONTA , T. 1989. K– Ar datings of lavas from Zao volcano, north-eastern Japan. Journal of the Geological Society of Japan, 95, 157– 170 (in Japanese with English abstract). T AMURA , Y. & T ATSUMI , Y. 2002. Remelting of an andesitic crust as a possible origin for rhyolitic

magma in oceanic arcs: an example from the Izu– Bonin arc. Journal of Petrology, 43, 1029– 1047. T IBA , T. 1961. Petrology of Zao volcano. Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists, 46, 74– 81 (in Japanese with English abstract). T OYA , N., B AN , M. & S HINJO , R. 2005. Petrology of Aoso volcano, northeast Japan arc: temporal variation of the magma feeding system beneath the Aoso volcano and nature of low-K amphibole andesite in the Aoso– Osore volcanic zone. Contributions to Mineralogy and Petrology, 148, 566– 581. T SUCHIYAMA , A. 1985. Dissolution kinetics of plagioclase in melt of the system diopside– albite–anorthite, and origin of dusty plagioclase in andesite. Contributions to Mineralogy and Petrology, 89, 1– 16. U MEDA , K., H AYASHI , S., B AN , M., S ASAKI , M., O HBA , T. & A KAISHI , K. 1999. Sequence of the volcanism and tectonics during the last 2.0 million years along the volcanic front in Tohoku district, NE Japan. Bulletin of the Volcanology Society of Japan, 44, 233 –249 (in Japanese with English abstract). V ANCE , J. A. 1965. Zoning in plagioclase: Patchy zoning. Journal of Geology, 73, 636–651. V OGEL , T. A., F LOOD , T. P. ET AL . 2006. Geochemistry of silicic magmas in the Macolod Corridor, SW Luzon, Philippines: evidence of distinct, mantle derived, crustal sources for silicic magmas. Contributions to Mineralogy and Petrology, 151, 267– 281. W IEBE , R. A. 1994. Silicic magma chambers as traps for basaltic magmas: the Cadillac Mountain intrusive complex, Mount Desert Island, Maine. Journal of Geology, 102, 423 –437. Y AMADA , Y., K OHNO , H. & M URATA , M. 1995. A low dilution fusion method for major and trace element analysis of geological samples. Advances in X-ray Analysis 26, 33– 44 (in Japanese with English abstract). Y AMAMOTO , M. & A BE , E. 1991. Residual and initial mantle for Quaternary basalts of northeast Japan arc. Bulletin of the Volcanology Society of Japan, 36, 149–159 (in Japanese with English abstract). Y OSHINAGA , A. & N AKAGAWA , M. 1999. Finding of primary basalt from Sannome-gata volcano, northeastern Japan, and its compositional variation. Journal of Mineralogy, Petrology and Economic Geology, 87, 433– 444 (in Japanese with English abstract). Z ELLMER , G. F., B LAKE , S., V ANCE , D., H AWKESWORTH , C. J. & T URNER , S. 1999. Plagioclase residence times at two island arc volcanoes (Kameni islands, Santorini, and Soufriere, St. Vincent) determined by Sr diffusion systematics. Contributions to Mineralogy and Petrology, 136, 345–357. Z ELLMER , G. F., S PARKS , R. S. J., H AWKESWORTH , C. J. & W IEDENBECK , M. 2003. Magma emplacement and remobilization timescales beneath Montserrat: insights from Sr and Ba zonation in plagioclagse phenocrysts. Journal of Petrology, 44, 1413– 1431.

Uranium-series isotope and thermal constraints on the rate and depth of silicic magma genesis A. DOSSETO1, S. P. TURNER1, M. SANDIFORD2 & J. DAVIDSON3 1

GEMOC National Key Centre, Department of Earth and Planetary Sciences, Macquarie University, Sydney, NSW 2109, Australia (e-mail: [email protected]) 2

School of Earth Sciences, University of Melbourne, Melbourne VIC 3010, Australia

3

Department of Earth Sciences, University of Durham, South Road, Durham DH1 3LE, UK Abstract: Uranium-series isotopes provide important constraints on the timescale of magma differentiation and this can be used to identify where in the crust and silicic magmas acquire their geochemical characteristics. Timescales of differentiation can be inferred from the observed co-variations of U-series disequilibria with differentiation indexes. When crustal assimilation of secular equilibrium material is involved, inferred timescales will generally decrease. In turn, they will increase if periodical recharge (.20 wt% relative volume) of the magma body occurs. If crustal assimilation and magma recharge occur concurrently, inferred timescales for differentiation can be similar to that of closed system differentiation. We illustrate the approach with data from Mount St Helens which suggest that dacitic compositions are produced in c. 2000 years. Combining this with recent evidence for an important role for amphibole fractionation suggests that differentiation of a c. 10 km3 magma body at this volcanic centre occurs at 8– 10 km depth in the crust.

The genesis of silicic magmas is important for understanding the growth of the continental crust and the origin of explosive eruptions because the upper crust is dominated by silicic igneous rocks, and evolved lavas are responsible for many of the most dangerous volcanic eruptions. Over the last two decades, numerous efforts have been undertaken to constrain how and where silicic magmas are generated (e.g. Davidson 1985; Huppert & Sparks 1988; Bergantz 1989; Bergantz & Dawes 1992; Laube & Springer 1998; Petford & Gallagher 2001; Orozco-Esquivel et al. 2002; Annen & Sparks 2002; Annen et al. 2006). In particular, recent thermal modelling has shown that repetitive intrusions of basalt can melt pre-existing intrusions in the lower crust and that the combination of residual magmas from basalt crystallization and these crust-derived melts in a deep crustal hot zone could produce silicic magmas and account for observed compositional variations in silicic volcanic systems (Annen & Sparks 2002; Annen et al. 2006). However, the timescales of magma differentiation can also be used to constrain the depth where silicic magmas acquire their geochemical characteristics and thus provide an important test of these models. An important aspect of the model of Annen et al. (2006) is that, whilst silicic magmas can be produced rapidly as the residue of basalt crystallization, the timescales for large volumes of melt to accumulate by melting of the crust are very long (c. 100 ka).

One constraint upon the timescales and relative proportions of liquid derived from recently emplaced basalt v. crustal melt in these hot zone models is that most silicic arc magmas contain significant 230Th/238U and 226Ra/230Th disequilibria (Cooper & Reid 2003; Turner et al. 2003a, b for a recent compilation). These disequilibria are believed to be derived from slab fluids (Turner et al. 2001; Bourdon et al. 2003). Thus, in order to preserve radioactive disequilibria, significantly less than 8 ka (five half-lives of 226Ra) can elapse during the evolution from mantle-derived basalts to andesitic and dacitic compositions and their eruption. Moreover, many suites of rocks from individual volcanoes show a correlated decrease in (226Ra/230Th) with increasing differentiation (as measured by SiO2 or Th content). This has been taken to indicate that differentiation occurred over a timescale proportional to the half-life of 226Ra and detailed studies have used the change in (226Ra/230Th) to infer the timescale in detail (e.g. Turner et al. 2003a, b; George et al. 2004). In this case, the silicic rocks must derive their geochemical characteristics almost entirely from crystallization of a single sill of basalt. Consequently, there is a need for a model that accounts for both the rapid generation of silicic magmas (in order to preserve Ra –Th disequilibrium) and their geochemical and isotopic diversity and evidence for crustal assimilation in many instances (e.g. Smith & Leeman 1987; Grove et al. 1988; Tepper et al. 1993; Bourdon

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 169–181. DOI: 10.1144/SP304.9 0305-8719/08/$15.00 # The Geological Society of London 2008.

170

A. DOSSETO ET AL.

et al. 2000). Moreover, if andesites are directly derived from crystallization of a basaltic sill, this requires no contribution from residual melts of previously crystallized intrusion. A second issue is that hydrous basalt cooling in the mid- to lower crust will crystallize amphibole and partial melts of earlier intruded basalt will probably form in the presence of residual amphibole. Thus, it now appears that amphibole fractionation may play a more critical role in the compositional evolution of arc magmas than the gabbroic assemblages with which they typically erupt (Davidson et al. 2007). Radium can be moderately compatible in amphibole (Blundy & Wood 2003) and so there is also a need to appraise the effects of amphibole fractionation on 226Ra disequilibria and to reconsider how this may affect our inferences about the timescales of differentiation. Accordingly, we have explored two endmember models. In the first, the radioactive disequilibrium in andesites and dacites is largely derived from zero-aged basalt mixed with crustal and residual melts just before eruption, since mixing has the potential to be an important process for the production of some intermediate silicic rocks (Zellmer et al. 2005). In the second model, crustal and residual melts are mixed with a primary basaltic magma during crystallization (assimilationfractional crystallization). The role of amphibole fractionation during this assimilation-fractional crystallization has also been assessed.

Model I: mixing of zero-aged basalt with crustal and residual melts just before eruption One end-member model to account for the geochemical characteristics of andesitic and dacitic magmas involves mixing of high-silica melts with ‘zero-aged’ basalt that imparts mantle-derived U-series disequilibria to the hybrid magma. Eruption takes place less than a few hundreds of years after mixing in order that Ra –Th disequilibrium is preserved. The high-silica melts are derived from a combination of partial melting of the crust and the residues of crystallization of the most recent magmatic intrusions. Annen & Sparks (2002) suggested that high-silica melts are produced in a significant volume only after successive magma emplacements over .105 years. In this case, the crustal melts will be in secular equilibrium for the 238 U/230Th/226Ra systems. Secondly, these crustal melts are produced by partial melting of amphibolite (Annen et al. 2006). Berlo et al. (2004) have shown that partial melting of amphibolite is unlikely to produce significant radioactive disequilibrium and so any observed disequilibria must be

inherited from the ‘zero-aged’ basalt. Dufek & Cooper (2005) suggested that it might be possible to produce 226Ra excess during continuous dehydration melting of the lower crust. However, this model cannot account for the positive correlation between (226Ra/230 Th) and Sr –Th ratios often observed in arc lavas (Dosseto et al. 2003). In order to quantify the contribution of zero-aged basalt required in the mixing model described above, we use the average composition of subductionrelated andesites and dacites for which SiO2, U, 226 Ra contents, (238U/230Th) and (226Ra/230Th) data are available (n¼36) and calculate the proportion of zero-aged basalt and the composition of zero-aged basalt and high-silica melt end-member that best reproduce this average andesite/dacite composition. The range of possible values for the SiO2 and U contents of the high-silica melt is 59–73 wt% and 1.5–4.5 ppm, respectively (average upper continental crust; Taylor & McLennan 1995), whereas for the SiO2 content of the zero-aged basalt it is 48– 52 wt% (average composition of subduction-related basalts measured for 238U/230Th/226Ra; n¼51). The range of possible values for the U content of the basalt endmembers is 0.01–4 ppm, which is the range exhibited by subduction-related basalts for which combined 238U/230Th/226Ra data are available. Three models were considered. In model Ia, the range of possible values for the (238U/230Th) and (226Ra/230Th) activity ratios of the basalt is simply the average values for subduction-related basalts: 1.1 + 0.2 and 2 + 1, respectively (1 standard deviation). In model Ib, the (238U/230Th) and (226Ra/230Th) activity ratios of the basalt are extreme values corresponding to theoretical compositions for slab-derived fluids: 8 and 17, respectively (Turner et al. 2003a, b). In these two models, activity ratios for the high-silica end-member are taken equal to unity (secular equilibrium) as explained above. In model Ic, the parental basalt is the same as that of model Ia but the high-silica melt has been allowed to have some radioactive disequilibria: (238U/230Th)¼(226Ra/230Th)¼1.2, those being the maximum values calculated by Berlo et al. (2004) for an amphibolitic melt. Starting compositions of the zero-aged basalt and silicic melt are compiled in Table 1 and results of the inversion are shown in Table 2 and Figure 1. Inversion of the average arc andesite/ dacite composition yields a SiO2 composition of the zero-aged basalt and the crustal melt at the higher end of the allowed range of values for all three models. As a consequence, because mixing end-members always have the same SiO2 content regardless of their activity ratios, mixing proportions are controlled by the SiO2 content of the end-member magmas.

THE RATES OF SILICIC MAGMA PRODUCTION

171

Table 1. Range of values allowed for the two components in the mixing models Model 1 SiO2 high-silica melt (wt%) SiO2 zero-aged basalt (wt%) U high-silica melt (ppm) U zero-aged basalt (ppm) (238U/230Th) high-silica melt (238U/230Th) zero-aged basalt (226Ra/230Th) high-silica melt (226Ra/230Th) zero-aged basalt

59–73 48–52 1.5–4.5 0.01–4 1 0.9–1.3 1 1–3

Model 2 59 – 73 48 – 52 1–5 0.01 – 4 1 8 1 17

Model 3 59 – 73 48 – 52 1–5 0.01– 4 1.2 0.9– 1.3 1.2 1–3

See text for explanation of the chosen component compositions.

For all three models, the U-series composition of an average arc andesite/dacite can be explained by the mixture between c. 45% of pre-existing silicic crustal melt (produced through basalt crystallization and/or wallrock melting) and c. 55% of zero-age basaltic magma. Sparks & Marshall (1986) showed that complete magma hybridization is only possible with proportions of mafic magma greater than 50%, which is the case here. However, this model assumes magma mixing without any requirement for crystal fractionation to occur. This is at odds with the numerous studies which have observed that the geochemistry of silicic lavas is at least partially controlled by crystal fractionation (Grove & Kinzler 1986 and references therein). For instance, this process is unable to explain the curvilinear trends so often observed between the abundance of elements such as TiO2 or P2O5 and differentiation index (Turner et al. 2003a, b; George et al. 2004; Zellmer et al. 2005). Therefore, in the following section, we consider more realistic model of coupled assimilationfractional crystallization (AFC). The effect of crustal assimilation, magma recharge and amphibole fractionation on U-series-derived timescales for magma differentiation is then evaluated.

Model II: assimilation of crustal/residual melts during differentiation + magma recharge Assimilation of secular equilibrium materials will act to reduce (226Ra/230Th) ratios, mimicking the effects of time and associated radioactive decay. Consequently, the magma differentiation timescale inferred from a specific dataset will represent maxima unless assimilation (which must be independently established) is quantitatively taken into account. In this section, we investigate quantitatively the effect of crustal assimilation during differentiation on timescales inferred from Ra –Th data. To illustrate this, we use U-series data from Mount St Helens (Fig. 2; Volpe & Hammond 1991), where the operation of AFC has been demonstrated (Halliday et al. 1983). Note that the objective of this discussion is not to re-appraise previous U-series interpretations (Volpe & Hammond 1991; Cooper & Reid 2003). Calculations performed for other volcanoes (e.g. Ruapehu, New Zealand, Price et al. 2007; Sangeang Api, Indonesia) yield similar conclusions to that presented below. We focused on whole-rock Ra –Th data, as it has been shown

Table 2. Results of the numerical modelling of mixing processes

Xbasalt (%) SiO2 high-silica melt (wt%) SiO2 zero-aged basalt (wt%) U high-silica melt (ppm) U zero-aged basalt (ppm) (238U/230Th) high-silica melt (238U/230Th) zero-aged basalt (226Ra/230Th) high-silica melt (226Ra/230Th) zero-aged basalt Errors are given at the 1s level.

Model 1

Model 2

Model 3

55.6 + 0.1 72.94 + 0.03 51.97 + 0.01 1.500 + 0.006 0.52 + 0.03 1 1.3 1 2.9 + 0.2

54 + 5 73 + 1 51.9 + 0.9 1.9 + 0.2 0.39 + 0.04 1 8 1 17

55 + 2 73.0 + 0.9 51.6 + 0.7 1.69 + 0.06 0.39 + 0.05 1.2 0.90 + 0.01 1.2 2.3 + 0.1

172

A. DOSSETO ET AL.

Fig. 1. (238U/230Th) and (226Ra/230Th) v. SiO2 (in wt%) mixing diagrams. The composition of the different end-members is shown: in black, zero-age basalt; in light grey, silicic melt derived from lower crust melting. Compositions are shown for the three different models as explained in the text. Tick marks show proportions of zero-age basalt in 25% intervals. It can be seen that the silica composition of the end-members and the proportion of zero-age basalt required to explain the average composition of subduction-related andesites and dacites (dark grey area) do not vary significantly from one model to another (see text for details).

Fig. 2. (226Ra/230Th) activity ratio v. 1/Th (in ppm21) diagram. (a) Positive trends are commonly observed for arc lavas: Tonga (crosses; Turner et al. 2001); Mount St Helens (squares; Volpe & Hammond 1991); Ruapehu, New Zealand (dots; Price et al. 2007). (b) Schema showing the effect of crystallization at various rates and mixing in a (226Ra/230Th) v. 1/Th diagram.

that the U-series composition of minerals may be complex to interpret (Zellmer et al. 2000; Cooper & Reid 2003). The variation of the mass of magma in the chamber with time is:

where f is the crystallization rate (in year21) and Qassimil is the rate of wallrock assimilation (in g year21). We use the non-dimensional parameter Qassimil so that r defined by DePaolo (1981): r ¼ f M Equation 1 becomes

dM ¼ f  M þ Qassimil dt

dM ¼ f  M  (r  1): dt

(1)

(2)

THE RATES OF SILICIC MAGMA PRODUCTION

The variation with time of the mass of a nuclide in the magma, Mi, is dMi dM dCi ¼ Ci : þM dt dt dt ¼ r  f  M  Ci,assimil  f  Di  M  Ci þ li1  M  Ci1  li  M  Ci

(3)

where Ci and Ci21 are respectively the concentrations of the nuclide and its parent in the magma (in g g21), Ci,assimil the concentration of the nuclide in the assimilated wallrock (in g g21), Di the partition coefficient of the nuclide and li and li21 are respectively the decay constants of the nuclide and its parent (in year21). We can re-write Equation 3 so that: dCi ¼ r  f  Ci,assimil  Ci  f (r þ Di  1) dt þ li1  Ci1  li  Ci :

(4)

This equation can be written for each nuclide (238U, 230Th, 226Ra and 232Th) and integrated so the evolution of activity ratios with time can be calculated.

173

The initial composition of the magma is assumed to be similar to the most primitive composition observed: Thi ¼1.7 mg g21 and (226Ra/230Th)i ¼1.6. The composition of the wallrock assimilated is assumed to be: Thcrust ¼5 mg g21 and (226Ra/230 Th)crust ¼1. Bulk partition coefficients used are 1025 for Ra (Blundy & Wood 2003) and 1022 for Th (for a mineralogical assemblage containing 10% clinopyroxene, which controls Th partitioning; Bacon & Druitt 1988). If no assimilation occurs (r¼0), it is possible to reproduce the positive trend observed between (226Ra/230Th) and 1/Th for Mount St Helens and infer that it took about 2000–2500 years for the magma to evolve from the most primitive to the most evolved compositions (Fig. 3). Note that we aim to reproduce the broad trend defined by the data and not each sample composition separately. This timescale is within the range of crystallization ages obtained by Cooper & Reid (2003) for plagioclase (c. 2 –4 ka) and pyroxene (c. 0.15 – 5.7 ka) in the St Helens lavas. The positive trend defined by the data is best reproduced for a crystallization rate of c. 2–31024 year21, which is comparable to that inferred from similar data from a number of volcanic systems from a diverse

Fig. 3. Closed system fractionation models for Mt St Helens. Squares are wholerock data from Volpe & Hammond (1991). Curves show the evolution of magma composition for different crystallization rates, assuming no crustal assimilation. The magma is assumed to have the following starting composition: Th¼1.7 ppm; (226Ra/230Th)¼1.6. Partition coefficients used are: DTh ¼0.01 for Th and DRa ¼1025 for 226Ra. The dotted line shows the effect of instantaneous mixing before eruption between the mafic magma and a crustal melt, with tick marks showing proportions of crustal melt in 10% intervals. The following composition is assumed for the crustal melt: Th¼5 ppm; (226Ra/230Th)¼1.

174

A. DOSSETO ET AL.

range of tectonic settings (Blake & Rogers 2005). As shown in Figure 3, higher crystallization rates produce curves with more gentle slopes than the best fit to the observed trend. Figure 4 shows that, if assimilation of a secular equilibrium crustal melt accompanies crystal fractionation (AFC), the time inferred to produce the most differentiated magma is significantly reduced (in other words the ‘true’ time would be overestimated if assimilation was not taken into account). For a ratio of assimilation to crystallization rate (r) of 0.3, suggested to be the highest value yet inferred (Tegner et al. 2005), the true timescale for differentiation is reduced to c. 1500 years (Fig. 4). As discussed for model I above, in the extreme case of two-end-member mixing, the timescale for differentiation approaches zero. As the rate of assimilation increases relative to crystallization, the inferred timescale for differentiation required to account for the Ra –Th disequilibrium decreases. This is not to say that increasing assimilation leads to faster differentiation. Rather, the timescale inferred from U-series isotopes increasingly overestimates the true timescale as the amount of assimilation increases if this effect is not taken into account in the interpretation. Glazner (2007) suggested that wallrock assimilation would increase the crystallization rate. In this case, if r¼0.3 and the

crystallization rate, f, increases from 2 to 41024 year21 during AFC, the most silicic composition is produced after c. 2000 years (not shown) instead of c. 1500 years if f is kept constant and equal to 31024 year21.

The role of amphibole Davidson et al. (2007) have recently shown that the REE composition of arc volcanics indicates an important role for amphibole fractionation in generating the diversity of arc magma geochemistry. Radium can be relatively compatible in amphibole (Blundy & Wood 2003) and so fractionation of amphibole-rich assemblages (or the presence of amphibole as a residual phase during the production of partial melts in the lower crust) could also influence (226Ra/230Th) ratios. Thus, we have also investigated AFC models in which amphibole is a major fractionating phase. The partition coefficient for Ba in amphibole can be up to 0.7 (LaTourrette et al. 1995), and assuming that DRa ¼0.08DBa (Blundy & Wood 2003), the highest predicted DRa is 0.056. As Figure 5 shows, because Ra remains relatively incompatible even in amphiboledominated assemblages, the effect on the inferred timescales is relatively modest (,5%).

Fig. 4. The effect of crustal assimilation on the evolution of magma composition and comparison with Mount St Helens data. The two upper curves are calculated for different ratios of assimilation over crystallization rates, r (labels on curves), and assuming a crystallization rate infinite compared with 226Ra half-life. The plain and dashed curves are calculated for r¼0 and 0.3, respectively, and with a crystallization rate of 31024 year21 (see Fig. 3). Labels on curve are durations of differentiation. Same symbols, partition coefficients and composition as in Figure 3.

THE RATES OF SILICIC MAGMA PRODUCTION

175

Fig. 5. Dashed and plain curves are calculated on the assumption that amphibole controls 226Ra abundance in the melt. Even if amphibole is an important phase during magma evolution, it can be seen that it does not greatly affect the (226Ra/230Th) of the magma. The composition for the crystallizing magma and the crustal melt are as in Figures 2 & 3. Tick marks on curves are at 500 year intervals.

The effect of recharge The models above assume no recharge of the magma chamber. Hughes & Hawkesworth (1999) have shown that this process can buffer the decay of 226Ra. In a model of assimilation-rechargefractional crystallization (ARFC), a primitive magma undergoes fractional crystallization with or without synchronous wallrock assimilation until a new batch of magma is added to the differentiating magma. New batches of magma are added periodically and mixed instantaneously with the resident magma. An equivalent mass of magma is erupted subsequent to this mixing in order to maintain a constant magma volume. Between each recharge event, fractional crystallization occurs with or without wallrock assimilation. Figure 6 shows that the periodic injection of a mass of primitive magma equivalent to 10% of the total mass of magma in the chamber every 500 years slightly increases the time required to produce silicic magmas. In the case of Mount St Helens, the inferred timescale for differentiation increases from c. 2000–2500 years (no recharge) to c. 2500–3000 years (with recharge) (Fig. 6). A significant increase in the inferred timescales would require larger volumes of replenishing magma. If more frequent recharge is invoked, a steady-state composition with a relatively low Th concentration

would be reached and the composition of the most evolved sample cannot be reproduced. It is worth noting that, if crustal assimilation and magma recharge occured concurrently, the resulting timescale for differentiation could be similar to that obtained for the simplest case where closed-system differentiation occurs (i.e. the effects of recharge and assimilation have opposite effects upon disequilibria). In the example of Mount St Helens, assuming the magma recharge scenario and crustal assimilation at a rate 10 times slower that of crystallization (i.e. r¼0.1), the time required to produce the most silicic compositions is c. 2500 years, similar to the timescale inferred when neither recharge or crustal assimilation occur.

Physical implications The timescale for differentiation inferred in the previous sections can, in principle, be used to place physical constraints on the depth of differentiation for a given geotherm and magma chamber volume and shape. For Mount St Helens (and other volcanoes such as Ruapehu and Sangeang Api), because differentiation and cooling appear to be fast (,2.5 ka), the magma body is likely to be small and/or located at shallow depths. This can be quantitatively constrained using a cooling

176

A. DOSSETO ET AL.

Fig. 6. Magma recharge increases the inferred timescale for differentiation. In the case of Mount St Helens (squares), whereas it takes c. 2000–2500 years to produce the most silicic compositions without magma recharge (plain curve), it takes c. 2500– 3000 years when 10 wt% of primitive magma is added every 500 years (bold curve). The composition of the primitive magma considered is: [Th]¼1.7 ppm; (226Ra/230Th)¼1.6. The crystallization rate between each magma addition is 31024 year21. Tick marks on curves are at 500 year intervals. Same partition coefficients as in Fig. 3.

model. We model the crystallization of a cooling magma body emplaced instantaneously at a temperature Tint, into isothermal crust at a temperature Twr. Our crystallization model is an analytical fit to the crystallization model compiled by Hawkesworth et al. (2000) (Fig. 7). We express the crystal mass fraction rate as   (Tint  Ttrans )2 exp  DT 2 pffiffiffiffiffiffiffiffiffiffiffiffi pDT 2 where the initial temperature of the magma body, Tint ¼1200 8C (Yoder & Tilley 1962), the temperature at which 50% crystallization is and DT¼63 8C. achieved, Ttrans ¼1137 8C Figure 7 shows that this parameter set fits well with the Hawkesworth et al. (2000) model, at least up to 85% cumulative crystallization. We calculate the cooling history of the magma body by solving the thermal energy balance in a spherical coordinate system, using the Finite element package COMSOL. We use a latent heat of crystallization of 400 kJ kg21, a thermal conductivity of 3 W m21 8C21, a heat capacity of 1 kJ kg21 8C21 and a density of 2800 kg m23. We

assume that the magma body cools simply by conduction. Convective cooling would not dramatically affect the inferred cooling/crystallization history, since similar timescales are obtained in models where convective heat flux is explicitly taken into account (Hawkesworth et al. 2000). Calculations were performed for different aspect ratios, AR¼diameter/thickness (AR ¼ 1, spherical magma body; AR.1, oblate-spheroidal magma body, i.e. sheet-like shape; AR,1, prolatespheroidal magma body, i.e. bottle-shaped) since some workers have argued that magma bodies have a high aspect ratio (Petford et al. 2000), although various observations of spherical or prolate-spheroidal intrusions have also been made (Iyer 1984; Marsh 1989; Higgins 1996; Husen et al. 2004; Iverson et al. 2006; Voight et al. 2006). The results are summarized in Figures 8 & 9. Note that similar results are obtained whether we consider a spherical (AR¼1) or prolatespheroidal (AR¼0.5) magma body. It is likely that the volume and shape of magma bodies vary for each volcanic centres. For this reason, we again use Mount St Helens to illustrate the approach. For Mount St Helens, geophysical studies suggest that the magma chamber has an aspect

THE RATES OF SILICIC MAGMA PRODUCTION

177

Fig. 7. (a) Cumulative crystallization (in %) and (b) crystallization rate (in %8C21) as a function of cooling below liquidus used in our thermal models. We use a simple analytical approximation (solid line) fitted to the model of Hawkesworth et al. (2000), shown in crosses.

ratio ,1 (e.g. Pallister et al. 1992; Iverson et al. 2006). For AR¼0.5, calculations suggest that crystallization of 50% of a magma body .5 km3 is achieved in c. 2000 years (see the previous section) if the wallrock temperature is ,500 8C (Fig. 8a). Assuming a geothermal gradient of 30 8C km21, this corresponds to an emplacement depth ,17 km. These are maximum values since at convergent margins the geothermal gradient is often higher than 30 8C km21, especially if the crust has been heated by repetitive magma intrusions (cf. Annen et al. 2006). To produce andesitic and dacitic compositions in the lower crust (Twr 700 8C) would require a very small magma

Fig. 8. Curves are isochrons (contoured in ka) showing the time required to precipitate 50% of crystals in (a) a prolate spheroidal magma body (aspect ratio¼0.5), (b) a spherical magma body and (c) an oblate spheroidal magma body (aspect ratio¼10) for variable magma volumes (x-axis) and temperatures of the intruded wallrock (in 8C; y-axis).

178

A. DOSSETO ET AL.

Fig. 9. (a) Time needed to crystallize 50% of a 10 km3 magma body as a function of the temperature of the intruded wallrock Twr (in 8C). Curves are shown for magma body aspect ratio¼0.5, 1, 5 and 10. (b) Time required for the crystallization of a 10 km3 magma body emplaced in a crust at 300 8C as a function of the magma body aspect ratio. Curves are shown for crystal content¼50 and 75%.

THE RATES OF SILICIC MAGMA PRODUCTION

body (,3 km3), which is unlikely (Pallister et al. 1992). Hence, our calculations suggest that at Mount St Helens geochemical diversity is produced in the upper crust. Davidson et al. (2007) identified amphibole as a crucial phase during magma chemical evolution. Amphibole can be stable in water-saturated basaltic or andesitic melts at pressures as low as 3 kbar (Beard & Lofgren 1991). Thus, in order to reconcile our cooling model with the stability field of amphibole, differentiation at Mount St Helens must occur in the lowermost pressures of the amphibole stability field (3 kbar). This would correspond to a depth of c. 10 km, which is in good agreement with previous geophysics and petrology studies (Lees 1992; Moran 1994; Blundy & Cashman 2005; Iverson et al. 2006). In turn, this implies that the size of the magma body is c. 10 km3, since a larger body would require shallower emplacement, incompatible with amphibole stability. This is also in agreement with previous estimates (5–7 km3; Pallister et al. 1992). The results are remarkably sensitive to magma body shape. For example, a magma volume of 50 km3 with aspect ratio of 10 (i.e. an oblate spheroidal body with a radius of 5 km and a thickness of 1 km) can undergo 50% crystallization in less than a few thousand years with a wallrock temperature as high as 700 8C. Hence, the ability to constrain the shape of magma bodies at a volcanic centre will be important in attempts to determine whether silicic magmas are produced in shallow spherical or prolate spheroidal magma bodies (e.g. Mount St Helens, Pallister et al. 1992; Mount Taranaki, New Zealand, Higgins 1996) or deep oblate spheroidal bodies (i.e. sheet-like sills). Finally, we note that other processes can potentially yield extremely short differentiation timescales. For instance, Blundy & Cashman (2005) argued that crystallization can be driven by (rapid) decompression and degassing. However, whether this could occur over thousands of years as indicated by the Ra –Th disequilibria data remains to be shown.

Conclusions We have quantitatively investigated several endmember models to account for the uranium-series isotopic variation often observed in co-genetic suites of andesites and dacites and used these to explore time and space constraints on the processes responsible for their generation. In the first model, the composition of silicic magmas is assumed to result solely from mixing between residual melts from differentiation and crustal melting with zero-aged basalt shortly before eruption. The

179

results suggest that about half of the erupted magma volume has to be derived from crustal and residual melts. Considering the few case studies in which pure two end-member mixing has been established, this scenario is deemed unlikely. In the second model, primary basalt and crustal melts are mixed during differentiation (AFC) rather than just before eruption (first model), and Ra –Th data from Mount St Helens are used as a case study to quantify the effect of crustal assimilation on inferred differentiation timescales. Crustal assimilation during crystallization reduces (226Ra/230Th) more rapidly than the effect of the passage of time alone. Including amphibole as a major crystallizing phase also decreases (226Ra/230Th) ratios but this effect is not dramatic. Thus, all of the processes considered here lead to reductions in the timescales of differentiation that have been inferred from negative (226Ra/230Th) 2 Th arrays assuming crystal fractionation was the only processes operating. Consequently, the timescales inferred from such studies should be considered as maxima. Frequent magma recharge can increase the inferred timescales of differentiation. However, this effect is significant only if large amounts of primitive magma frequently replenish the chamber (.20 wt% of the magma in the chamber every 500 years). Because of the short timescales inferred, simple thermal models for magma cooling may require that differentiation occurs shallowly, depending on magma chamber geometry. By combining this with the suggestion that amphibole was an important crystallizing phase (Davidson et al. 2007), we infer that magma emplacement below Mount St Helens occurred at a depth of c. 10 km and that the size of the magma body is c. 10 km3. This is consistent with recent geodetic data (Iverson et al. 2006). Hence, production of silicic magma in spherical or prolate spheroidal (i.e. bottle-shaped) magma bodies probably occurs in the mid-upper crust. Oblate spheroidal magma bodies (i.e. sheet-like sills) can cool over shorter timescales (a few thousand years) at deeper depths. Future studies combining geophysical imagery and geochemical tools such as U-series isotopes at other volcanic centres should provide important constraints on the mechanisms and locations of silicic magma production. We would like to thank T. Rushmer and C. O’Neill for helpful discussions. We would also like to thank the editor (C. Annen) and the reviewers (G. Zellmer and Ch. Hawkesworth) for their helpful comments that largely improved the manuscript. S.P.T. acknowledges an ARC Federation Fellowship. This research was funded by ARC grant DP0451704. This is GEMOC publication no. 507.

180

A. DOSSETO ET AL.

References A NNEN , C., B LUNDY , J. D. & S PARKS , R. S. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology, 47, 505–539. A NNEN , C. & S PARKS , R. S. J. 2002. Effects of repetitive emplacement of basaltic intrusions on thermal evolution and melt generation in the crust. Earth and Planetary Science Letters, 203, 937–955. B ACON , C. R. & D RUITT , T. H. 1988. Compositional evolution of the zoned calcalkaline magma chamber of Mount-Mazama, Crater Lake, Oregon. Contributions to Mineralogy and Petrology, 98, 224–256. B EARD , J. S. & L OFGREN , G. E. 1991. Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3, and 6.9 kb. Journal of Petrology, 32, 365–401. B ERGANTZ , G. W. 1989. Underplating and partial melting: implications for melt generation and extraction. Science, 245, 1093– 1095. B ERGANTZ , G. W. & D AWES , R. 1992. Aspects of magma generation and ascent in continental lithosphere. In: R YAN , M. P. (ed.) Magmatic Systems. Academic Press, San Diego, CA, 291–317. B ERLO , K., T URNER , S., B LUNDY , J. & H AWKESWORTH , C. 2004. The extent of U-series disequilibria produced during partial melting of the lower crust with implications for the formation of the Mount St. Helens dacites. Contributions to Mineralogy and Petrology, 148, 122 –130. B LAKE , S. & R OGERS , N. 2005. Magma differentiation rates from (226Ra– 230Th) and the size and power output of magma chambers. Earth and Planetary Science Letters, 236, 654– 669. B LUNDY , J. & C ASHMAN , K. 2005. Rapid decompression-driven crystallization recorded by melt inclusions from Mount St. Helens volcano. Geology, 33, 793–796. B LUNDY , J. & W OOD , B. 2003. Mineral-melt partition of Uranium, Thorium and their daughters. In: B OURDON , B., H ENDERSON , G. M., L UNDSTROM , C. C. & T URNER , S. P. (eds) Uranium-series Geochemistry. Geochemical Society, Mineralogical Society of America, Washington, DC, 52, 59– 123. B OURDON , B., W O¨ RNER , G. & Z INDLER , A. 2000. U-series evidence for crustal involvement and magma residence times in the petrogenesis of Parinacota volcano, Chile. Contributions to Mineralogy Petrology, 139, 458– 469. B OURDON , B., T URNER , S. & D OSSETO , A. 2003. Dehydration and partial melting in subduction zones: constraints from U-series disequilibria. Journal of Geophysical Research B: Solid Earth, ECV 3-1–3-19. C OOPER , K. M. & R EID , M. R. 2003. Re-examination of crystal ages in recent Mount St. Helens lavas: implications for magma reservoir processes. Earth and Planetary Science Letters, 213, 149– 167. D AVIDSON , J. 1985. Mechanisms of contamination in Lesser Antilles island arc magmas from radiogenic and oxygen isotope relationships. Earth and Planetary Science Letters, 72, 163–174. D AVIDSON , J., T URNER , S., H ANDLEY , H., M ACPHERSON , C. & D OSSETO , A. 2007. An amphibole “sponge” in arc crust? Geology, 35, 787– 790.

D E P AOLO , D. J. 1981. Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization. Earth and Planetary Science Letters, 53, 189–202. D OSSETO , A., B OURDON , B., J ORON , J.-L. & D UPRE´ , B. 2003. U– Th–Pa–Ra study of the Kamchatka arc: new constraints on the genesis of arc basalts. Geochimica et Cosmochimica Acta, 67, 2857– 2877. D UFEK , J. & C OOPER , K. M. 2005. 226Ra/230 Th excess generated in the lower crust: Implications for magma transport and storage timescales. Geology, 33, 833–836. G EORGE , R., T URNER , S., H AWKESWORTH , C., B ACON , C. R., N YE , C., S TELLING , P. & D REHER , S. 2004. Chemical versus temporal controls on the evolution of tholeiitic and calc-alkaline magmas at two volcanoes in the Alaska– Aleutian Arc. Journal of Petrology, 45, 203– 219. G LAZNER , A. F. 2007. Thermal limitations on incroporation of wall rock into magma. Geology, 35, 319–333. G ROVE , T. L. & K INZLER , R. J. 1986. Petrogenesis of andesites. Annual Review of Earth and Planetary Sciences, 14, 417–454. G ROVE , T. L., K INZLER , R. J., B AKER , M. B., D ONNELLY -N OLAN , J. M. & L ESHER , C. E. 1988. Assimilation of granite by basaltic magma at Burnt Lava Flow, Medicine Lake Volcano, Northern California – decoupling of heat and mass-transfer. Contributions to Mineralogy and Petrology, 99, 320–343. H ALLIDAY , A. N., F ALLICK , A. E., D ICKIN , A. P., M ACKENZIE , A. B., S TEPHENS , W. E. & H ILDRETH , W. 1983. The isotopic and chemical evolution of Mount St. Helens. Earth and Planetary Science Letters, 63, 241–256. H AWKESWORTH , C. J., B LAKE , S., E VANS , P., H UGHES , R., M ACDONALD , R., T HOMAS , L. E., T URNER , S. P. & Z ELLMER , G. 2000. Time scales of crystal fractionation in magma chambers – integrating physical, isotopic and geochemical perspectives. Journal of Petrology, 41, 991– 1006. H IGGINS , M. D. 1996. Crystal size distributions and other quantitative textural measurements in lavas and tuff from Egmont volcano (Mt. Taranaki), New Zealand. Bulletin of Volcanology, 58, 194– 204. H UGHES , R. D. & H AWKESWORTH , C. J. 1999. The effects of magma replenishment processes on 238 U – 230 Th disequilibrium. Geochimica et Cosmochimica Acta, 63, 4101–4110. H UPPERT , H. E. & S PARKS , R. S. 1988. The generation of granitic magmas by intrusion of basalt into the continental crust. Journal of Petrology, 29, 599–624. H USEN , S., S MITH , R. B. & W AITE , G. P. 2004. Evidence for gas and magmatic sources beneath the Yellowstone volcanic field from seismic tomographic imaging. Journal of Volcanology and Geothermal Research, 131, 397– 410. I VERSON , R. M., D ZURISIN , D. ET AL . 2006. Dynamics of seismogenic volcanic extrusion at Mount St Helens in 2004–05. Nature, 444, 439–443. I YER , H. M. 1984. Geophysical evidence for the locations, shapes and sizes, and internal structures of magma chambers beneath regions of Quaternary volcanism.

THE RATES OF SILICIC MAGMA PRODUCTION Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences, 310, 473– 510. L A T OURRETTE , T., H ERVIG , R. L. & H OLLOWAY , J. R. 1995. Trace element partitioning between amphibole, phlogopite, and basanite melt. Earth and Planetary Science Letters, 135, 13– 30. L AUBE , N. & S PRINGER , J. 1998. Crustal melting by ponding of mafic magmas: a numerical model. Journal of Volcanology and Geothermal Research, 81, 19– 35. L EES , J. M. 1992. The magma system of Mount St. Helens: non-linear high-resolution P-wave tomography. Journal of Volcanology and Geothermal Research, 53, 103–116. M ARSH , B. D. 1989. Magma chambers. Annual Review of Earth Planetary Sciences, 17, 439– 474. M ORAN , S. C. 1994. Seismicity at Mount St. Helens, 1987–1992: evidence for repressurization of an active magmatic system. Journal of Geophysical Research, 99, 4341– 4354. O ROZCO -E SQUIVEL , M. T., N IETO -S AMANIEGO , A. F. & A LANIZ -A LVARES , S. A. 2002. Origin of rhyolitic lavas in the Mesa Central, Mexico, by crustal melting related to extension. Journal of Volcanology and Geothermal Research, 118, 37–56. P ALLISTER , J. S., H OBLITT , R. P., C RANDELL , D. R. & M ULLINEAUX , D. R. 1992. Mount St Helens a decade after the 1980 eruptions: magmatic models, chemical cycles, and a revised hazards assessment. Bulletin of Volcanology, 54, 124–146. P ETFORD , N. & G ALLAGHER , K. 2001. Partial melting of mafic (amphibolitic) lower crust by periodic influx of basaltic magma. Earth and Planetary Science Letters, 193, 483–499. P ETFORD , N., C RUDEN , A. R., M C C AFFREY , K. J. W. & V IGNERESSE , J.-L. 2000. Granite magma formation, transport and emplacement in the Earth’s crust. Nature, 408, 669–673. P RICE , R. C., G EORGE , R. ET AL . 2007. U– Th–Ra fractionation during crustal-level andesite formation at Ruapehu volcano, New Zealand. Chemical Geology, 244, 437– 451. S MITH , D. R. & L EEMAN , W. P. 1987. Petrogenesis of Mount St. Helens dacitic magmas. Journal of Geophysical Research, 92, 10313–10334. S PARKS , R. S. J. & M ARSHALL , L. A. 1986. Thermal and mechanical constraints on mixing between mafic and silicic magmas. Journal of Volcanology and Geothermal Research, 29, 99–124.

181

T AYLOR , S. R. & M C L ENNAN , S. M. 1995. The geochemical evolution of the continental crust. Reviews in Geophysics, 33, 241– 265. T EGNER , C., W ILSON , J. R. & R OBINS , B. 2005. Crustal assimilation in basalt and jotunite: constraints from layered intrusions. Lithos, 83, 299–316. T EPPER , J. H., N ELSON , B. K., B ERGANTZ , G. W. & I RVING , A. J. 1993. Petrology of the Chiilwack batholith, North Cascades, Washington. Generation of calc-alkaline granitoids by melting of mafic lower crust with variable water fugacity. Contributions to Mineralogy and Petrology, 113, 333 –351. T URNER , S., E VANS , P. & H AWKESWORTH , C. 2001. Ultrafast source-to-surface movement of melt at island arcs from 226Ra-230Th systematics. Science, 292, 1363–1366. T URNER , S., B OURDON , B. & G ILL , J. 2003a. Insights into magma genesis at convergent margins from U-series isotopes. In: B OURDON , B., H ENDERSON , G. M., L UNDSTROM , C. C. & T URNER , S. P. (eds) Uranium-Series Geochemistry. Geochemical Society, Mineralogical Society of America, Washington, DC, 52, 255– 315. T URNER , S., F ODEN , J., G EORGE , R., E VANS , P., V AME , R., E LBURG , M. & J ENNER , G. 2003b. Rates and processes of potassic magma generation at Sangeang Api volcano, east Sunda arc, Indonesia. Journal of Petrology, 44, 491– 515. V OIGHT , B., L INDE , A. T. ET AL . 2006. Unprecedented pressure increase in deep magma reservoir triggered by lava-dome collapse. Geophysical Research Letters, 33, L03312. V OLPE , A. M. & H AMMOND , P. E. 1991. 238 U– 230Th—226Ra disequilibria in young Mount St. Helens rocks: time constraint for magma formation and crystallization. Earth and Planetary Science Letters, 107, 475– 486. Y ODER , H. S., J R & T ILLEY , C. E. 1962. Origin of basalt magmas: an experimental study of natural and synthetic rock systems. Journal of Petrology, 3, 342–532. Z ELLMER , G., T URNER , S. & H AWKESWORTH , C. 2000. Timescales of destructive plate margin magmatism: new insights from Santorini, Aegean volcanic arc. Earth and Planetary Science Letters, 174, 265– 281. Z ELLMER , G. F., A NNEN , C., C HARLIER , B. L. A., G EORGE , R. M., T URNER , S. P. & H AWKESWORTH , C. J. 2005. Magma evolution and ascent at volcanic arcs; constraining petrogenetic processes through rates and chronologies. Journal of Volcanology and Geothermal Research, 140, 171–191.

Long-term geochemical variability of the Late Cretaceous Tuolumne Intrusive Suite, central Sierra Nevada, California WALT GRAY1,2, ALLEN F. GLAZNER2, DREW S. COLEMAN2 & JOHN M. BARTLEY3 1

Engineering Dynamics Department, Southwest Research Institute, PO Drawer 28510, San Antonio, TX 78228, USA

2

Department of Geological Sciences, Mitchell Hall, 104 South Road, University of North Carolina, Chapel Hill, NC 27599-3315, USA (e-mail: [email protected])

3

Department of Geology and Geophysics, University of Utah, Salt Lake City, UT 84112, USA Abstract: This study investigates the internal anatomy and petrogenesis of the Tuolumne Intrusive Suite (TIS), which comprises metaluminous, high-potassium, calc-alkaline granitoids typical of the Sierra Nevada batholith. Although the TIS has often been cited as an example of a large magma chamber that cooled and fractionated from the margins inward, its geochemistry is inconsistent with closed-system fractionation. Most major elements are highly correlated with SiO2, but the scattered nature of trace elements and variations of initial Sr and Nd isotopic ratios indicate that fractional crystallization is not the predominant process responsible for its chemical evolution. Isotopic data suggest mixing between melts of mantle-like rocks and a granitic melt similar in composition to the highest-silica TIS unit. Monte Carlo models of magma mixing confirm that such processes can reproduce the observed variations in major elements, trace elements and isotopic ratios. Thermobarometry suggests emplacement at depths near 6 km and crystallization temperatures ranging from 660 to 750 8C. Feldspars, hornblende, biotite and magnetite exhibit evidence of extensive low-temperature subsolidus exsolution. The TIS as a whole trends toward more evolved isotopic compositions and younger U –Pb zircon ages passing inward. This pattern indicates a general increase in the proportion of felsic, crustally derived melt in the mixing process, which may have resulted from net accumulation of heat added to the lower crust by intrusion of mantle-derived mafic magma. However, the bulk geochemical and isotopic compositions of the equigranular Half Dome Granodiorite, the porphyritic Half Dome Granodiorite and the Cathedral Peak Granodiorite overlap one another and the contacts between them are commonly gradational. We interpret these map units to represent a single petrological continuum rather than distinct intrusive phases. The textural differences that define the units probably reflect thermal evolution of the system rather than distinct intrusive events.

Normally zoned intrusive suites are common in the plutonic record and characteristically have mafic margins that grade toward a felsic core (Buddington 1959; Pitcher 1993). The origin of the zonation is interpreted to reflect either inward-progressing fractional crystallization (e.g. Bateman & Chappell 1979; Walawender et al. 1990) or magma mixing between end members represented by marginal and core facies (e.g. Reid et al. 1983; Kistler et al. 1986). The notion that all plutons and intrusive suites were largely molten at one time so that fractional crystallization and mixing could operate throughout has recently been challenged, and is at odds with geophysical observations and a growing set of geochronological data (e.g. Deniel et al. 1987; Glazner et al. 2004). Consequently, the origin of petrographic and geochemical variation in zoned suites must be reexamined. The Tuolumne Intrusive Suite (TIS) of the Sierra Nevada batholith (SNB), California, is a

particularly well-studied and well-exposed example of a zoned intrusive suite and was originally interpreted to reflect inward fractional crystallization (Frey et al. 1978; Bateman & Chappell 1979). However, subsequent work by Kistler et al. (1986) revealed isotopic variation across the suite, and a mixing origin for the petrological and chemical variation in the TIS became widely accepted. More recently, Coleman et al. (2004) demonstrated a regular age variation across the TIS and, most problematically for the mixing hypothesis, a difference of at least 8 Ma in the ages of the outermost mafic (.93 Ma) and innermost felsic (c. 85 Ma) members of the suite. These authors argued that the age difference was too large for the two magmas to ever have been molten at the same time and, therefore, an origin involving mixing between the exposed units was discredited. Here, we return to the TIS and build on an already large petrographic, chemical, isotopic

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 183–201. DOI: 10.1144/SP304.10 0305-8719/08/$15.00 # The Geological Society of London 2008.

184

W. GRAY ET AL.

and geochronological dataset in an attempt to understand the origin of chemical variation in the suite. Data include major-element, trace-element and isotopic data, as well as the first comprehensive set of mineral analyses published for the TIS. Mineral compositions were originally measured to test the fractional crystallization and mixing hypotheses; however, as our work progressed, we realized that these options were not tenable. A full discussion of the data and the evolution of thought regarding the TIS are found in Gray (2003). This work is based on analysis of approximately 120 samples; additional datasets and thermal modelling are in Gray (2003).

Geological setting The TIS is one of a number of zoned plutonic bodies emplaced along the eastern Sierra Nevada during the Late Cretaceous (Stern et al. 1981; Chen & Moore 1982). Related nested plutonic suites include the John Muir and Mount Whitney Intrusive Suites to the south and the Sonora plutonic complex to the north (Fig. 1). Together these bodies comprise over 2500 km2 of exposed rock that is among the youngest in the Mesozoic SNB and that varies in composition from gabbro to granite. Magmatism that created the SNB occurred in three major episodes from approximately 220 to 80 Ma (Stern et al. 1981; Bateman 1992). The largest volume was emplaced during the middle to Late Cretaceous (c. 120 to 80 Ma, Stern et al. 1981; Chen & Moore 1982; Coleman & Glazner 1997) along the present axis of the batholith. The TIS was emplaced over a period of approximately 8 Ma, with concordant zircon U –Pb ages decreasing from approximately 93 Ma at its margins to 85 Ma near its centre (Coleman et al. 2004; Matzel et al. 2005). It intruded between older Cretaceous granitoid rocks (intrusive suite of Yosemite Valley, Sentinel Granodiorite) to the west and south, and Paleozoic to Jurassic metasedimentary and metavolcanic rocks to the north and east (Fig. 1). The TIS is a concentrically zoned suite consisting of five mapped units (Calkins 1930; Bateman et al. 1983, 1988; Bateman 1992). From the margins inward the units are: (1) the granodiorite of Kuna Crest; (2) the equigranular Half Dome Granodiorite; (3) the porphyritic Half Dome Granodiorite; (4) the Cathedral Peak Granodiorite; and (5) the Johnson Granite Porphyry (Fig. 1). The rocks are predominantly medium- to coarse-grained, except the Johnson Granite Porphyry, which has a highly variable, but generally aplitic to alaskitic, texture. The TIS contains a consistent suite of minerals, but abundance and grain size varies among

the units. The primary mineralogy is plagioclase (andesine to oligoclase), K-feldspar (microcline perthite), quartz, hornblende and biotite, with accessory titanite, magnetite, zircon, apatite and allanite (Bateman & Chappell 1979). In general, the proportions of mafic minerals and corresponding color indices decrease from the margins inward. Hornblende and biotite together constitute approximately 20% of the granodiorite of Kuna Crest, but less than 1% of the Johnson Granite Porphyry. The largest decrease occurs in traverses across the Kuna Crest and equigranular Half Dome Granodiorite (Bateman & Chappell 1979). Modal layering of mafic minerals is commonly observed in the granodiorite of Kuna Crest and at the outer margins of both the equigranular Half Dome and Cathedral Peak granodiorites, but is generally sparse elsewhere. Mafic magmatic enclaves are common in the Kuna Crest and Half Dome granodiorites. A moderately strong margin-parallel foliation is present, as indicated by the preferred alignment of mafic minerals and flattened mafic enclaves (Bateman 1992; Zak et al. 2007). Evidence for solid-state deformation is locally marked by elongated quartz grains and the wispy appearance of linear disaggregated hornblende and biotite. The Half Dome Granodiorite is easily recognized by euhedral hornblende phenocrysts up to 20 mm long, and large (typically up to 4 mm and locally larger) euhedral titanite crystals. The porphyritic Half Dome Granodiorite is also distinguished by the presence of abundant K-feldspar phenocrysts and rare megacrysts (crystals .50 mm in length). K-feldspar megacrysts also are abundant in the Cathedral Peak Granodiorite, but they are typically larger (up to 100 mm in length) and more blocky than in the adjacent porphyritic Half Dome Granodiorite (Johnson et al. 2006). The Johnson Granite Porphyry is characterized by a nearly equal mixture of fine-grained sodic plagioclase, K-feldspar and quartz, with minor biotite and a general absence of hornblende. Locally it contains miarolitic cavities and sparse spherical blobs, typically 10–20 cm across, of K-feldspar megacrysts surrounded by Cathedral Peak-like matrix (Titus et al. 2005). Concentric zonation and gradational contacts led early workers to hypothesize that emplacement occurred in three separate pulses, each successive pulse having intruded into the still molten previous one, which was still mobile enough to be pushed out of the way without forming a significant solid-state fabric (Evernden & Kistler 1970; Frey et al. 1978; Bateman & Chappell 1979). As envisioned by Bateman & Chappell (1979), the Kuna Crest was emplaced first, followed by the Half Dome Granodiorite, then the Cathedral Peak Granodiorite and

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

185

Fig. 1. Geology of the TIS, central Sierra Nevada, California (after Bateman 1992) and plutons comprising the Late Cretaceous Sierra Crest event (Coleman & Glazner 1997). The TIS is one of a number of compositionally zoned pluton suites intruded along the Sierra crest during the Late Cretaceous.

186

W. GRAY ET AL.

Johnson Granite Porphyry. Diapiric rise and shouldering aside of wall rocks was considered the most likely emplacement scenario for the early units, with subsequent units emplaced by stretching of partially crystallized outer margins (ballooning), as evidenced by foliation in the Kuna Crest and progressive weakening of foliation inward (Bateman et al. 1983; Bateman & Chappell 1979). Contacts between units vary from sharp to gradational and dip moderately to steeply away from the centre of the suite (Bateman 1992; Coleman et al. 2006). Contact orientations are difficult to determine in many places owing to broad, gradational contacts. The contact between the inner Johnson Granite Porphyry and Cathedral Peak Granodiorite is generally sharp (Titus et al. 2005), whereas contacts between the Cathedral Peak, equigranular and porphyritic Half Dome and Kuna Crest are typically gradational over tens to hundreds of metres (Johnson et al. 2006). The steeply dipping margin-parallel foliation, so pronounced in the outer granodiorite of Kuna Crest, becomes progressively weaker inward and is scarcely visible in the Johnson Granite Porphyry (Bateman 1992; Zak et al. 2007).

Petrology of the Tuolumne Intrusive Suite

For isotope analyses, a single dissolution was used with 500 mg of powdered sample loaded into Teflonw bombs with a mixture of HF þ HNO3, and placed in an oven at 180 8C for 5 days. After dissolution, the solution was dried, dissolved in hydrochloric acid and separated into three aliquots for Sr, Nd and Pb isotope analysis. Each aliquot was then dried and re-dissolved in the appropriate starting acid solution for ion exchange chemistry. The aliquots were then subjected to ion-exchange chromatography followed by analysis using a Micromass VG Sector 54 thermal ionization mass spectrometer (TIMS) at the University of North Carolina. Strontium isotopic measurements were normalized to 86Sr/88Sr ¼ 0.1194, and Nd isotopes to 146Nd/144Nd ¼ 0.7219. Six replicate analysis of standards during the study period yielded a mean 87Sr/86Sr ¼ 0.710257 + 0.000022 (2s) for NBS 987, a mean 143Nd/144Nd ¼ 0.512112 + 0.000011 (2s) for JNdi-1, and a mean 206Pb/ 207 Pb ¼ 1.0940 + 0.0003 (2s) for NBS 981 with a mean fractionation correction of 0.098 + 0.008% per amu. Isotopic ratios were corrected to the mean age of the host rock (Coleman et al. 2004). All data tables (mineral analyses, whole-rock major- and trace-element analyses, and Sr, Nd and Pb isotope analyses) are available online at http:// www.geolsoc.org.uk/SUP18320.

Methods

Mineral chemistry

Mineral analyses were performed by wavelengthdispersive X-ray spectrometry using the Cameca CAMEBAX electron microprobe at Duke University. X-ray intensities were corrected utilizing the methods and correction factors of Bence & Albee (1968). All analyses were performed using an accelerating voltage of 15 kV and a sample current of 15 nA. A beam diameter of 20 mm was used in each analysis except during K-feldspar analyses, during which a beam diameter of 50 mm was used to better average exsolution lamellae. Where applicable, concentrations of Na, Cl and F were determined first to minimize the effect of volatile loss and element migration. Mineral compositions are reported in weight-percent oxides and represent the average of at least five analyses of each mineral grain. Analysis for major and trace elements used a combination of fusion ICP, total digestion ICP, ICP-MS and INAA techniques. Approximately 1–2 kg of material was collected at each location and reduced to powder in a steel jaw crusher and alumina shatterbox. Based on replicate analyses of two samples, analytical precision is estimated to be approximately +2% for major oxides and trace elements, and +3% for rare earth elements.

Fifteen polished thin sections representing all units of the TIS and two mafic enclaves were analysed to determine representative mineral compositions. Feldspar, titanite, biotite and hornblende molecular proportions were calculated on the basis of 8, 4, 22 and 24 oxygens, respectively. All iron in feldspar, titanite and biotite was assumed to be Fe2þ, whereas Fe3þ was calculated in hornblende from charge balance by the method of Cosca et al. (1991). Hornblende normalization to 24 oxygens was chosen over the more traditional 23 oxygens because the latter scheme has been shown to overestimate Fe3þ. In addition, normalization to 24 oxygens is the recommended procedure for hornblende and plagioclase thermobarometry (Blundy & Holland 1990; Holland & Blundy 1994; Anderson & Smith 1995). The K-feldspars measured in this study are potassium-rich and vary only from Or86 to Or90 (Fig. 2; see also Kerrick 1969; Johnson et al. 2006). Orthoclase content does not correlate with whole-rock silica or location within the intrusive suite. Anorthite and iron concentrations are uniformly low, less than 0.3 and 0.1 mol%, respectively. Plagioclase occurs as both optically zoned and unzoned grains, the zoned grains displaying both

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

187

Fig. 2. Feldspar compositions in the TIS do not agree with the experimentally determined compositions of Elkins & Grove (1990), and suggest pervasive subsolidus modification. Pressure and temperature estimates for crystallization of the TIS from this study. See text for discussion.

normal and oscillatory zoning. Anorthite content in unzoned grains decreases systematically inward across the TIS from approximately An50 in the marginal granodiorite of Kuna Crest to An15 in the Johnson Granite Porphyry (see also Bateman & Chappell 1979). Average An content in the zoned plagioclase grains is similar, but ranges to higher An values. Plagioclase Or content is less than 2.5 mol% in all grains analysed. Iron content is slightly higher than that observed in the K-feldspars, reaching 0.22 mol% in one sample (HD01-6). Textural evidence for relict grains (resorbed and embayed edges, overgrowths, etc.) and cumulate grains is generally lacking. However, synneusis and boxy-cellular texture (Hibbard 1995) are common. Hornblendes are calcic and plot in the magnesiohornblende to actinolite fields on the classification diagram of Leake (1978). Although hornblende SiO2 generally increases with host-rock SiO2, the increase is not systematic and considerable variability is observed within individual TIS units. FeO generally decreases with host-rock SiO2, similar to the trend observed for whole-rock FeO, but hornblende MgO increases, contrary to the whole-rock trend. Other oxides show no correlation with host-rock chemistry, consistent with a previous study of central SNB hornblendes (Dodge et al. 1968). Hornblende Fe/(Fe þ Mg) correlates negatively with whole-rock SiO2, decreasing from approximately 0.42 in the granodiorite of Kuna Crest to 0.30 in the Cathedral Peak Granodiorite (the Johnson Granite Porphyry contains no appreciable hornblende). Fluorine and chlorine are present in only minor amounts, typically less than 0.3 wt%. Except for the Kuna Crest Granodiorite, hornblende

Ti contents are not correlated with Ti in biotite from the same rock (Fig. 3). Biotite contains nearly constant molecular abundances of Fe and Mg, resulting in a narrow range of Fe/(Fe þ Mg) ratios (approximately 0.48 –0.41) that decrease with whole-rock SiO2. Variations with whole-rock chemistry mimic those of hornblende, but each oxide varies by no more than 1 wt% over the full range of whole-rock SiO2 (55 –74 wt%). Only in the Johnson Granite Porphyry does F content exceed 1 wt%; in all other units, F content is typically less than 0.25 wt%. Chlorine is present in only trace amounts. In contrast to the other minerals, titanite major-element concentrations vary between samples by less than 0.6 wt%, and are uncorrelated with host-rock chemistry.

Fig 3. Titanium (Ti) compositions of the TIS hornblende and biotite. Lack of correlation in Ti composition is thought to indicate subsolidus re-equilibration (Ague & Brimhall 1988).

188

W. GRAY ET AL.

A significant aspect of hornblende and biotite chemistry is the decreasing Fe/(Fe þ Mg) with whole-rock SiO2, which indicates crystallization under increasing fO2 conditions (Czamanske & Wones 1973; Czamanske et al. 1981). Elevated fO2 is consistent with the assemblage magnetite þ titanite þ quartz (Wones 1989). Wones (1989) estimated that the reaction hedenbergite þ ilmenite þ oxygen ¼ titanite þ magnetite þ quartz lies about 2 log units above the Ni –NiO buffer.

Thermobarometry Estimates of pressure and temperature at the time of crystallization were made by simultaneous solution of the plagioclase –amphibole thermometer of Blundy and Holland (Blundy & Holland 1990; Holland & Blundy 1994), and the temperaturedependent Al-in-hornblende barometer of Anderson & Smith (1995). Estimated plagioclase temperatures vary from approximately 750 + 40 8C in the granodiorite of Kuna Crest and mafic enclaves, to approximately 700 + 40 8C in the Cathedral Peak Granodiorite (Fig. 4). Zircon saturation temperatures (Watson & Harrison 1983) range from 710 to 770 8C, in general agreement with the plagioclase thermometry and suggesting late-magmatic crystallization of zircon. Although estimated pressures range from 0.35 to 0.03 GPa, the majority of samples indicate pressures from 0.1 to 0.2 GPa,

with an average of 0.17 GPa (corresponding to a depth of 6 km), consistent with previous estimates for central Sierra Nevada plutons (Bateman 1992). Ague & Brimhall (1988) also obtained pressures of 0.05–0.23 GPa for two Cathedral Peak Granodiorite samples from Al-in-hornblende barometry. The estimated temperatures resemble experimentally determined solidus temperatures for synthetic granites and granodiorites (Whitney 1988) and natural dacite melts (Scaillet & Evans 1999) at 0.2 GPa confining pressure. At these temperatures and pressures, hornblende on the solidus suggests an H2O content of at least 4 wt% (Scaillet & Evans 1999). TIS samples have plagioclase and hornblende compositions well-matched to the compositions used in the thermobarometric calibrations. In addition, the mineral assemblage assumed by Anderson & Smith (hornblende þ biotite þ plagioclase þ quartz þ alkali feldspar þ titanite þ magnetite or ilmenite) is identical to that found in the TIS. However, hornblendes in the TIS display lower Fe/(Fe þ Mg), which decreases with wholerock SiO2. A decreasing Fe/(Fe þ Mg) 2 SiO2 trend, combined with a high Mg content, indicates an elevated oxidation state in the hornblende (Wones & Gilbert 1982; Wones 1989). Progressive oxidation of the hornblende decreases its total Al content, resulting in a potential underestimate of pressure (Anderson & Smith 1995). Thus, the calculated pressures for the TIS may be slightly low.

Major- and trace-element geochemistry

Fig. 4. Estimated pressures and temperatures from Al-in-hornblende barometry (Anderson & Smith 1995), and amphibole – plagioclase thermometry (Holland & Blundy 1994). Average results suggest an emplacement depth of approximately 6 km. Error of methods estimated as +0.06 GPa, and +40 8C.

New major-oxide and trace-element data are presented for 56 samples collected from locations within all units of the TIS, as well as three mafic enclaves and one aplite dyke. Rocks of the TIS are metaluminous, high-potassium, calc-alkaline granitoids, similar to many other plutonic suites within the SNB (Bateman 1992). The TIS displays a broad compositional range, with SiO2 content varying from approximately 59 wt% in the marginal granodiorite of Kuna Crest, to 75 wt% in the central Johnson Granite Porphyry. Compositional variability within individual units is more restricted, except for the equigranular Half Dome Granodiorite in which compositions overlap all other units except the Johnson Granite Porphyry (Fig. 5). A small compositional gap exists between the Johnson Granite and the older units, and the major and trace-element contents of the Johnson closely approach the composition of the aplite dyke (Fig. 5). Major oxides are well-correlated with SiO2, except Na2O and K2O, possibly the result of low-temperature re-equilibration (see Discussion).

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

189

Fig. 5. Major element v. SiO2 (wt%) plots. All symbols as in the first panel. All major elements except Na2O and K2O display strong linear trends.

However, most trace elements display little or no correlation with SiO2 (Fig. 6). Notable exceptions include the transition metals (V, Co, Ni and Sc) and Sr, which display strong negative correlations. Zr and Hf display weak negative correlations, and Rb displays a weak positive correlation. However, considerable scatter is observed in these plots. When plotted against each other, most incompatible trace elements show only weak correlations. However, strong correlations exist for Zr –Hf (Zr–Hf is nearly constant across all units) and

Nb –Ta, as is expected from their similar chemical behaviour. All units of the TIS are depleted in heavy rare earth elements (REE) relative to light REE (Fig. 7). Light REE abundances are similar in all units and overlap in normalized plots, but heavy REE abundances decrease systematically from the margins inward and correlate negatively with SiO2. Correlation of REE abundance with wholerock SiO2 changes with atomic number: light REE display no significant correlation, middle

190

W. GRAY ET AL.

Fig. 6. Plots of representative trace elements v. SiO2. All symbols as in the first panel. Most trace elements display weak to no correlations with SiO2. Notable exceptions include the transition metals and Sr.

REE are strongly negatively correlated and heavy REE are weakly negatively correlated. Although the depletion of heavy and middle REE relative to light REE could be explained by ubiquitous removal of zircon or titanite, evidence from the TIS is mixed. Both Zr and Hf decrease in the more felsic units, but for other trace elements commonly associated with these minerals, namely U, Th and Y, such trends are not apparent. The similarity of TIS REE patterns to those of the mafic enclaves may suggest significant inheritance

of these patterns from the source, especially if garnet is present. The absence of pronounced Eu anomalies, except in isolated cases, is a ubiquitous feature of the TIS. Minor Eu anomalies occur in a few samples of the granodiorite of Kuna Crest and Johnson Granite Porphyry, but are absent in the other units. It is most significant that the one aplite sample does not show the Eu anomaly that might be expected in this highly differentiated rock (Glazner et al. 2008).

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

191

Fig. 7. Chondrite-normalized REE diagrams for the TIS. Heavy REE abundance decreases with increasing SiO2, and toward the centre of the suite, but light REE abundance is uncorrelated with SiO2 and position within the suite. Like most other plutons in the Sierra Nevada, the TIS is characterized by depletion of heavy REE relative to light REE, and a general absence of Eu anomalies. REE compositions normalized to values of Anders & Grevesse (1989).

Radiogenic isotope geochemistry Twenty-six samples were analysed to determine whole-rock Sr, Nd and Pb isotopic compositions. Rocks of the TIS display considerable variability in whole-rock Sr and Nd isotopic composition with a negative correlation between initial Sr ratio and 1Nd(t). Initial 87Sr/86Sr varies from 0.7056 to 0.7073 and 1Nd(t) varies from approximately 23 to 28 (Fig. 8a). Data from the granodiorite of Kuna Crest, equigranular Half Dome Granodiorite and Johnson Granite Porphyry plot in distinct and separate fields that do not overlap with other units. However, there is a continuous variation in the isotopic compositions of the suite as a whole, and the isotopic compositions of the porphyritic Half Dome and Cathedral Peak granodiorites are similar and plot within a common region. The

negative trend shown by the Kuna Crest samples is less pronounced than the rest of the TIS and resembles the trend defined by adjacent and older intrusive rocks on the west side of the TIS (Sentinel Granodiorite and intrusive suite of Yosemite Valley; Kistler et al. 1986; Ratajeski et al. 2001; Fig. 8b). With the exception of one porphyritic Half Dome sample, the remaining units appear to share a common but more negative trend. At present it is unclear whether the composition of this one sample is anomalous or can be explained by mixing between potential source rocks or melts (see the section below on Source of Isotopic Variability). Considerable variability was also observed in initial Pb isotopic compositions with 206Pb/204Pbi varying from 18.8 to 19.3, 207Pb/204Pbi from 15.5 to 15.8, and 208Pb/204Pbi from 38.2 to 38.9

192

W. GRAY ET AL.

Fig. 8. Strontium, Nd and Pb isotopic trends in the TIS. Symbols as in panel (a). (a) 1Nd(t) v. 87Sr/86Sri. (b) Relationship between TIS isotopic compositions and those of the Sentinel Granodiorite, intrusive suite of Yosemite Valley and potential source rocks. The isotopic compositions of the granodiorite of Kuna Crest are distinct from the remaining units of the TIS, lying along a trend defined by the Sentinel Granodiorite and intrusive suite of Yosemite Valley. Data: TIS from this study; hypothetical basalt and felsic source, Kistler et al. (1986); Sentinel Granodiorite, Kistler et al. (1986) and Coleman & Glazner (1997); xenoliths, Ducea & Saleeby (1998); intrusive suite of Yosemite Valley, Ratajeski (1999). (c) Initial Pb isotopic data. Linear regression of 206Pb/204Pb and 207Pb/204Pb yields a slope corresponding to an age of 1.96 Ga, interpreted as the mean age of lithosphere beneath the Sierra Nevada. (d) Sr isotopes v. 1/Sr, including data from Kistler et al. (1986). No simple mixing relationship is apparent. (e) Sr isotopic ratios generally increase with silica, but again, no simple relationship is apparent.

(Fig. 8c). These compositions plot within the same general field as other SNB granitoids and close to the composition of primitive island-arc lavas (Doe & Zartman 1979; Chen & Tilton 1991; Ratajeski et al. 2001). Linear regression of 206 Pb/204Pbi v. 207Pb/204Pbi results in a slope of 0.1208 + 0.018, equivalent to a 207Pb/206Pb* age of 1.96 + 0.24 Ga, consistent with ages calculated from previous Pb isotopic studies, and is interpreted

as the mean age of a lithospheric component beneath the SNB (Doe & Delevaux 1973; Chen & Tilton 1991). On a plot of 87Sr/86Sri v. 206 Pb/204Pbi, the different trends displayed by samples from the granodiorite of Kuna Crest and the other TIS units are again apparent. Scattered possible mixing trends are observed in all isotopic systems. The trend is most apparent on plots of 1Nd(t) v. 87Sr/86Sri (Fig. 8a),

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE 87

Sr/86Sri v. 1/Sr (Fig. 8d) and 87Sr/86Sri v. SiO2 (Fig. 8e). The convex-upward curvature of the 1Nd(t) v. initial Sr is unusual and requires the more radiogenic end member to have higher Sr/Nd than the less radiogenic end member (DePaolo & Wasserburg 1979). This contrasts sharply with other SNB granitoids where the more radiogenic end member typically has lower Sr/Nd (DePaolo 1981). Alternatively, the convex upward trend may be the combination of two trends, the flatter trend of the granodiorite of Kuna Crest and the more negative trend of the remaining TIS units, suggesting that (1) the Kuna Crest is isotopically unrelated to the rest of the TIS, (2) there are more than two mixing end members or (3) a more complex mixing process is responsible for the pattern. The two mafic enclaves have higher 87Sr/86Sri and lower 1Nd(t) than the granodiorite of Kuna Crest, even though they are more mafic in composition. In each case the isotopic composition is similar to the host granitoid (porphyritic Half Dome and Cathedral Peak). Similarity in isotopic composition between host granitoid and enclave has been observed in many other plutons (Pin et al. 1990; Allen 1991; Barbarin 1991), and is interpreted to result from either (1) isotopically similar source magmas, or (2) interactions leading to partial isotopic equilibrium.

Discussion Low-temperature mineral equilibration Evidence for subsolidus mineral modification is apparent from the mineral chemistry. Feldspar compositions do not agree with the experimental equilibrium compositions of Elkins & Grove (1990) and suggest pervasive subsolidus modification (Fig. 2; Johnson et al. 2006). Although plagioclase An contents are consistent with equilibration with a granitic melt (Johannes 1989), the uniformly Or-rich content of the alkali feldspars and Or-poor content of the plagioclase feldspars indicate thorough subsolidus re-equilibration with respect to potassium (Brown & Parsons 1989). Local threefeldspar assemblages that avoid the peristerite gap suggest equilibration at temperatures below 500 8C (Johnson et al. 2006). Unzoned plagioclase crystals of the same composition as the average composition of associated zoned crystals suggest local internal homogenization of originally zoned crystals, perhaps during the same period of subsolidus modification. Magnetite crystals have exsolved nearly all of their Ti ulvo¨spinel component. The general lack of correlation between hornblende and biotite Ti content in all units except the Kuna

193

Crest Granodiorite also suggests subsolidus modification (Fig. 3). Ague and Brimhall (1988) found a similar relationship among other SNB granitoids and argued for low-temperature re-equilibration. Thus, TIS hornblende, biotite, magnetite and feldspar chemistry are all consistent with significant re-equilibration at temperatures below the solidus. This suggests that P–T estimates derived from the hornblende –plagioclase system should be viewed with caution, at least in long-lived, slowly cooled systems.

Spatial and temporal variations in composition The spatial distribution of major and trace elements was examined using data from three traverses across some or all of the TIS (Fig. 9). The data reveal a progressive increase in SiO2 from the margins inward. All other major elements except Na2O and K2O also correlate with position, reflecting their well-defined correlations with SiO2. Trace element correlations are mixed. Rubidium shows the strongest correlation, with an almost linear increase inward across the TIS, followed by the transition metals that decrease in abundance from the margins inward. The remaining trace elements show weak to no correlation with position, again reflecting their lack of correlation with SiO2. In contrast to the trace elements, there is a clear relationship between the isotopic composition and location within the TIS (Fig. 9). Overall, 87 Sr/86Sri increases and 1Nd(t) decreases from the margins inward, reaching maximum and minimum values, respectively, in the Johnson Granite Porphyry. Because age decreases continuously inward (Coleman et al. 2004), this correlation is linked to the variation in age. This conclusion contrasts with an earlier study of Chen & Tilton (1991), who found no correlation between age and isotopic composition of rocks elsewhere in the central Sierra Nevada. However, they based this conclusion on a single age from each pluton sampled and did not examine variation within individual plutons.

Origin of chemical variation in the Tuolumne Intrusive Suite The obvious concentric textural and compositional zoning, gradational contacts between the units, linear major-element trends on SiO2 plots, systematic changes in mineral compositions and whole-rock isotopic compositions, and the trend toward more felsic compositions inward noted in this and previous studies (Bateman & Eaton 1967; Frey et al. 1978; Bateman & Chappell 1979; Reid et al. 1983; Kistler et al. 1986; Coleman & Glazner

194

W. GRAY ET AL.

Fig. 9. Composition of selected elements and initial isotopic composition v. position within the TIS. Limited data from three traverses reveal a progressive increase in SiO2 from the margins inward. All other major elements except Na2O and K2O also correlate with position, but the trace-element compositions are mixed. Both initial Sr and Nd ratios are well correlated with position, becoming more radiogenic inward. Because age also decreases continuously inward (Coleman et al. 2004), we can infer that the isotopic compositions vary with age, the youngest rocks having the most evolved isotopic compositions. Distance is measured from the western margin of the TIS.

1997) led earlier researchers to consider the TIS an example of emplacement followed by fractional crystallization, magma mixing or some combination thereof. However, Coleman et al. (2004)

and Glazner et al. (2004) noted that these processes cannot be the sole processes at work because the outer units of the suite are at least 8 Ma older than the inner units, and thermal modelling indicates

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

that even large magma bodies will cool to the solidus in a matter of hundreds of thousands of years. Consequently, a new model that is consistent with geochemical, isotopic and geochronologic data is needed. Any model for formation of the TIS must incorporate the regular temporal and spatial trends observed, but also account for the apparent linear trends in major-element and scatter in traceelement data. One model that could explain the observations is simply derivation of the various units from an evolving partial-melting zone (hot zone) in the deeper crust (Ratajeski et al. 2005; Annen et al. 2006). If the source region evolves via thermal maturation, progressive movement (upward?) and depletion of source materials, then the overall geochemical evolution of the suite and lack of correlation among trace elements could simply be a result of changes in the source region. In this admittedly ad hoc model the various magmas are related only in that they were derived from nearby areas during the same long-term thermal event and came to freeze in the same general part of the upper crust. An alternative process that honours a stronger genetic relation among the components of the TIS and can produce the observed correlations (and lack thereof ) is mixing of partial melts. Sisson et al. (2005) demonstrated that partial melting of Sierran gabbros and diorites can produce a range of melt compositions from gabbro to granite. They also found a nearly linear relationship between partial melt fraction and the major-oxide content of the resulting melt for most oxides. Using this as a starting point, we developed a predictive model to evaluate partial melting and mixing as candidate processes for evolution of the Tuolumne trends. The model assumes a two-step process in which melting of a mafic metaigneous crustal rock is followed by mixing of the partial melts thus produced. Major-element, trace-element and isotopic compositions are calculated using Monte Carlo methods. The model generates three uniformly distributed random numbers drawn from [0 1] to represent two partial melt fractions (F) and the proportion of the first melt in the resulting mixture. The F-values are used to calculate the composition of each melt, and the final composition is determined from binary mixing equations. Variability in compositions were examined by conducting approximately 400 Monte Carlo calculations (two partial melts and one mixture composition from each calculation) for several major and trace elements. Because of the approximately linear relationship between melt percentage and major-element content, random mixing of any number of these partial melts results in major-oxide compositions that lie on the same trend (Fig. 10a). Thus, the model predicts linear major-element behaviour

195

consistent with that observed in the TIS (Fig. 10b). The exception is Na2O; owing to the curvilinear partial melt trend of Na2O, random mixing produces some scatter, but this also is consistent with the variation of Na2O in the TIS (Fig. 5). A batch-melting model was used to predict trace-element trends, assuming that liquids were equilibrated with their sources before extraction. Calculation of the bulk distribution coefficients is somewhat problematic in that knowledge of the minerals and proportions melted from the source rock is required. In the case of the TIS, the source rock is poorly known, and thus calculation of welldetermined D-values is not possible. However, if D is far from unity, the general pattern of chemical behaviour is relatively insensitive to the precise value of D. Therefore, trace-element behaviour was simulated using a D of 0.01 for incompatible elements and a D of 10 for compatible elements. Starting trace-element compositions also have to be selected, so we assumed typical values from partial melts of representative mafic rocks in the region (Ratajeski 1999), although we do not necessarily imply that these rocks provided the TIS source magmas. On plots of trace-element concentrations v. whole-rock SiO2, incompatible trace elements are enriched in high SiO2 (low volume) partial melts, whereas compatible trace elements are depleted (Fig. 10c). Again, owing to the curvature of each trend, the composition of a random mixture will lie along a line offset from the predicted melting trajectory (Fig. 10c & d). This introduces considerable scatter, especially in the incompatible elements. The random nature of the mixing process also produces decoupling of the incompatible elements from one another owing to different values of partition coefficients. A comparison of model predictions with major- and trace-element data from the TIS reveals that this process is consistent with the observed data (Fig. 10e– h). Thus, the partial melt and random mixing model predicts trace-element scatter comparable to the TIS, while preserving the relative linearity of the major-element trends. We emphasize that we are not trying to reproduce observed trends with this model, but to explore processes that can explain the features of these trends.

Source of isotopic variability Although the observed major- and trace-element trends can be reproduced by randomly mixing partial melts from a single source rock, isotopic data clearly require at least one additional source to reproduce the isotopic trends (Fig. 8). From a study of initial Sr and Nd isotopic ratios, Kistler et al. (1986) argued that mixing between mafic and felsic end members is the most plausible

196

W. GRAY ET AL.

Fig. 10. Monte Carlo partial melting and mixing model predictions compared to data from the TIS. Model predictions are shown as solid circles, data from the TIS shown as open circles. (a) Predicted major-oxide compositions after partial melting and random mixing. Owing to the linear nature of the partial melt composition v. SiO2 relationship (see text), final mixed compositions are confined to a line. (b) Model predictions for CaO compared with TIS data. (c) Monte Carlo batch melting model predicts curvilinear trends for incompatible and compatible trace elements. Random mixing of any two melt compositions produces scatter away from the melting trends. Cm/Co is the ratio of the concentration of an element in the melt to its composition in the source rock. (d) Predicted compositions after random mixing of partial melts (incompatible elements shown as solid circles, compatible elements shown as solid triangles). Model predictions for Rb (e), Ta (f), Sr (g) and Zr (h) compared with TIS data. Assumed model starting compositions: Rb ¼ 25 ppm, Ta ¼ 0.15 ppm, Sr ¼ 850 ppm and Zr ¼ 325 ppm.

explanation for the range of intermediate compositions observed in the TIS. Although liquid-crystal fractionation may have been active on a local scale, its signature is masked by the more predominant mixing process. Kistler et al. (1986) argued that

covariation of initial Sr and Nd ratios could be modelled by simple mixing of basalt with 48 wt% SiO2, 87 Sr/86Sri ¼ 0.7047 and 143Nd/144Ndi ¼ 0.51269 with a granitic magma with 73.3 wt% SiO2, 87 Sr/86Sri ¼ 0.7068 and 143Nd/144Ndi ¼ 0.51212

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

(Fig. 8b). Whether these two hypothetical compositions represent realistic source magmas for the TIS is difficult to assess, as neither corresponds to any specific rock unit presently observed in the SNB. Although Kistler’s granitic composition is similar to the Johnson Granite Porphyry, the basaltic composition lies considerably off the overall TIS trend (Fig. 8b). Studies of peridotite xenoliths from Cenozoic volcanic rocks suggest that the mantle beneath of the SNB at the time of TIS emplacement possessed on average a higher initial 87Sr/86Sr and lower 143Nd/144Nd than proposed by Kistler and his coworkers (Beard & Glazner 1995; Coleman & Glazner 1997; Ducea & Saleeby 1998). Thus, Kistler’s basalt may not be the best candidate for the isotopically less-evolved end member (Fig. 8b). Another possible candidate for the less-evolved end member are melts derived from sub-Sierran peridotite. With 87Sr/86Sri ¼ 0.7058 to 0.7062 and 143 Nd/144Ndi from 0.5122 to 0.5125, the xenoliths plot near the middle of the TIS trend (Fig. 8b). Mixing of Kistler’s granitic end member with peridotite melts could explain the trend between the Half Dome Granodiorites, Cathedral Peak Granodiorite and Johnson Granite Porphyry, but cannot account for the granodiorite of Kuna Crest. This opens the possibility that three different magmas were mixed to produce the TIS. Underplating by basaltic liquids similar in isotopic composition to the peridotite xenoliths melted the older Sentinel and Yosemite Valley rocks that then mixed to produce the granodiorite of Kuna Crest. Mixing one or several Kuna Crest compositions with the crustal granitic melt produced the remaining TIS units. However, all components would have been molten at the same time, and it seems unlikely that cross trends (scatter away from the mixing line) could have been avoided. A more likely scenario is basaltic underplating that melted the Sentinel and Yosemite Valley rocks, followed by mixing of the components to produce the granodiorite of Kuna Crest. A later but less-intense pulse of basaltic underplating mixed with partial melts of dioritic to gabbroic lower crust (of varying SiO2 content; Reid et al. 1993; Ratajeski et al. 2005) producing the remainder of the TIS. To date the identities of source magmas responsible for the TIS (and the SNB as a whole) remain largely a topic of speculation and controversy. We used the end-member isotopic compositions estimated by Kistler et al. (1986) to determine how well the random mixing model fared in producing the observed variability. Kistler et al. did not estimate the Nd concentration in the end members, and therefore we used typical values from the granodiorite of Kuna Crest and Johnson Granite

197

Porphyry of 24 and 19 ppm, respectively, in model calculations. In our model (Fig. 11), although the isotopic compositions of the two end members are fixed, the Sr and Nd concentrations of melts vary in accordance with the trace-element systematics described earlier. For simplicity, the major- and trace-element composition of the basaltic liquid is held constant although, in reality, local fractionation and assimilation would probably modify this composition as well. The model indicates that this partial melting and mixing process can create scatter about twocomponent isotopic mixing trends without requiring additional source components (Fig. 11a). If Sr –Nd ratios of the end members were constant, a mixing hyperbola with no cross-trend variability would

Fig. 11. Monte Carlo partial melting and random mixing model isotopic composition predictions compared with data from the TIS. Model predictions are shown as solid circles; data from the TIS are shown as open circles. The model predicts that the partial melting and random mixing process can create considerable scatter about two-component isotopic mixing trends without the involvement of additional source components. Assumed source compositions: basalt 87 Sr/86Sri ¼ 0.7047, 143Nd/144Ndi ¼ 0.51259, Sr ¼ 555 ppm, Nd ¼ 24 ppm; crustal 87Sr/86Sri ¼ 0.7068, 143 Nd/144Ndi ¼ 0.51212, Sr ¼ 475, Nd ¼ 19 ppm (Kistler et al. 1986).

198

W. GRAY ET AL.

result, but if Sr/Nd ratios in the melts are modified by partial melting, the curvature changes and random mixtures plot away from the main trend. Thus, partial melting followed by random mixing can create significant isotopic scatter even with only two end members. The amount of scatter predicted by the model is not as great as that observed in the TIS (Fig. 11b), but a small amount of variability in the crustal end member would increase the scatter.

Time-space patterns of geochemical variation Major- and trace-element modelling suggests that both partial melting and mixing are required to explain the chemical evolution of the TIS because neither process alone can account for all of the observed trends. However, although clear trends are absent from individual units, the TIS as a whole becomes more felsic inward. Further, initial Sr and Nd isotopic compositions are systematically more evolved passing inward. Thus, the magma mixtures which formed the interior units contained progressively larger proportions of a crustal, felsic and isotopically evolved end member. The TIS also systematically decreases in age inward (Coleman et al. 2004), and therefore the proportion of the crustal component increased with time (Gray 2003). This pattern is difficult to reconcile with a purely random mixing model, and therefore the emplacement process may not have been as random as implied by the modelling. We propose that individual melt aliquots were generated in the upper mantle and lower crust and buoyantly separated from their sources; mantleand crust-derived components were then mixed at the crustal source and/or during ascent to the upper crust, and incrementally assembled and consolidated in the upper crust. The systematic increase with time of a crustal magmatic component may reflect progressive heating of the lower-crustal source region by prolonged, repeated injection of mantle-derived mafic magma (Kemp et al. 2007). Heat accumulation in the lower-crustal magma source would cause later mafic injections to raise an increasing volume of surrounding lower crust above its solidus, and thus make an increasing volume of crustal melt available to mix and mingle with the mantle-derived component (Annen et al. 2006).

Significance of mapped plutonic units The porphyritic facies of the Half Dome Granodiorite is geochemically and isotopically nearly identical to the Cathedral Peak Granodiorite, and Kistler & Fleck (1994) argued that the two therefore should be reassigned to the same rock unit.

However, the geochemical and isotopic properties of the equigranular Half Dome also overlap both of those units such that the only consistent differences between the three units are textural (Johnson et al. 2006). Further, contacts between all three rock units are commonly gradational in the field, and geochronologic evidence indicates progressive assembly of the three units over a period of several Ma (Coleman et al. 2004; Matzel et al. 2005). We therefore infer that the porphyritic Half Dome shares petrological properties with the equigranular Half Dome and the Cathedral Peak units because the three form a single evolutionary continuum. Mapping such a continuum as three distinct rock units arises from imposition of the discrete rock unit definitions needed for field mapping onto a continuous range of petrological variation produced by a protracted incremental process. Such arbitrary unit definitions are routinely used as heuristic tools for mapping continuously variable rocks of all sorts, but it is vitally important not to confuse mapping conventions with actual geological phenomena. The question of whether the porphyritic Half Dome would be better assigned to the same rock unit as the Cathedral Peak thus appears to have little fundamental significance other than field convenience. Nonetheless, the anomalously large feldspar grains in a rock that otherwise resembles typical Half Dome Granodiorite is intriguing. One possible explanation is that the porphyritic facies represents a mixture between the equigranular Half Dome and Cathedral Peak Granodiorites, a hypothesis based largely on the presence of both K-spar megacrysts characteristic of the Cathedral Peak and euhedral hornblende crystals characteristic of the equigranular Half Dome (Wallace & Bergantz 2002). This alternative is contradicted both by geochronological data indicating that the granodiorites differ in age sufficiently that they never were molten at the same time, and by the observation that rocks with megacrysts lack most of the smaller crystals that are present in non-megacrystic units. Textural coarsening offers an alternative explanation (Higgins 1999; Johnson et al. 2006). Heating of inner parts of the Half Dome Granodiorite by emplacement of the adjacent Cathedral Peak Granodiorite, and flushing the system with water released during crystallization of water-rich magmas (Sisson & Layne 1993; Carmichael 2002), may have resulted in textural coarsening of adjacent Half Dome Granodiorite. Another related alternative is that the textural changes that define the transitions from equigranular Half Dome to porphyritic Half Dome to Cathedral Peak represent thermal evolution over time (Coleman et al. 2006). The equigranular Half Dome contains kilometre-scale lithological cycles that we interpret to record an

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE

earlier evolutionary stage during which the magma system periodically froze up to a significant degree; the megacrystic Cathedral Peak seems to lack lithological cycles and probably represents a later stage when intergranular melt was continuously present. In this interpretation, the textural features that define the three mapped units reflect the evolving thermal state of the TIS and do not relate directly to intrusive events that assembled the TIS.

Summary Rocks of the TIS are metaluminous, highpotassium, calc-alkaline granitoids, identical to many other plutonic suites within the SNB. Mineralogy and thermobarometry indicate hornblende – plagioclase and zircon saturation temperatures of 660 –770 8C at c. 6 km depth. Evidence of lowtemperature subsolidus exsolution is observed in compositions of feldspars, hornblende, biotite and magnetite. Although most major elements display strong linear trends with SiO2, scatter of trace-element data and variability of initial Sr and Nd isotopic ratios are inconsistent with fractional crystallization as the predominant process responsible for its chemical evolution. Monte Carlo simulation of randomly mixed melts derived by varying degrees of partial melting of a mafic crustal source indicates that such a mixing process could produce the observed major- and trace-element trends. Isotopic data, however, suggest mixing between melts of mantle-like rocks and a granitic melt similar in composition to the Johnson Granite Porphyry. The sources cannot be identified with certainty, but potential candidates include mantle peridotite, lower crustal diorite and older granitoids including the Sentinel Granodiorite and intrusive suite of Yosemite Valley. Isotopic data from the granodiorite of Kuna Crest are distinct from other TIS units and plot on a trend that corresponds to the Sentinel Granodiorite and intrusive suite of Yosemite Valley. This suggests that (1) the Kuna Crest is isotopically more closely related to these rocks than to other units of the TIS, or (2) more than two magma sources were involved in the chemical evolution of the TIS. The remaining units appear to share a common but more negative Sr–Nd isotopic trend more consistent with twocomponent mixing. The general temporal trend toward more evolved isotopic compositions may result from thermal maturation of the source region (Annen et al. 2006). The petrochemical and isotopic properties of the equigranular Half Dome Granodiorite, porphyritic Half Dome Granodiorite and Cathedral Peak Granodiorite largely overlap one another and their texturally defined contacts are generally gradational. We thus interpret these map units to

199

represent a single petrological continuum rather than distinct intrusive phases. The textural features that define the map units may reflect thermal evolution of the system rather than indicating distinct intrusive events during its assembly. The TIS shares time – space – composition features common to most large zoned intrusive suites. Consequently, it seems likely that processes responsible for the generation of the TIS may have operated in other systems as well. If geochronological data from other zoned suites reveal long (millions of years) intrusive histories, traditional ideas regarding the evolution of such suites will need to be abandoned in favour of geologically plausible scenarios. This work has been supported by grants from the National Science Foundation (EAR-9526803, EAR-9814787, EAR9814788, EAR-9814789, EAR-0336070, EAR-0337351, EAR-053812, EAR-0538094) and by student grants from the Geological Society of America, the University of California’s White Mountain Research Station and the University of North Carolina Martin-McCarthy fund. The manuscript was greatly improved by incorporation of suggestions from reviewers B. Jicha, C. Miller and B. Scaillet. We gratefully acknowledge generous cooperation and logistical support from the US National Park Service, US Forest Service and US Geological Survey. In particular, our work has been greatly aided by T. Frost, T. Sisson, R. Kistler, J. van Wagtendonk and P. Moore.

References A GUE , J. J. & B RIMHALL , G. H. 1988. Regional variations in bulk chemistry, mineralogy, and the compositions of mafic and accessory minerals in the batholiths of California. Geological Society of America Bulletin, 100, 891 –911. A LLEN , C. M. 1991. Local equilibrium of mafic enclaves and granitoids of the Turtle Pluton, Southeast California; mineral, chemical, and isotopic evidence. American Mineralogist, 76, 574– 588. A NDERS , E. & G REVESSE , N. 1989. Abundances of the elements; meteoritic and solar. Geochimica et Cosmochimica Acta, 53, 197– 214. A NDERSON , J. L. & S MITH , D. R. 1995. The effects of temperature on the Al-in-hornblende barometer. American Mineralogist, 80, 549– 559. A NNEN , C., B LUNDY , J. D. & S PARKS , R. S. J. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology, 47, 505– 539. B ARBARIN , B. 1991. Enclaves of the Mesozoic calc-alkaline granitoids of the Sierra Nevada batholith, California. In: D IDIER , J. & B ARBARIN , B. (eds) Enclaves and Granite Petrology. Elsevier, New York, 135–153. B ATEMAN , P. C. 1992. Plutonism in the Central Part of the Sierra Nevada Batholith, California. US Geological Survey Professional Papers, 1483. B ATEMAN , P. C. & C HAPPELL , B. W. 1979. Crystallization, fractionation, and solidification of the Tuolumne

200

W. GRAY ET AL.

intrusive series, Yosemite National Park, California. Geological Society of America Bulletin, 90, 465– 482. B ATEMAN , P. C. & E ATON , J. P. 1967. Sierra Nevada batholith. Science, 158, 1407–1417. B ATEMAN , P. C., C HAPPELL , B. W., K ISTLER , R. W., P ECK , D. L. & B USACCA , A. J. 1988. Tuolumne Meadows Quadrangle, California; analytic data. US Geological Survey Bulletin, 1819, 43. B ATEMAN , P. C., K ISTLER , R. W., P ECK , D. L. & B USACCA , A. J. 1983. Geologic Map of the Tuolumne Meadows Quadrangle, Yosemite National Park, California. US Geological Survey Map GQ-1570, scale 1:62,500. B EARD , B. L. & G LAZNER , A. F. 1995. Trace element and Sr and Nd isotopic composition of mantle xenoliths from the Big Pine volcanic field, California. Journal of Geophysical Research, 100, 4169–4179. B ENCE , A. E. & A LBEE , A. L. 1968. Empirical correction factors for the electron microanalysis of silicates and oxides. Journal of Geology, 76, 382–403. B LUNDY , J. D. & H OLLAND , T. J. B. 1990. Calcic amphibole equilibria and a new amphibole – plagioclase geothermometer. Contributions to Mineralogy and Petrology, 104, 208– 224. B ROWN , W. L. & P ARSONS , I. 1989. Alkali feldspars: ordering rates, phase transformations and behavior diagrams for igneous rocks. Mineralogical Magazine, 53, 25– 42. B UDDINGTON , A. F. 1959. Granite emplacement with special reference to North America. Geological Society of America Bulletin, 70, 671–747. C ALKINS , F. C. 1930. The Granitic Rocks of the Yosemite Region. US Geological Survey Professional Papers, 160, 120–129. C ARMICHAEL , I. S. E. 2002. The andesite aqueduct; perspectives on the evolution of intermediate magmatism in west-central (105– 99 degrees W) Mexico. Contributions to Mineralogy and Petrology, 143, 641– 663. C HEN , J. H. & M OORE , J. G. 1982. Uranium-lead isotopic ages from the Sierra Nevada batholith, California. Journal of Geophysical Research, 87, 4761–4784. C HEN , J. H. & T ILTON , G. R. 1991. Applications of lead and strontium isotopic relationships to the petrogenesis of granitoid rocks, central Sierra Nevada batholith, California. Geological Society of America Bulletin, 103, 439–447. C OLEMAN , D. S. & G LAZNER , A. F. 1997. The Sierra crest magmatic event: rapid formation of juvenile crust during the Late Cretaceous in California. International Geology Review, 39, 768 –787. C OLEMAN , D. S., B ARTLEY , J. M., G LAZNER , A. F. & J OHNSON , B. R. 2006. Incremental growth and consolidation of the Half Dome Granodiorite, Tuolumne Intrusive Suite. EOS, Transactions of the American Geophysical Union, 87, Abstract V22A-08. C OLEMAN , D. S., G RAY , W. & G LAZNER , A. F. 2004. Rethinking the emplacement and evolution of zoned plutons: geochronologic evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology, 32, 433–436. C OSCA , M. A., E SSENE , E. J. & B OWMAN , J. R. 1991. Complete chemical analyses of metamorphic hornblendes; implications for normalizations, calculated H2O

activities, and thermobarometry. Contributions to Mineralogy and Petrology, 108, 472– 484. C ZAMANSKE , G. K. & W ONES , D. R. 1973. Oxidation during magmatic differentiation, Finnmarka Complex, Oslo Area, Norway; Part 2, the mafic silicates. Journal of Petrology, 14, 349–380. C ZAMANSKE , G. K., I SHIHARA , S. & A TKIN , S. A. 1981. Chemistry of rock-forming minerals of the Cretaceous – Paleocene batholith in southwestern Japan and implications for magma genesis. Journal of Geophysical Research, 86, 10 431– 10 469. D ENIEL , C., V IDAL , P., F ERNANDEZ , A., F ORT , P. & P EUCAT , J.-J. 1987. Isotopic study of the Manaslu granite (Himalaya, Nepal): inferences on the age and source of Himalayan leucogranites. Contributions to Mineralogy and Petrology, 96, 78–92. D E P AOLO , D. J. 1981. A neodymium and strontium isotopic study of the Mesozoic calc-alkaline granitic batholiths of the Sierra Nevada and Peninsular ranges, California. Journal of Geophysical Research, 86, 10 470–10 488. D E P AOLO , D. J. & W ASSERBURG , G. J. 1979. Petrogenetic mixing models and Nd– Sr isotopic patterns. Geochimica et Cosmochimica Acta, 43, 615– 627. D ODGE , F. C. W., P APIKE , J. J. & M AYS , R. E. 1968. Hornblendes from granitic rocks of the central Sierra Nevada batholith, California. Journal of Petrology, 9, 378–410. D OE , B. R. & D ELEVAUX , M. H. 1973. Variations in lead– lead isotopic compositions in Mesozoic granitic rocks of California: a preliminary investigation. Geological Society of America Bulletin, 84, 3513– 3526. D OE , B. R. & Z ARTMAN , R. E. 1979. Plumbotectonics, the Phanerozoic. In: B ARNES , H. L. (ed.) Geochemistry of Hydrothermal Ore Deposits. WileyInterscience, New York, 22–70. D UCEA , M. N. & S ALEEBY , J. B. 1998. The age and origin of a thick mafic–ultramafic keel from beneath the Sierra Nevada batholith. Contributions to Mineralogy and Petrology, 133, 169 –185. E LKINS , L. T. & G ROVE , T. L. 1990. Ternary feldspar experiments and thermodynamic models. American Mineralogist, 75, 544–559. E VERNDEN , J. F. & K ISTLER , R. W. 1970. Chronology of Emplacement of Mesozoic Batholithic Complexes in California and Western Nevada. US Geological Survey Professional Papers, 623. F REY , F. A., C HAPPELL , B. W. & R OY , S. D. 1978. Fractionation of rare-earth elements in the Tuolumne Intrusive Series, Sierra Nevada batholith, California. Geology, 6, 239–242. G LAZNER , A. F., B ARTLEY , J. M., C OLEMAN , D. S., G RAY , W. & T AYLOR , R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today, 14, 4– 11. G LAZNER , A. F., C OLEMAN , D. S. & B ARTLEY , J. M. 2008. Tenuous connection between high-silica rhyolites and granodiorite plutons. Geology, 36, 183– 186. G RAY , W. M. 2003. Chemical and thermal evolution of the Late Cretaceous Tuolumne Intrusive Suite, Yosemite National Park, California. PhD thesis, University of North Carolina at Chapel Hill, NC. H IBBARD , M. J. 1995. Petrography to Petrogenesis. Prentice-Hall, Englewood Cliffs, NJ.

EVOLUTION OF TUOLUMNE INTRUSIVE SUITE H IGGINS , M. D. 1999. Origin of Megacrysts in Granitoids by Textural Coarsening; a Crystal Size Distribution (CSD) Study of Microcline in the Cathedral Peak Granodiorite, Sierra Nevada, California. Geological Society Special Publications, 168, 207– 219. H OLLAND , T. J. B. & B LUNDY , J. 1994. Non-ideal interactions in calcic amphiboles and their bearing on amphibole – plagioclase thermometry. Contributions to Mineralogy and Petrology, 116, 433– 447. J OHANNES , W. 1989. Melting of plagioclase – quartz assemblages of 2 kbar water pressure. Contributions to Mineralogy and Petrology, 103, 270– 276. J OHNSON , B. R., G LAZNER , A. F. & C OLEMAN , D. S. 2006. Significance of K-feldspar Megacryst Size and Distribution in the Tuolumne Intrusive Suite, California. Geological Society of America Abstracts with Programs, 38, 93. K EMP , A. I. S., H AWKESWORTH , C. J. ET AL . 2007. Magmatic and crustal differentiation history of granitic rocks from Hf–O isotopes in zircon. Science, 315, 980–983. K ERRICK , D. M. 1969. K-feldspar megacrysts from a porphyritic quartz monzonite, central Sierra Nevada, California. American Mineralogist, 54, 839– 848. K ISTLER , R. W. & F LECK , R. J. 1994. Field Guide for a Transect of the Central Sierra Nevada, California; Geochronology and Isotope Geology. US Geological Survey Open File Report, 94-0267. K ISTLER , R. W., C HAPPELL , B. W., P ECK , D. L. & B ATEMAN , P. C. 1986. Isotopic variation in the Tuolumne intrusive suite, central Sierra Nevada, California. Contributions to Mineralogy and Petrology, 94, 205 –220. L EAKE , B. E. 1978. Nomenclature of amphiboles. American Mineralogist, 63, 1023– 1052. M ATZEL , J., M UNDIL , R., P ATERSON , S., R ENNE , P. & N OMADE , S. 2005. Evaluating Pluton Growth Models using High Resolution Geochronology: Tuolumne Intrusive Suite, Sierra Nevada, California. Geological Society of America Abstracts with Programs, 37, 131. P IN , C., B INON , M., B ELIN , J. M., B ARBARIN , B. & C LEMENS , J. D. 1990. Origin of microgranular enclaves in granitoids: equivocal Sr-Nd evidence from Hercynian rocks in the Massif Central (France). Journal of Geophysical Research, 95, 17 281– 17 828. P ITCHER , W. S. 1993. The Nature and Origin of Granite. Chapman & Hall, London. R ATAJESKI , K. 1999. Field, Geochemical and Experimental Study of Mafic to Felsic Plutonic Rocks Associated with the Intrusive Suite of Yosemite Valley, California. PhD thesis, University of North Carolina at Chapel Hill, NC. R ATAJESKI , K., G LAZNER , A. F. & M ILLER , B. V. 2001. Geology and geochemistry of mafic and felsic plutonic rocks in the Cretaceous intrusive suite of Yosemite Valley, California. Geological Society of America Bulletin, 113, 1486– 1502. R ATAJESKI , K., S ISSON , T. W. & G LAZNER , A. F. 2005. Experimental and geochemical evidence for derivation of the El Capitan Granite, California, by partial

201

melting of hydrous gabbroic lower crust. Contributions to Mineralogy and Petrology, 149, 713–734. R EID , J. B., E VANS , O. C. & F ATES , D. G. 1983. Magma mixing in granitic rocks of the central Sierra Nevada, California. Earth and Planetary Science Letters, 66, 243– 261. S CAILLET , B. & E VANS , B. W. 1999. The 15 June 1991 eruption of Mount Pinatubo. I. Phase equilibria and pre-eruption P – T– fO2 –fH2O conditions of the dacite magma. Journal of Petrology, 40, 381–411. S ISSON , T. W. & L AYNE , G. D. 1993. H2O in basalt and basaltic andesite glass inclusions from four subductionrelated volcanoes. Earth and Planetary Science Letters, 117, 619– 635. S ISSON , T. W., R ATAJESKI , K., H ANKINS , W. B. & G LAZNER , A. F. 2005. Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology, 148, 635 –661. S TERN , T. W., B ATEMAN , P. C., M ORGAN , B. A., N EWELL , M. F. & P ECK , D. L. 1981. Isotopic U– Pb Ages of Zircons from the Granitoids of the Central Sierra Nevada. US Geological Survey Professional Papers, 1071. T ITUS , S. J., C LARK , R. & T IKOFF , B. 2005. Geologic and geophysical investigation of two fine-grained granites, Sierra Nevada Batholith, California; evidence for structural controls on emplacement and volcanism. Geological Society of America Bulletin, 117, 1256– 1271. W ALAWENDER , M. J., G ASTIL , R. G. ET AL . 1990. Origin and Evolution of the Zoned La Posta-type Plutons, Eastern Peninsular Ranges Batholith, Southern and Baja California. Geological Society of America Memoirs, 174, 1 –18. W ALLACE , G. S. & B ERGANTZ , G. W. 2002. Waveletbased correlation (WBC) of zoned crystal populations and magma mixing. Earth and Planetary Science Letters, 202, 133– 145. W ATSON , E. B. & H ARRISON , T. M. 1983. Zircon saturation revisited; temperature and composition effects in a variety of crustal magma types. Earth and Planetary Science Letters, 64, 295–304. W HITNEY , J. A. 1988. The origin of granite: the role and source of water in the evolution of granitic magmas. Geological Society of America Bulletin, 100, 1886– 1897. W ONES , D. R. 1989. Significance of the assemblage titanite þ magnetite þ quartz in granitic rocks. American Mineralogist, 74, 744– 749. W ONES , D. R. & G ILBERT , M. C. 1982. Chapter 3, Amphiboles in the igneous environment; Introduction. In: V EBLEN , D. R. & R IBBE , P. H. (eds) Amphiboles; Petrology and Experimental Phase Relations (Reviews in Mineralogy 9B), Mineralogical Society of America, Washington, DC, 355– 357. Z AK , J., P ATERSON , S. R. & M EMETI , V. 2007. Four magmatic fabrics in the Tuolumne batholith, central Sierra Nevada, California (USA): implications for interpreting fabric patterns in plutons and evolution of magma chambers in the upper crust. Geological Society of America Bulletin, 119, 184–201.

Construction, solidification and internal differentiation of a large felsic arc pluton: Cathedral Peak granodiorite, Sierra Nevada Batholith S. D. BURGESS & J. S. MILLER Department of Geology, San Jose State University, 1 Washington Square, San Jose, CA 95192, USA (e-mail: [email protected]) Abstract: The Tuolumne Batholith (TB), Sierra Nevada Batholith (USA), is an archetypal large, zoned arc intrusion (c. 1200 km2). Previous work proposed that compositional zonation observed in the TB was produced in-situ by inward differentiation of a large magma chamber and/or largescale, intrachamber magma mixing. Recent geochronology shows that the TB was intruded over 8– 9 Ma, making single pulse fractionation or mixing in a magma chamber of TB dimensions unlikely. We examine processes responsible for compositional variation in the Cathedral Peak Granodiorite, which is the largest mapped unit of the TB. New field, geochemical and geochronological work along a roughly contact-perpendicular 5 km transect indicates: (1) magmatic foliation is steeply-dipping (.608); (2) field evidence for repeated separation of crystals from melt and local magma mixing is observed; (3) U –Pb zircon ages at opposing ends of the transect are indistinguishable within error (c. 87.5 Ma); (4) bulk composition varies only modestly but trace elements show variable degrees of scatter; (5) 1Nd(t) and 87Sr/86Sr(i) have small variation compared with that in the whole TB. Geochemical and isotopic data are compatible with fractionation of major silicates and accessory minerals. However, the geochemical spatial variation, minor isotopic variation and field evidence suggest that fractionation was highly disorganized and also involved mixing with new input magma and remobilization of crystal mush as the pluton solidified. Our observations are consistent with the construction of a large and dynamic magma system within the last c. 1 Ma of TB growth.

Compositionally zoned igneous intrusions have a direct bearing on continental crust construction and magma genesis processes, and they also provide natural laboratories for studying magma systems. Despite extensive research over the past few decades, many questions remain in understanding the petrogenesis of zoned intrusions: (1) are zoned intrusions an amalgamation of smaller plutons or are they formed from solidification of single, large magma chambers? (2) If multiple intrusions are involved, how is the compositional zoning developed and how large are the intrusive pulses and increments? (3) To what extent do discrete intrusions interact and how closely spaced in time are successive magma pulses? Most commonly, these intrusions are normally zoned, meaning that they are typically most mafic at their outer margins and increasingly more felsic inward. Classically, such zonation has been attributed to fractional crystallization, restite unmixing or hybridization of compositionally variable magmas (e.g. Compton 1955; Vance 1961; Ragland & Butler 1972; Chappell & White 1974; Bateman & Nokleberg 1978; Bateman & Chappell 1979; McCarthy & Groves 1979; Halliday et al. 1980; Hill 1988; Hill et al. 1988; Walawender et al. 1990; Sawka et al. 1990; Bateman 1992; Reid et al.

1993). These processes are inferred to operate in-situ and on relatively large batches of magma. However, more recent work in some zoned arc plutons suggests that they may be constructed from successive emplacement of numerous pulses or increments over time spans that are likely to be longer than the lifetime of any single magma chamber (Coleman et al. 2004; Glazner et al. 2004). In this case, magma chamber processes are inferred to be second-order processes that modify magma increments, whose bulk compositions are largely imparted at the magma source. The Tuolumne Batholith (TB) in eastern California, USA is one such zoned intrusion which has been studied for the past three decades in some detail, with much current work aimed at answering the above questions (Bateman 1992; Reid et al. 1993; Gray 2003; Coleman et al. 2004; Glazner et al. 2004; Zak & Paterson 2005; Fig. 1). The TB is enigmatic because it has a seemingly simple zoned and nested map pattern but in detail displays complicated and variable contact relations, and highly variable trace element and isotope geochemistry. Although end member hypotheses have been developed for TB petrogenesis, detailed chemical and isotopic data are still lacking in many areas, and challenges remain in correlating

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 203–233. DOI: 10.1144/SP304.11 0305-8719/08/$15.00 # The Geological Society of London 2008.

204

S. D. BURGESS & J. S. MILLER

Fig. 1. (a) Generalized map of the USA denoting California and the Sierra Nevada Batholith. (b) Map of the Sierra Nevada Batholith delineating major intrusive suites. (c) Expanded map of the Tuolumne Batholith showing the study area and including previously published U– Pb zircon ages of TB units.

chemical, isotopic and field-based data sets. Additional challenges remain in quantitatively describing the size, number and location of any pulses other than those defined by major unit boundaries. In particular, in order to evaluate the hypothesis that the TB was constructed in small increments, it is necessary to conduct studies at a relatively small spatial scale and to conduct more detailed and high-precision geochronology.

Geological setting The Tuolumne Batholith is a 93.5– 85.4 Ma felsic to intermediate zoned intrusion that is crudely elliptical and elongated in a NNW–SSE trend. It covers an area of approximately 1200 km2 within the eastcentral part of the much larger Jurassic-Cretaceous Sierra Nevada Batholith of western North America (Fig. 1). The units of the TB are progressively younger inward (Stern et al. 1981; Kistler & Fleck 1994; Fleck et al. 1996; Coleman & Glazner 1997; Coleman et al. 2004). From outer to inner, these units are: Kuna Crest Granodiorite, Half Dome Granodiorite (equigranular and then

porphyritic), Cathedral Peak Granodiorite and Johnson Granite porphyry (Bateman & Chappell 1979). Collectively, the units of the TB show a gradation from mafic-intermediate to felsic and generally a regular inward increase in SiO2, Na2O and K2O, and decrease in Al2O3, CaO, FeO and MgO (Bateman et al. 1988). Bateman & Chappell (1979) originally suggested that the entire Tuolumne Batholith (TB) was emplaced as one large intrusion that differentiated largely via fractional crystallization to produce the smooth chemical variation observed. However, inner units cut sharply across preceding units in some areas, whereas in other areas the contacts are gradational. Bateman & Chappell (1979) suggested that these variable contact relations were caused by the cutting out of earlier solidified or partially solidified rocks as the melt-rich core surged upward and outward. Systematic variation in initial 87Sr/86Sr and 143 Nd/144Nd v. SiO2 within the TB was later documented by Kistler et al. (1986), who suggested that mixing between low 87Sr/86Sri, high143Nd/144Nd mafic magmas and high 87Sr/86Sri, low 143Nd/ 144 Nd felsic melts resulted in magmas of

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

intermediate composition, which were then modified by closed-system fractional crystallization. A mixing origin was also favoured by Reid et al. (1993) because, in addition to the isotopic variability observed by Kistler et al. (1986), the most obvious cumulate rocks observed in the TB (ladder ‘dykes’ and mafic schlieren) constitute only a minor fraction of the whole intrusion, and their chemical compositions generally are not compatible with derivation by fractionation from the main TB units. More recently, Coleman et al. (2004) and Glazner et al. (2004) concluded that the 8 –10 Ma interval of intrusion, and the fairly ‘monotonic’ inward decrease in age make shallow, singlechamber differentiation and/or mixing impossible. As an alternative, these workers proposed that the TB was constructed and grew incrementally by many small-volume inputs into an ephemeral, central magma chamber that produced the observed monotonic decrease in age. They also suggested that the chemical and isotopic variability is the result of heterogeneity in the source and/or variations in melt fraction and residual mineralogy.

Methods Sampling strategy In this study we have focused efforts solely on the interior, volumetrically dominant (at the level of exposure) Cathedral Peak unit. This constrains our interpretations to this part of the TB, but does not preclude their applicability elsewhere in the batholith. We conducted two east– west sampling transects within the Cathedral Peak Granodiorite, extending from near the contact with the Johnson Granite porphyry to the outer-most Cathedral Peak exposures nearest the contact with porphyritic Half Dome Granodiorite (Fig. 2). The transects run roughly perpendicular to the contacts between these three units, and provide a data set that documents within-unit and across-unit chemical, compositional, petrographic and isotopic variation. Detailed field observations and descriptions were made along and adjacent to these transects and focused on schlieren, enclaves, xenoliths, felsic sheets, dykes, mafic troughs and magmatic foliation. Samples were collected at approximately 500 m increments along each transect, and additional samples were taken in cases where internal contacts or significant rock heterogeneities were noted.

205

sample homogeneity. Splits of each sample were then processed at the Washington State University GeoAnalytical Laboratory for major (Table 1) and trace elements by XRF (Table 2) and ICP-MS (Table 3) (see: www.wsu.edu/~geology/Pages/ Services/Geolab.html for complete XRF and ICP-MS procedures). The reported values were checked for accuracy and precision using replicate splits of samples and standards and reproducibility and internal consistency of trace elements using different methods was checked by comparison of elements analysed by both XRF and ICP-MS (e.g. Ba, Sr, Zr). Reproducibility is better than 1% of the amount present for major elements and better than 5% of the amount present for trace elements. Fourteen whole rock samples were also analysed for Sr and Sm –Nd isotopes at the University of North Carolina, Chapel Hill (Table 4). Two hundred milligrams of rock powder from each sample was spiked with mixed 150Nd– 147Sm spike and dissolved in a mixture of HF and HNO3 in a sealed Teflon bomb at 220 8C for 5–7 days. After conversion to chlorides, separation of Sm and Nd was accomplished using a two-column REE separation and RE-spec and LN-spec resin. Strontium was separated using Sr-spec column chemistry and HNO3. Neodymium was loaded on single Re filaments with dilute HCl, and Sm was loaded on single Ta filaments with H3PO4. Strontium was loaded with H3PO4 and TaCl5 emitter on a Re filament. All analyses were performed on the Micromass Sector-54 mass spectrometer at UNC. Neodymium was analysed in dynamic multicollector mode as NdO using an oxygen bleed valve at 1V, and Sm was analysed in static multicollector mode with 147Sm ¼ 200 mV. Strontium was analysed in dynamic multicollector mode with 88 Sr ¼ 3V. Neodymium data are normalized to 146 Nd/144Nd ¼ 0.7219. Strontium data are normalized to 86Sr/88Sr ¼ 0.1194. Replicate analysis of SRM-987 yielded 87Sr/86Sr ¼ 0.710246 + 0.000010. Replicate analysis of the UNC J–Nd standard during the period of analysis gave 143 Nd/144Nd ¼ 0.512099 + 0.000005, and data are referenced to this standard, which is in turn referenced to La Jolla Nd (143Nd/144Nd ¼ 0.511853), which is run less often. Initial Sr ratios were calculated using l87Rb ¼ 1.42  10211 year21, and using Rb and Sr concentrations measured by ICP-MS.

Mineral chemistry of plagioclase Analytical methods A total of 27 samples were analysed for whole-rock and trace-element compositions. For coarse-grained samples, including those with large K-feldspar crystals, an excess of 10 kg was crushed to ensure

Plagioclase compositions were measured using a Jeol JXA-8800L electron microprobe at the United States Geological Survey in Menlo Park, CA (see supplemental table 1, available online at http://www.geolsoc.org.uk/SUP18321). Operating

206

S. D. BURGESS & J. S. MILLER

Fig. 2. Sample locations in the study area. Each diamond represents the location of a field station and a sample processed for whole rock/trace element geochemistry and/or isotopic analysis. The two circles in the Cathedral Peak unit represent samples analysed for U–Pb geochronology. Irregular grey polygons represent zones of abundant schlieren.

conditions were 15 kV accelerating voltage and 10 nA beam current with a 2 mm beam diameter. Matrix effects were corrected by using CITZAF data reduction. Synthetic and natural standards were used, and included Tiburon Albite (Si, Na), Barite (Ba), ORIA (K), Anorthite 100 (Al, Ca) and RGSC (Mg).

Geochronology Two samples of Cathedral Peak granodiorite were collected from either end of the northern sampling

transect for U/Pb zircon geochronology (Table 5, Fig. 2). Single zircon crystals were hand-picked in ethanol from bulk fractions and cleaned in distilled HNO3. All analysed grains were aggressively abraded using standard air abrasion for over 72 h. Zircon weights were estimated using a video camera and scale, and are known to within 15%, then digested using 29 M HF. Uranium and Pb were spiked with a mixed 205Pb – 233U – 236U tracer, separated, and then loaded onto zone-refined rhenium filaments using clean H3PO4 and silica gel. All analyses were accomplished with a Micromass

Table 1. Major element oxide concentration determined by X-ray fluorescence Sample no.

Megacryst-poor Gdt Gdt Enclave Gdt Granitic dike Gdt Gdt Schlieren Schlieren Gdt Gdt Mafic xenolith Aplite dike Leucocratic Gdt Leucocratic Gdt with Sch Gdt Gdt Gdt Gdt Gdt Gdt Gdt Gdt Gdt Gdt Gdt Gdt

Weight % oxide SiO2

TiO2

Al2O3

FeO

MnO

MgO

CaO

Na2O

K2O

P2O5

70.9 70.5 60.3 70.9 74.9 71.0 70.4 48.3 62.5 69.5 70.4 67.9 77.6 76.8 71.1 73.6 71.8 71.0 69.8 70.9 70.0 68.7 69.8 70.3 67.7 68.9 70.1

0.349 0.446 0.845 0.345 0.198 0.351 0.390 2.815 1.185 0.432 0.410 0.332 0.062 0.179 0.380 0.234 0.327 0.362 0.499 0.415 0.445 0.428 0.415 0.435 0.533 0.475 0.469

15.3 15.1 17.7 15.4 13.9 15.2 15.5 9.91 14.8 15.7 15.3 15.6 12.7 12.6 15.0 14.5 15.0 15.2 15.3 15.0 15.5 16.1 15.5 15.3 16.2 15.8 15.1

2.1 2.7 5.9 2.1 1.2 2.2 2.3 21.0 7.7 2.6 2.4 2.6 0.4 0.9 2.5 1.3 1.9 1.9 3.0 2.6 2.6 2.6 2.5 2.5 3.1 2.8 2.7

0.054 0.066 0.171 0.053 0.030 0.055 0.051 0.378 0.170 0.055 0.055 0.100 0.017 0.025 0.073 0.036 0.051 0.059 0.066 0.060 0.059 0.055 0.055 0.057 0.065 0.066 0.058

0.596 0.746 2.090 0.620 0.318 0.664 0.694 2.150 2.220 0.728 0.743 1.640 0.049 0.202 0.713 0.377 0.517 0.604 0.821 0.773 0.798 0.885 0.707 0.749 0.954 0.806 0.726

2.41 2.96 4.49 2.44 1.74 2.52 2.95 4.05 3.39 2.91 2.74 1.20 0.691 1.10 2.17 2.00 2.30 2.47 3.03 3.06 3.04 3.05 2.85 2.89 3.44 3.11 2.98

4.23 4.36 5.42 4.26 3.68 4.15 4.41 2.42 3.75 4.32 4.21 3.21 3.80 2.96 4.44 4.08 4.31 4.31 4.24 4.24 4.32 4.03 4.19 4.32 4.20 4.27 4.03

4.00 2.97 2.69 3.75 4.02 3.75 3.11 2.91 3.87 3.59 3.58 7.34 4.63 5.22 3.51 3.83 3.69 3.97 3.15 2.86 3.10 4.02 3.79 3.24 3.60 3.58 3.58

0.141 0.190 0.406 0.133 0.079 0.143 0.162 0.500 0.404 0.183 0.166 0.109 0.011 0.058 0.127 0.077 0.120 0.142 0.193 0.171 0.182 0.166 0.165 0.174 0.213 0.198 0.195

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

BTL 9 BTL 10 BTL 18 BTL 25 BTL 27 BTL 28 BTL 30 BTL 31A BTL 31B BTL 36 BTL 37 BTL 39 BTL 41 BTL 42A BTL 42B BTL 43 BTL 45 BTL 48 BTL 55 BTL 56 BTL 58 BTL 59 BTL 68 BTL 69 BTL 79 BTL 81 BTL 82

Type

Gdt ¼ granodiorite, Sch ¼ schlieren.

207

208

Table 2. Unnormalized trace element concentrations determined by X-ray fluorescence Sample

0.81 0.83 3.30 3.50 1.20 0.78 3.80 0.54 2.0 0.68 2.30 3.90 4.0 4.70 3.20 4.30 3.30 2.50 3.80 2.50 4.30 5.30 6.0 3.60 2.10 3.20 2.20

0.3 0.6 1.7 23.0 0.2 0.0 0.0 11.0 4.2 0.0 2.0 48.0 1.7 0.9 1.6 1.5 0.0 1.1 2.6 0.0 1.9 24.0 0.0 3.3 0.0 0.0 0.0

3.1 3.7 13.0 3.3 2.1 3.8 3.5 14.0 9.3 3.8 3.6 6.6 0.9 2.5 4.7 1.7 2.5 4.2 3.8 3.6 4.0 4.1 4.2 2.9 4.2 4.5 4.3

36 46 100 35 21 38 40 400 130 42 42 34 5.9 15 38 23 32 31 50 43 44 45 40 44 54 50 47

850 557 269 879 297 597 609 246 451 945 787 1120 10 746 166 607 519 749 572 410 609 1182 993 631 1170 777 872

149 120 179 142 162 147 114 148 192 115 142 392 232 210 186 137 166 135 131 128 126 126 136 134 121 138 127

612 649 546 599 415 608 684 235 487 697 634 434 49 316 413 487 545 559 608 625 632 681 683 630 758 671 648

121 155 206 119 75 120 132 1401 417 148 130 132 51 72 157 82 122 126 165 137 141 121 138 142 159 156 154

7.0 9.3 14.0 6.9 5.8 8.1 8.0 61.0 21.0 9.1 7.5 6.3 2.4 5.3 7.7 4.9 7.2 7.1 10.0 8.7 7.8 7.3 8.7 8.4 11.0 10.0 10.0

7.0 8.2 17.0 6.8 5.9 6.7 8.3 51.0 20.0 8.1 6.5 4.3 4.6 5.8 8.3 4.9 8.0 7.1 9.2 8.2 6.6 6.1 8.9 8.2 9.6 10.0 9.5

19 21 29 21 20 22 22 30 27 22 20 16 20 17 24 20 21 21 21 21 22 19 20 22 19 23 20

5.3 4.3 42.0 5.0 5.4 6.3 4.4 8.7 3.4 4.6 3.4 47.0 3.7 4.2 4.4 3.6 5.0 4.2 4.7 3.7 6.0 5.6 6.2 4.2 3.2 6.9 6.2

60 64 145 57 34 59 57 383 173 59 62 102 14 23 74 38 53 49 63 63 62 54 59 58 65 63 57

18 16 19 20 22 18 15 10 15 15 18 34 32 25 21 19 20 19 19 16 17 18 18 17 17 16 18

32 34 54 30 19 33 32 217 92 54 29 31 19 23 36 22 31 32 35 40 33 28 35 36 41 33 37

46 66 106 46 32 51 58 436 153 70 54 44 15 28 47 35 50 54 67 57 59 40 50 60 67 68 64

17 15 31 17 23 14 12 105 59 41 17 14 26 16 27 16 19 18 17 21 18 14 18 18 17 27 15

18 23 38 16 8 14 20 184 54 22 21 12 0 9 16 13 17 20 27 19 24 15 20 21 29 24 28

S. D. BURGESS & J. S. MILLER

BTL 9 BTL 10 BTL 18 BTL 25 BTL 27 BTL 28 BTL 30 BTL 31A BTL 31B BTL 36 BTL 37 BTL 39 BTL 41 BTL 42A BTL 42B BTL 43 BTL 45 BTL 48 BTL 55 BTL 56 BTL 58 BTL 59 BTL 68 BTL 69 BTL 79 BTL 81 BTL 82

Ni Cr Sc V Ba Rb Sr Zr Y Nb Ga Cu Zn Pb La Ce Th Nd (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm)

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Sector 54 thermal ionization mass spectrometer at the University of North Carolina (UNC). Lead was analysed using a single Daly collector, and analyses are corrected by 0.18%/amu for mass fractionation. After subtraction of blank Pb (,5 pg), common Pb corrections were unnecessary for most fractions. For fractions with total common Pb in excess of 5 pg, corrections were made using Stacey & Kramers (1975) initial Pb. Uranium was analysed using a dynamic peak-hopping routine. Decay constants used were 238U ¼ 0.155125  1029 year21, and 235U ¼ 0.98485  1029 year21 (Steiger & Ja¨ger 1977).

Field relations and petrography Schlieren Schlieren are common but heterogeneously distributed throughout the field area. Schlieren are typically layered and defined by alternating decimetre-scale sets or sequences of mafic minerals (biotite and hornblende) and felsic minerals (feldspars and quartz). They display sedimentary-like features such as size grading (most commonly reversely graded), ‘cross bedding’, and layer truncations. Subhedral to euhedral hornblende and biotite are commonly imbricated. In the Cathedral Peak unit, schlieren layers dominantly strike NW–SE. The average strike of schlieren within the entire study area is 1578 but can vary up to 508 locally (Fig. 3). The average strike of the contact between Cathedral Peak and Half Dome granodiorites is 1708, and so in general the schlieren are slightly oblique to this major mapped contact. Schlieren layering dips 608 on average, but varies within the area from vertical to horizontal. Schlieren are commonly associated with K-feldspar concentrations and generally form irregularly shaped boundaries around these features. K-feldspar crystals are commonly aligned parallel to schlieren layering, but discordance is also observed. Schlieren zones occur both near to and well away from the Cathedral Peak– Half Dome unit contact. Generally, regardless of featureassociation, some schlieren are wispy, folded and highly irregular in shape, and are only continuous over a few metres (Fig. 4b). Others are more planar, and laterally continuous for many tens of metres (Fig. 4a). Within the map area, a semi-coherent zone of schlieren was documented which strikes ENE –SSW and is continuous along strike for greater than 1km (Fig. 2). The across-strike extent of the schlieren-rich zone is approximately 2 km, and schlieren in individual outcrops range in thickness from 30 to ,1 m. The zone is composed of both planar (Fig. 4a) and highly irregular schlieren.

209

Although not mapped in detail, several smaller isolated zones, together with the largest continuous outcrop, define a broad SW-dipping swath (c. 558).

Microgranitoid enclaves Enclaves are fairly common within the studied area, both away from and near internal contacts (such as schlieren and interaction zones) but represent much less than 1% of total outcrop. Enclaves mimic the mineral assemblage of the host granodiorite, but contain a greater fraction of mafic minerals, and commonly have hornblende and plagioclase phenocrysts, ranging from 5 to 8mm long. Host rock –enclave interaction is limited, and felsic rims of varying thickness (3 cm) locally surround enclaves. Unlike some intermediate plutons (e.g. Wiebe et al. 2007), enclaves in the Cathedral Peak are not clearly associated with schlieren and although often in proximity to one another, these two features seem physically unlinked. Sparse solitary enclaves are observed, as are small swarms containing multiple enclaves (5– 10) in close proximity (within 1– 2 m). Enclaves in the small swarms typically vary in aspect ratio, and do not show any consistent elongation direction.

Aplitic dykes Aplite dykes are present within the study area and range from 1 cm to 3 m wide. Dykes crosscut nearly all features within the Cathedral Peak unit, and the majority have sharp contacts. However, some dykes have irregular margins and, in a few cases, small dykes of aplite emanate from a larger dyke and dissipate into host granodiorite. Aplites are dominantly fine-grained and homogeneous, although some dykes, generally the larger ones, have pegmatitic cores composed of quartz, Kfeldspar and plagioclase.

Cathedral Peak petrography The Cathedral Peak unit is a porphyritic granodiorite containing abundant large (1– 20 cm) K-feldspar megacrysts and has been described in detail by Bateman (1992). Average granodiorites observed in the study area do not differ from his general description. In thin section, plagioclase is volumetrically dominant (40– 50 modal per cent) and exhibits both oscillatory and concentric zoning. Grains are subhedral and range from fresh to highly altered (sericite) and fractured grains. Commonly, anhedral resorption surfaces can be seen truncating albite twins. K-feldspars (16–25 modal per cent) are dominantly orthoclase with less abundant microcline and can be divided into two separate categories. The first consists of megacrysts that

210

S. D. BURGESS & J. S. MILLER

Table 3. Trace element concentration determined by inductively coupled plasma mass spectrometry. All values are in ppm

BTL 9 BTL 10 BTL 18 BTL 25 BTL 27 BTL 28 BTL 30 BTL 31A BTL 31B BTL 36 BTL 37 BTL 39 BTL 41 BTL 42A BTL 42B BTL 43 BTL 45 BTL 48 BTL 55 BTL 56 BTL 58 BTL 59 BTL 68 BTL 69 BTL 79 BTL 81 BTL 82

La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

29.1 34.4 51.4 29.7 18.6 35.1 32.1 221.0 95.7 53.3 30.1 32.8 18.1 20.4 38.2 22.4 31.5 34.0 38.6 37.2 34.4 27.0 30.0 36.1 48.2 29.0 30.8

46.1 58.0 87.4 47.0 26.8 55.3 53.6 421.7 154.5 75.8 50.6 45.3 16.7 29.9 54.2 33.6 47.4 51.6 64.3 58.8 55.1 43.5 49.2 56.2 79.3 47.3 51.1

4.62 6.10 8.83 4.71 2.51 5.59 5.54 46.30 15.70 7.29 5.28 4.0 0.96 2.89 5.02 3.26 4.69 5.06 6.68 6.01 5.66 4.51 5.07 5.65 8.41 4.90 5.24

16.4 22.0 32.1 16.9 8.8 19.9 20.2 172.2 56.2 25.4 19.2 12.9 2.1 10.1 17.3 11.6 16.5 18.1 24.3 21.6 20.4 16.4 18.4 20.5 30.8 17.9 19.2

2.8 3.8 5.5 2.9 1.5 3.4 3.5 30.0 9.6 4.2 3.3 2.0 0.2 1.7 2.8 1.9 2.7 3.0 4.2 3.8 3.6 3.0 3.3 3.5 5.4 3.1 3.4

0.7 0.9 1.1 0.7 0.4 0.8 0.9 6.0 2.0 1.0 0.8 0.2 0.0 0.4 0.7 0.5 0.7 0.7 1.0 0.9 0.9 0.8 0.8 0.9 1.3 0.8 0.8

1.8 2.5 3.9 1.9 1.0 2.2 2.3 20.0 6.4 2.8 2.2 1.3 0.1 1.1 1.9 1.3 1.8 2.0 2.9 2.6 2.4 2.0 2.2 2.4 3.9 2.1 2.3

0.23 0.32 0.5 0.25 0.12 0.29 0.29 2.59 0.82 0.35 0.28 0.17 0.02 0.15 0.24 0.16 0.23 0.26 0.38 0.32 0.31 0.27 0.28 0.30 0.48 0.27 0.30

1.2 1.6 2.6 1.3 0.6 1.5 1.5 13.3 4.2 1.8 1.5 0.9 0.1 0.8 1.2 0.8 1.2 1.3 1.9 1.7 1.7 1.4 1.5 1.5 2.5 1.4 1.5

0.21 0.29 0.47 0.22 0.11 0.27 0.27 2.38 0.76 0.32 0.26 0.17 0.03 0.14 0.23 0.15 0.21 0.24 0.34 0.31 0.30 0.25 0.27 0.28 0.45 0.25 0.27

0.56 0.78 1.20 0.60 0.30 0.70 0.66 6.0 2.0 0.82 0.65 0.44 0.11 0.36 0.61 0.41 0.56 0.62 0.89 0.80 0.75 0.63 0.68 0.71 1.10 0.63 0.72

0.08 0.11 0.17 0.09 0.05 0.10 0.10 0.88 0.29 0.12 0.10 0.07 0.02 0.06 0.09 0.06 0.08 0.09 0.13 0.12 0.11 0.09 0.10 0.10 0.16 0.09 0.10

0.55 0.72 1.13 0.57 0.34 0.66 0.64 5.60 1.90 0.77 0.61 0.44 0.24 0.38 0.65 0.40 0.55 0.59 0.84 0.77 0.72 0.58 0.63 0.65 0.99 0.61 0.64

are typically .3 cm in diameter and enclose all other minerals in the assemblage. Plagioclase is the most abundant inclusion. Fine-grained inclusions of minerals such as biotite and plagioclase are commonly concentrated along growth surfaces within megacrysts, resulting in roughly concentric zones of inclusions from rim to core. These megacrysts have Carlsbad twins and are commonly exsolved to perthite. The second population is dominantly interstitial and chiefly composed of orthoclase but includes minor microcline. Mafic minerals make up 10% of the Cathedral Peak mineral assemblage. Hornblende is sparse (,3 modal per cent) in the granodiorite and is generally small (1 –2 mm) and sub/euhedral. Biotite (3–4 mm) is the dominant mafic phase (3–5 modal per cent typically) and is found both as fresh grains and chloritized fragments. Titanite is the most common accessory phase (,1 modal per cent) and is dominantly euhedral. Sub-millimetre oxides are also found in small concentrations adjacent to and sometimes in contact with titanite crystals, and euhedral titanite (,1 mm) and oxides are commonly included within biotite. Accessory zircon and apatite are present, and, a few samples from the inner part of the Cathedral Peak along

the transect (,1 km from the Johnson Granite contact) also contain allanite. The compositional outliers in the area include microgranitoid enclaves, mafic schlieren and aplite dykes. Enclave compositions are dominantly plagioclase (50 –60 modal per cent), amphibole (25– 30 modal per cent) and biotite (10–15 modal per cent). Plagioclase is sericitized and has commonly reacted to form myrmekite. Amphibole is the dominant mafic phase in enclaves, and biotite is slightly less abundant. Hornblende is acicular and bladed, presumably due to rapid undercooling of the enclave. Biotite is commonly small (,0.5 mm) and fresh, but some grains are altered to chlorite. Accessory titanite is common and can be larger (up to 5 mm) than biotite and hornblende. Oxides compose a larger modal percentage of enclaves than granodioritic samples (up to 2%). Overall, enclaves are much finer-grained than the majority of Cathedral Peak rocks. Mafic schlieren consist predominantly of mafic silicates, oxides, and plagioclase. Quartz and Kfeldspar are present in the mafic layers but typically are small (,3 mm), interstitial and comprise ,10 modal per cent. The colour index ranges typically from 20 to as high as 60. Hornblende and biotite

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

211

Table 3. (Continued) Lu 0.09 0.12 0.19 0.09 0.06 0.11 0.11 0.93 0.31 0.13 0.10 0.09 0.06 0.07 0.12 0.07 0.10 0.10 0.14 0.13 0.12 0.10 0.11 0.11 0.16 0.10 0.11

Ba

Th

Nb

Y

Hf

Ta

U

Pb

Rb

Cs

Sr

Sc

Zr

854.4 562.5 252.6 871.5 276.6 566.2 608.7 226.6 442.4 871.7 757.9 1118.0 13.1 719.8 159.7 601.5 507.2 727.3 565.1 385.8 603.8 1164.0 991.1 606.9 1096.0 798.5 936.0

14 16 29 15 19 18 14 109 65 38 16 14 25 14 26 16 23 18 17 27 17 13 15 17 20 16 14

7.1 9.1 16.0 7.3 4.9 8.5 8.2 62.0 22.0 9.5 8.2 4.4 4.7 6.4 10.0 5.6 7.5 8.8 11.0 9.3 9.0 6.8 7.8 9.0 11.0 7.5 7.8

6.48 8.76 13.70 6.59 3.60 8.09 8.0 65.8 21.4 9.4 7.4 4.31 1.32 4.03 6.88 4.44 6.22 6.99 10.10 9.10 8.49 7.19 7.51 8.06 12.70 7.25 7.96

3.5 4.3 6.2 3.4 2.3 4.0 3.7 42.0 12.0 4.5 3.4 3.8 2.8 2.9 4.9 2.5 3.7 3.8 4.8 4.4 4.0 3.3 3.7 4.0 4.4 3.1 3.1

0.6 0.7 1.0 1.1 0.4 0.7 0.6 6.0 2.0 0.8 1.9 0.4 2.0 2.0 2.0 2.0 0.6 2.0 2.0 0.7 2.0 1.0 0.6 2.0 0.9 0.6 0.6

4 5 8 4 5 4 3 21 12 6 5 5 11 13 9 5 6 5 5 8 4 3 5 5 7 5 4

19 16 19 18 20 17 16 13 17 16 17 35 30 22 21 19 21 19 17 16 16 17 18 17 16 17 17

147.3 119.2 178.4 134.9 153.4 127.8 107.4 148.8 190.4 110.6 131.1 380.7 219.7 193.5 176.2 128.5 156.0 127.3 127.1 125.0 121.2 119.6 128.1 125.2 114.5 130.8 126.2

5.1 4.6 13.7 4.3 4.9 5.3 3.4 7.6 7.3 4.0 8.6 20.0 4.2 5.0 6.0 3.2 5.1 3.4 5.2 9.0 4.3 5.3 4.6 5.1 5.9 6.1 5.3

597.3 630.3 559.2 598.7 400.0 583.8 643.6 252.4 512.8 663.6 614.9 427.6 48.61 311.3 413.6 481.8 509.2 554.1 618.1 606.0 639.5 687.5 666.9 622.2 732.0 662.0 635.6

2.9 3.8 12.5 3.0 1.8 3.5 3.4 17.8 10.5 3.9 3.5 7.5 1.1 2.2 5.0 1.8 2.9 3.3 4.4 3.7 3.8 3.8 3.6 4.5 5.1 3.4 3.5

117 146 204 109 60.7 131 128 1520 432 153 112 125 49 68 148 78 112 121 157 145 132 115 121 133 142 100 100

are generally 2 –4 mm in size and either subequal in abundance or have biotite greater than hornblende. Both minerals are subhedral, and commonly have inclusions of titanite and oxides; they can also be intergrown, or occur in mineral clots with oxides and titantite. Plagioclase shows oscillatory zoning and commonly has inclusions of hornblende, biotite, titanite, oxides, zircon and apatite. In the mafic-rich layers biotite and hornblende are generally 2–3 mm whereas in the felsic layers they are more commonly 4–6 mm in size. Perhaps the most notable feature of the mafic schlieren is their distinct enrichment in heavy minerals (titanite, oxides, zircon and apatite). Particularly in dark schlieren, it is not uncommon for titanite and oxides to comprise 30 modal per cent of the rock and rare millimetre-scale layers of nearly pure oxides and titanite can also occur (see also Reid et al. 1993). Trace amounts of sub-euhedral, zoned allanite were found in BTL031a.

Ladder dykes Ladder dykes are common within the Cathedral Peak unit and are typically but not always associated with schlieren-rich zones (Reid et al. 1993; Weinberg et al. 2001; Zak & Paterson 2005).

Ladder dykes are delineated by a crescent-shaped arrangement of dark and light layers arranged in concave-up stacks (Reid et al. 1993). Lighter layers are composed chiefly of plagioclase with subordinate quartz, K-feldspar, hornblende and biotite. Dark layers contain dominantly hornblende, biotite, plagioclase and abundant titanite. Ladder dykes within the study area are typically 2–3 m long (Fig. 4d) and dip steeply into the exposed surface, similar to schlieren in the unit. In some cases, although the bulk of the ladder dyke is concentrated in a small area and is sharply bounded, the periphery of the dyke is observed to dissipate gradually into host rock. Most of the ladder dykes are well within the Cathedral Peak Granodiorite and not directly adjacent to any major unit contact.

Interaction zones Along the northern transect, two zones comprising at least several tens of square metres display evidence of intimate intrusive interaction between felsic materials of contrasting mineralogy and texture. In each case, a more felsic, K-feldspar megacryst-poor granite has interacted with a less felsic, megacryst-rich granodiorite (Fig. 4f–h). Both zones are irregularly shaped and dominantly

212

Table 4. Isotope data Sample

Nd (ppm)

Sm–Nd weight ratio

2.96 3.15 8.76 3.77 1.76 2.41 1.92 0.69 3.45 4.41 3.52 3.33 4.51 3.55

19.84 20.40 57.68 24.76 11.76 16.59 12.89 19.87 22.13 27.72 22.27 21.52 27.76 22.71

0.0900 0.0932 0.0918 0.0920 0.0908 0.0879 0.0898 0.0210 0.0942 0.0961 0.0956 0.0936 0.0981 0.0945

147

Rb, Sr concentrations determined by ICP-MS and given in Table 3. Errors on 1Nd are +0.3 1 units for all samples. Errors on initial Sr are 2  1025 for all samples.

Sm/144Nd ratio 0.1490 0.1542 0.1519 0.1523 0.1502 0.1454 0.1486 0.0347 0.1558 0.1590 0.1582 0.1548 0.1623 0.1564

Measured Nd/144Nd

143

0.5122 0.5122 0.5122 0.5123 0.5123 0.5122 0.5122 0.5118 0.5123 0.5123 0.5123 0.5123 0.5122 0.5123

2s error on measured 143 Nd/144Nd 0.000090 0.000008 0.000008 0.000009 0.000007 0.000008 0.000008 0.000007 0.000008 0.000009 0.000008 0.000006 0.000010 0.000006

Initial Nd/144Nd T ¼ 88 Ma

1 Nd at time T

87 Sr/86Sr measured

Calculated initial 87 Sr/86Sr

0.512190 0.512195 0.512189 0.512204 0.512206 0.512189 0.512198 0.511804 0.512214 0.512207 0.512246 0.512217 0.512174 0.512080

26.5 26.4 26.6 26.3 26.2 26.5 26.4 26.2 26.1 26.2 25.4 26.0 26.9 26.2

0.70728 0.70705 0.70779 0.70726 0.70855 0.70791 0.70704 0.70731 0.70723 0.70718 0.70699 0.70720 0.70712 0.70722

0.70640 0.70645 0.70636 0.70645 0.70614 0.70628 0.70637 0.70644 0.70645 0.70646 0.70632 0.70643 0.70655 0.70648

143

S. D. BURGESS & J. S. MILLER

BTL028 BTL030 BTL031B BTL037 BTL042A BTL042B BTL043 BTL048 BTL055 BTL058 BTL059 BTL069 BTL079 BTL081

Sm (ppm)

Fractions Weight* (mg)

Concentrations

206

Pb‡ Pb

204 †

U (ppm)

Pb (ppm)

208

Pb§ Pb

206

206

Pb§ U

238

% Error

2s errors % (%) Error 207 Pb 235 U

207 206

Pb Pb

% Error

Age (Ma) 206

Pb U

238

Common

207

207

235

206

Pb U

Pb Pb

corr. coef.

Pb (pg)

BTL045 _8z _3z _4z _1z _5z _6z

0.0050 0.0050 0.0050 0.0050 0.0030 0.0030

1067.6 901.0 1074.6 1040.7 2108.6 1357.7

18.5 12.8 15.5 14.1 29.9 21.1

810.16 808.94 727.69 1432.12 861.51 343.15

0.107 0.122 0.124 0.112 0.122 0.131

0.01654 0.01370 0.01365 0.01357 0.01356 0.01344

0.51 0.72 0.60 0.63 0.51 0.79

0.12684 0.09046 0.08984 0.08941 0.08900 0.08884

0.60 0.89 0.67 0.83 0.56 1.06

0.05560 0.04790 0.04772 0.04777 0.04759 0.04794

0.30 0.50 0.29 0.51 0.20 0.67

105.8 87.7 87.4 86.9 86.8 86.1

121.3 87.9 87.4 87.0 86.6 86.4

436.5 94.2 85.5 88.1 79.2 96.3

0.864 0.827 0.902 0.788 0.931 0.772

7.1 5.0 6.6 3.2 6.5 10.7

BTL056 _6z _7z _3z _1z _8z

0.0050 0.0050 0.0050 0.0050 0.0060

485.9 1224.3 1208.6 1041.1 1016.6

7.4 17.8 17.0 14.1 16.1

334.60 891.26 1163.60 1432.12 1112.50

0.119 0.143 0.135 0.112 0.086

0.01367 0.01371 0.01360 0.01357 0.01577

1.30 0.53 0.54 0.63 0.47

0.09073 0.09048 0.08976 0.08938 0.10518

1.54 0.69 0.61 0.83 0.51

0.04814 0.04785 0.04788 0.04777 0.04836

0.80 0.41 0.28 0.51 0.21

87.5 87.8 87.1 86.9 100.9

88.2 88.0 87.3 86.9 101.5

106.0 91.8 93.3 88.1 117.2

0.855 0.799 0.893 0.788 0.913

6.7 6.1 4.6 3.2 5.6

*Estimated, †radiogenic Pb. ‡Measured ratio corrected for fractionation only. §Corrected for fractionation, spike, blank, and initial common Pb.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Table 5. U –Pb zircon analytical data for BTL045 and BTL056

213

214

S. D. BURGESS & J. S. MILLER

Cathedral Peak Granodiorite contact), consists of three rock types, differentiated in the field by colour index, mineralogy and texture. One rock type closely resembles the most common Cathedral Peak Granodiorite in terms of grain size and mineralogy, but generally has fewer K-feldspar megacrysts. The second rock type is leucocratic (colour index, CI  5), and contains subhedral quartz and plagioclase phenocrysts, similar to the Johnson Granite Porphyry, but different in that it contains Kfeldspar megacrysts (Fig. 4h). Boundaries between this leucocratic material and the Cathedral Peak host range from sharp to diffuse. The third rock type consists of granite which is intimately intermingled with the other two rock types and is characterized by abundant faint and discontinuous layered schlieren that are commonly folded into chaotic patterns. All of these granitic ‘facies’ display very complex geometric patterns and mutually crosscutting intrusive relationships with each other, and the overall ‘soupy’ appearance of the contacts suggests that the different rock types were rheologically capable of mingling and partially hybridizing.

Potassium feldspar concentrations

Fig. 3. (a) Contoured stereonet plot of poles to foliation within the Cathedral Peak Granodiorite (n ¼ 66). (b) Contoured stereonet plot of poles to schlieren layering in the Cathedral Peak Granodiorite (n ¼ 50).

non-planar, rather showing a blob-like geometry. The textures in each interaction zone show mutually cross-cutting relationships of the magmatic compositions. Interaction boundaries range from sharp to diffuse, commonly displaying cuspate/lobate geometries and ‘flame’ structures. In a few cases, centimetre-scale veinlets of leucocratic melt from a larger mass of granite have permeated the host granodiorite, where they appear to dissipate in less than a metre from the primary contact between the granite and host granodiorite (Fig. 4f). The larger of the two zones, cropping out in the eastern edge of the area (near the Johnson Granite–

K-feldspar megacrysts ranging up to 10 cm long are common in the Cathedral Peak unit. In addition to being scattered throughout the unit as single crystals, K-feldspar megacrysts occur in dense concentrations of nearly 100% megacrysts (Fig. 4e). Individual megacrysts within a concentration are similar to isolated crystals in size, shape and internal structure, showing inclusions of biotite and hornblende along concentric growth rings. Within the study area, K-feldspar concentrations exhibit a wide variety of geometries, which can be roughly classified as tubes, troughs, irregularly shaped clusters, dykes or small diapirs (cf. Zak & Paterson 2005). In general, small elliptical to circular masses (or pipes in three dimensions) can be ,10 cm in diameter, whereas some dyke-like concentrations can be .1 m in width and continuous for several metres or more along strike. Imbrication of K-feldspar megacrysts is common in the larger concentrations, which are found intruding the porphyritic Half Dome unit (on the western side of the area) as irregularly shaped schlieren-bounded masses. Marginal schlieren are commonly associated with the larger concentrations, and may be the result of a process similar to that invoked by Wiebe et al. (2007) to explain schlieren associated with enclaves. A noteworthy feature of the K-feldspar concentrations is that the immediately adjacent host granodiorite typically does not show obvious depletion in K-feldspar relative to areas lacking or far away from the K-feldspar concentrations.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

215

Fig. 4. (a– d) Differing mafic schlieren geometries within the Cathedral Peak Granodiorite. (a) Steeply and dipping planar schlieren outcrop from within the semi-continuous schlieren zone (grey zone in Fig. 2). (b) Cuspate ladder dyke exposure. (c) Magmatic fault offsetting steeply dipping schlieren. Dotted lines demarcate schlieren-layer boundaries offset by the fault. (d) Ladder dyke truncated by aplite dyke. (e) Monomineralic K-feldspar concentration. Dotted lines delineate sharp concentration boundary with host rock. (f – h) Examples of interactions between rocks of variable composition. All three photos display the interaction between a relatively mafic, schlieren-like rock, a megacryst-poor Cathedral Peak-like rock and a felsic, quartz-rich rock. (f) A broad zone of interaction between these three compositions. (g) A closer view of a delicate contact within this same area showing the highly irregular contact geometries. (h) A close-up view of the delicate contacts between these three compositions. This photo (h) highlights the schlieren-like bearing rock and the quartz-rich megacryst-poor rock.

Magmatic faults Magmatic faults (i.e. those whose movement took place prior to complete pluton solidification) occur in several places within the Cathedral Peak unit. These faults commonly offset material of the same composition, making them difficult to observe, and so they may be more abundant than is apparent in the field (Fig. 4c). Faults sometimes offset other structures such as ladder dykes. At two locations, the exposed fault plane shows sinistral-oblique normal motion, whereby the extensional component is smaller than the strike –slip component in both cases. The dilated area is

healed by either K-feldspar-rich material, similar to that found in ‘pipes’ and concentrations elsewhere in the Cathedral Peak unit, or by aplite.

Magmatic fabrics Two magmatic foliations were recognized within the Cathedral Peak unit along the sample transects and in near vicinity (Fig. 3). Both are defined by aligned biotite primarily and by hornblende. One foliation strikes NNE –SSW, roughly parallel to major unit contacts at the latitude of the transects, and the other strikes WNW –ESE. The WNW– ESE striking fabric cuts across the more contact-parallel

216

S. D. BURGESS & J. S. MILLER

fabric in many places. The strike of dominant foliation varies between stations, but dips are uniformly steep (average 778). Mineral lineation was not generally observed and therefore not systematically recorded, but in the few cases where it could be measured it was generally steeply plunging (70–908).

Results Geochemistry Major element variation Including compositional outliers such as mafic enclaves, mafic schlieren and aplite dykes, the total range of silica concentration for samples along the northern and southern transects ranges from 48 wt% (mafic schlieren) to 78 wt% (aplite), but the volumetrically dominant granodiorites along the two transects display a much narrower compositional range, from 67 to

Fig. 5. Major elements v. silica.

72 wt% SiO2 (Table 1). The greatest major element scatter is observed at low (,65%) SiO2. These samples constitute a very small fraction of the rocks and greatly alter trends on the Harker plots (Fig. 5). However, their compositions are important in evaluating the possible roles of mixing and fractionation within the suite and are thus relevant to the later discussion. Alkalis (Na2O, K2O) generally show appreciable scatter v. SiO2, while the other major element oxides are fairly well-correlated and show linear trends with some scatter in the low to intermediate SiO2 range. From the outer margin of the Cathedral Peak unit inward, major elements generally show sub-linear trends (Bateman 1992). Within the study area, silica increases inward while calcium decreases over the same interval, similar to the pattern observed by Bateman & Chappel (1979) (Fig. 5). Patterns for FeO, MgO, TiO2 and Al2O3 are similar to CaO. Alkalis also increase inward, but show appreciable scatter.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Trace element variation Excluding mafic schlieren and enclaves, feldspar compatible trace elements (e.g. Rb, Sr, Ba) show differing trends with respect to SiO2 (Fig. 6, Tables 2 & 3). Strontium concentration is generally well correlated with silica, decreasing linearly with increasing SiO2. Rubidium generally increases with increasing SiO2, but shows a greater degree of scatter than Sr. Barium generally decreases with increasing SiO2, but again shows appreciable scatter throughout the compositional range. Mafic schlieren and enclaves clearly lie off the main Cathedral Peak trend. Compatible elements (e.g. Sc, Y) decrease linearly with increasing SiO2 (Fig. 7), showing scatter similar to Rb. Zirconium shows appreciable scatter with SiO2, but the most felsic rocks (75 wt% SiO2) generally have lower Zr than most of the lower silica granodiorites. Plotting U against Th (two elements incompatible in nearly all phases in the assemblage) shows an overall positive correlation, but the data generally fan out from the origin (Fig. 8). Plotting incompatible v. compatible (e.g. Th, Sc) elements shows no clear trend (Fig. 8). Large-ion lithophile elements show spatial differences from the western side of the study area towards the centre of the TB (Fig. 9). Rubidium and Sr trends are scattered, but generally show increases and decreases respectively moving toward the centre with the Sr trend being much more definitive. Barium values (not shown) are scattered, but in general the range of Ba values increases inward in the unit if very high-silica rocks and schlieren are excluded. Strongly incompatible elements (e.g. U, Th) show scatter similar to Rb (Fig. 9). Compatible elements (e.g. Sc) are scattered about a main trend, but generally decrease slightly inward. Rare earth element variation Chondrite-normalized rare earth element patterns display a remarkable degree of similarity in overall shape from 67 to 73 wt% silica (Fig. 10). All samples in this compositional range show light rare earth element (LREE) enrichment (100–300  chondrite) and lack Eu anomalies (Table 3). Leucocratic samples from the Cathedral Peak unit (.73 wt% silica) also lack Eu anomalies but one aplite dyke is strongly depleted in middle REEs relative to LREEs and HREEs, which is denoted by the concave-up REE pattern (Fig. 10) (e.g. Nabelek 1986). Overall REE concentrations decrease with increasing SiO2 values (e.g. compare Fig. 10c & d). An enclave from the Cathedral Peak unit shows a slight negative Eu anomaly and, like the granites and granodiorites, exhibits LREE enrichment. The two schlieren samples are characterized by overall REE enrichment reaching 800 –1000  chondrite for LREEs in the most mafic schlieren. Schlieren also display very slight negative Eu

Fig. 6. Feldspar compatible elements v. silica.

217

218

S. D. BURGESS & J. S. MILLER

Fig. 8. (a) Incompatible v. incompatible trace elements and (b) incompatible v. compatible trace elements.

anomalies. The amphibolite xenolith (BTL039) has the largest negative Eu anomaly. Excluding the strongly fractionated aplites and schlieren, increasing Ce/Yb in all other samples indicate a general LREE enrichment toward the centre of the Cathedral Peak (Fig. 11; see also Frey et al. 1978).

Sr and Nd isotopes Fig. 7. Trace elements v. silica.

Fourteen samples were analysed for Sr and Nd isotopic composition (Table 4). Initial strontium

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Fig. 9. Trace elements v. longitude.

219

220

S. D. BURGESS & J. S. MILLER

Fig. 10. Chondrite normalized rare earth element plots. (a) Low-silica samples (,68 wt% silica). (b) Samples with between 68.01 and 70 wt% silica. (c) Samples with 70.01–73 wt% silica. (d) Samples with .73 wt% silica. Samples from all four plots come only from the Cathedral Peak unit and plots (b) and (c) contain only granodioritic Cathedral Peak samples.

(87Sr/86Sr(i)) ranges from 0.70640 to 0.70655 for these samples. At the two extremes are a leucocratic, megacryst-poor Cathedral Peak sample from the interaction zone (BTL042A) and a relatively low-silica Cathedral Peak sample from the lower sampling transect (BTL079) (Fig. 12). In general, higher silica samples, which correspond to the inner portions of the field area, show slightly lower 87Sr/86Sr(i) values. However, for the highestsilica samples with high Rb/Sr, the initial ratio is very sensitive to the age correction, and initial ratios are only precise to +5% (Rb and Sr concentrations determined by ICP-MS). We therefore hesitate to attach too much significance to the modest variation in 87Sr/86Sr(i) without precise age information for each sample, especially for higher silica samples. In general though, the range of 87 Sr/86Sr(i) from the study area is small in comparison to the range reported elsewhere for the entire TB (e.g. Kistler et al. 1986; Coleman & Glazner

1997; Gray 2003) (Fig. 12). The schlieren sample from the Cathedral Peak unit has 87Sr/86Sr(i) just slightly lower than other granodiorite and granite samples. Initial epsilon Nd (1Nd(t)) values range from 24 to 26.9 but most are less than 25. In the Cathedral Peak Granodiorite, 1Nd(t) is not correlated (within error) with increasing silica (Fig. 12). This is also reflected in constant 1Nd(t) with respect to longitude. As with 87Sr/86Sr(i), there is very limited variation for 1Nd(t) along the transects compared with the TB as a whole (e.g. Kistler et al. 1986; Coleman & Glazner 1997), which ranges from 87Sr/86Sr(i) ¼ 0.7057; 1Nd(t) ¼ 23.2 to 87Sr/86Sr(i) ¼ 0.7067; 1Nd(t) ¼ 28.0 (Kistler et al. 1986). This is in general agreement with Bateman (1992), who also observed that the spread in initial isotopic ratios occurs mainly in the outer units of the Tuolumne Batholith (e.g. Kuna Crest and equigranular Half Dome Granodiorites).

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

221

Fig. 11. Rare earth elements v. longitude. Large plots include compositional outliers. Insets omit schlieren samples in order to show the trend defined by granodioritic samples.

Feldspar chemistry A total of 31 plagioclase grains from eight samples at various locations along both sampling transects

span the compositional range of the Cathedral Peak unit, from schlieren to leucocratic granodiorite. Anorthite content among all samples ranges from An48 in mafic schlieren to An1 in leucogranite.

222

S. D. BURGESS & J. S. MILLER

Fig. 12. (a) Epsilon Nd v. initial 87Sr/86Sr covariation showing the relative variance in Cathedral Peak samples (open circles enclosed in grey) from this study as compared with the TB (crosses). Tuolumne Batholith data from Kistler et al. (1986), Gray (2003) and this study. (b) Initial 87Sr/86Sr v. silica and (c) and 1 Nd v. silica. Error bars on (b) & (c) are +0.3 epsilon units for Nd and +2.0  1025 for initial Sr ratios.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Excluding schlieren, the highest An content (An39) was from the outer Cathedral Peak (see supplementary table 1). The majority of plagioclase grains are normally zoned with respect to anorthite. Where detailed traverses were conducted, the interior of the grains (cores) have a relatively constant An content, but are overgrown by rims showing approximately linear (i.e. monotonic) decrease in An content, out to the very edge of the grain. Only one instance of reverse zoning was documented. The average molar anorthite concentrations for both cores and rims decrease systematically with increasing silica of the host rock. Anorthite contents in both rim and core positions also decreases with decreasing longitude (i.e. toward the interior of the Cathedral Peak, see also Bateman 1992). Core compositions range from about An35 (excluding schlieren plagioclase) to An25. Rim compositions vary from An20 to An2 but most are between An15 and An20. The compositions of plagioclase analysed from a mafic schlieren overlap with those from granodioritic Cathedral Peak samples, but one core composition from the schlieren is appreciably more calcic (An47) than any from the granodiorites and is a clear outlier. Schlieren plagioclase grains have more variable core compositions that those from inner and outer Cathedral Peak samples. We also analysed plagioclase grains for Sr, which show zonation similar to An content, with more Sr-rich cores as compared to rims. Spikes in Sr concentration are common within a single crystal, but are not consistent from grain to grain within a single thin section in most cases.

U– Pb geochronology Two Cathedral Peak Granodiorite samples were chosen to represent the innermost and outermost positions along the northern transect (Fig. 2). These locations represent the largest potential age gap along the transects, assuming progressive younging inward from the margins of the Tuolumne Batholith. Both inner (BTL045) and outer (BTL056) samples have ages clustered about concordia (Fig. 13; Table 5). A weighted mean 206Pb – 238U age of the four most concordant fractions from the outer sample (BTL056) is 87.3 + 0.7 Ma. This age is indistinguishable (within error) from the inner sample, BTL045, which is 87.0 + 0.7 Ma based on a weighted mean 206Pb – 238U age of the five most concordant fractions. Weighted means have MSWD substantially .1, which means that the scatter is more than can be attributed to analytical errors. Ages spread along concordia by up to or slightly greater than 1 Ma, suggesting that zircons are a heterogeneous population. Given that all zircons were extensively abraded, the dispersion is

223

most likely attributable to minor inheritance of slightly older cores, but minor lead loss cannot be completely ruled out. One grain in sample BTL045 shows clear inheritance. A discordia line through the cluster of concordant or near concordant grains and this discordant grain gives an upper intercept age of 1.4 Ga. If the dispersion of ages along concordia for these two samples is attributed to inheritance, then the minimum age from the array is likely to be the age that most closely approximates to the final solidification of the sample, again assuming that lead loss has not affected the grains substantially.

Discussion Scale is fundamental to understanding the petrogenetic development of the batholith. If the individual mapped units were constructed as a series of numerous, discrete magma inputs (e.g. Glazner et al. 2004), then what is the geometry, frequency and size of each input? What effect, if any, might new inputs have on the compositional characteristics and physical evolution of the resident (i.e. ‘chambered’) magma and how can field observations, geochronology, geochemistry and isotopic analysis help delineate different inputs? These questions apply particularly to the issue of whether a persistent magma chamber was present in the midcrust during all or a substantial part of the intrusive history. In general, igneous differentiation processes such as magma hybridization and fractional crystallization, which might occur in a mid-crustal magma chamber, are influenced by the rate of magma input relative to solidification rate. Fractional crystallization requires solidification on a timescale that is longer than that needed for processes that separate melt from crystals (e.g. crystal settling, boundary layer melt migration and filter pressing), which will vary depending on the local geothermal gradient, and the size and overall compositional characteristics of the magma chamber (e.g. Reid 1993). Magma hybridization would be favoured by overall high magmatic flux, where ample heat remains in the system such that the magma remains above the solidus between successive inputs. A long interval of construction with a slow rate of intrusion (i.e. low magma flux) would limit the possibility of extensive interaction, particularly on the scale of the individual intrusions that comprise the batholith. Coleman et al. (2004) and Kistler & Fleck (1994) showed that the entire Tuolumne Batholith was assembled over 8–10 Ma, which probably precludes simple fractional crystallization or a magma mixing scenario in a single unitary magma

224

S. D. BURGESS & J. S. MILLER

Fig. 13. Single zircon U– Pb geochronology. Sample BTL045 was collected from the innermost Cathedral Peak unit (closest to the Johnson Granite Porphyry) covered by the sampling transect. Sample BTL056 was collected from the outer Cathedral Peak unit (near the porphyritic Half Dome boundary). Insets show close-up of most concordant fractions.

chamber, because even for an intrusion the size of the entire batholith, simple cooling by conduction will occur on a much shorter timescale (cf. Glazner et al. 2004).

Geochronological data shows that a significant fraction of the 8–10 Ma duration of intrusion applies to the outer units of the batholith (Sentinel, Kuna Crest and equigranular Half Dome

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Granodiorites), including the oldest age (95.0 + 1 Ma), which is from the Sentinel Granodiorite. Some researchers consider the Sentinel to be genetically related to the rest of the Tuolumne Batholith (Kistler & Fleck 1994; Coleman et al. 2004; Glazner et al. 2004) while others (primarily based on map relations) do not (Bateman & Chappell 1979; Bateman 1992; Zak & Paterson 2005). Among the individual mapped units, the most convincing case for a long duration of construction comes from the equigranular Half Dome unit, which spans approximately 3 Ma (Coleman et al. 2004). The porphyritic Half Dome Granodiorite and the Cathedral Peak Granodiorite, which together make up the largest volume of exposed Tuolumne rocks, span a maximum of 3.3 Ma and a minimum of 0.8 Ma based on U –Pb ages and age errors of this study and Coleman et al. (2004), and Kistler & Fleck (1994). Thus, estimating the duration of construction and possible intrusion rate for these units is more equivocal than for the TB as a whole. No significant difference in U –Pb zircon age can be discerned over 4 km of roughly contactperpendicular map distance between the two samples dated in this study (Fig. 13). The weighted mean 206Pb – 238U ages for the concordant zircons in both samples are the same within error, excluding the highly discordant data point for BTL045. The slightly older zircon grains within these samples are most easily interpreted as recycled zircon antecrysts from previously emplaced magma pulses, in which case the minimum age from the array probably gives the age of final solidification, i.e. 86.5– 87 Ma (see discussion in Miller et al. 2007). It could be argued that, over the area of the sampling transects, the Cathedral Peak Granodiorite is actually a sill-like intrusion, in which case no age difference would be expected between the two samples (i.e. they are from the same structural position in the sill). However, universally steep magmatic fabrics and moderately to steeply oriented schlieren suggest otherwise, and more recently Matzel et al. (2005, 2006) have obtained additional U –Pb zircon data on Cathedral Peak from over 10 km to the north, that are in agreement with the age data presented here. Therefore, it is reasonable to conclude that the two dated samples represent approximately the oldest and youngest possible portions of the Cathedral Peak unit along the transects.

Internal contacts In the model of Bateman & Chappell (1979) and Bateman (1992), the mapped contacts in the Tuolumne Batholith are viewed as the result of upward and outward surging of the more

225

mobile melt-rich core of a large batch of inwardly solidifying magma. These contacts generally represent compositional/mineralogical changes within the rocks, although they range from sharp to gradational over several tens to hundreds of metres. For example, the contact between the porphyritic Half Dome and Cathedral Peak units is 2–15 m wide throughout the study area, and is marked by the appearance of K-feldspar megacrysts (generally .4–5 cm long diameter) in the Cathedral Peak unit, abundant and steeply oriented schlieren, and a decrease in the hornblende to biotite ratio. Within the Cathedral Peak Granodiorite, internal zones of schlieren juxtaposed against granodioritic host material (Fig. 4) represent locations in the magma system where crystals have been segregated in some fashion from melt, and the rocks are compositionally cumulates marked in some cases by extraordinarily high REE contents and much higher Zr contents (from accumulated zircon) than average Cathedral Peak Granodiorite (e.g. Reid et al. 1993; Miller et al. 2007) (Fig. 11). This zone is at a high angle to the sampling transects (Fig. 1), so it is difficult to constrain its overall contiguity. Nevertheless, it is a large and previously unrecognized area within the Cathedral Peak unit, where crystals were segregated from melt repeatedly, given that younger layered sets show truncations of older layers in this zone as they do elsewhere in the Tuolumne Batholith (Reid et al. 1993; Zak & Paterson 2005). The outcrop patterns of the schlieren layers suggest widely varying rheology in the granodiorite during and after their formation. Straight and relatively steep continuous layering may represent areas of fairly rigid (high crystallinity) ‘walls’ in the magma, along which flow and segregation of crystals from melt could occur. However, it is not clear from the layers whether magma was primarily moving up (i.e. because of positive buoyancy) or down, perhaps as a high particle concentration gravity flow (e.g. Barierre 1981). Nevertheless, the strong rheological contrast resulted in a fairly ordered sequence of schlieren, where each set of mineral layers represents a separate small-scale flow-segregation event. In contrast, areas where layering is poorly or subtly developed and layers are folded and mingled with and/or disrupted by granodiorite represent areas in the magma where either the rheologic contrast between schlieren and granodiorite was modest, or wholesale collapse of a semi-rigid mush occurred. This may have occurred in conjunction with partial mixing of magmas during intrusion or a new pulse of magma (e.g. Bergantz 2000). In addition to schlieren, monomineralic Kfeldspar concentrations suggest accumulation of

226

S. D. BURGESS & J. S. MILLER

crystals in contact with host granodiorite. These contacts are commonly sharp and fairly linear (Fig. 4). K-feldspar megacryst concentrations can result from flow of megacryst-bearing melt that is ‘necked’ down, effectively decreasing the diameter of the conduit through which the same volume of material is attempting to flow. They are often associated with pipe-like schlieren (ladder dykes), suggesting a process of flow through some sort of magmatic conduit or constriction, which would create a ‘logjam’ of megacrysts, resulting in outcrop-scale concentrations (Weinberg et al. 2001; Zak & Paterson 2005). In many cases, clusters of megacrysts are detached and segregated from adjacent larger K-feldspar masses. These features could either represent brittle separation and tearing of a small portion of megacrysts from a larger mass into a more melt-rich environment, or alternatively could represent smaller local ‘logjams’ resulting from grain to grain interference and flow within crystal-rich magma, perhaps akin to glomerophyric clots in lava flows. Unlike mafic schlieren contacts and K-feldspar concentrations, which separate accumulated crystals from granodiorite, the felsic magma interaction zones represent areas where granodioritic to granitic liquids with subtle compositional differences have interacted. Contacts range from gradational to abrupt and are generally highly irregular over a short distance. The intricate and irregular contacts between different rock compositions are indicative of a partially molten state. They may be the product of variable degrees of fractionation within a larger mass of granitic mush. The presence of abundant and highly disrupted wispy schlieren (Fig. 4f) may indicate that, even in these zones, fractionation of felsic melt from granitic mush occurred. Alternatively, the different varieties of granite may represent separate inputs of new magma, where boundaries between different compositions represent contacts between resident mush and new magma. Disruption of the faint schlieren by granite is indicative of mingling between felsic and less-felsic magmas and partially (but perhaps only slightly) accumulative magma. The geochemistry of some of these felsic masses suggests that they are highly fractionated liquids from some less silicic parent, whereas other compositions suggest mixing of crystals into highly evolved magma. The very late aplite dykes cut surrounding granodiorite resulting in sharp, planar contacts in most cases. This relationship suggests that the aplite was either a late addition to the pluton, and that the host material was near or below its solidus at the time of intrusion, or it was a late-evolved felsic melt extracted from the granodiorite when it was at near-solidus conditions.

Compositional variation in Cathedral Peak: fractional crystallization, magma mixing or melt source variability? We have argued above that the operation of magma chamber processes on a fairly wide scale is not precluded by existing geochronology on the Cathedral Peak Granodiorite, and that at least some of the magmatic structures and features preserved in the field record crystal accumulation and mixing between texturally and/or compositionally distinct magmas and/or crystal mush. Thus, the geochemical variation that is observed in the Cathedral Peak Granodiorite may potentially involve fractional crystallization (e.g. of schlieren), mixing of magmas, magma–‘mush’ mixing, and/or successive emplacement (with limited interaction or mixing) of granodiorite liquids representing variable partial melt fractions of likely Sierran granitoid protoliths (e.g. Sisson et al. 2005) (Fig. 5). Here we consider the possible role of these processes in producing the observed geochemical variability in light of the field relations and geochronology. Major element mixing models Major element trends from the Cathedral Peak are potentially compatible with fractionation of mafic silicates, plagioclase, and accessory minerals, and the schlieren are the most obvious candidates for possible cumulates. However, as Reid et al. (1993) noted, removal of mafic schlieren from a granodioritic parent magma would drive both Al2O3 and SiO2 up (because mafic layers are generally feldspar poor), whereas Al2O3 decreases with increasing SiO2 overall in TB rocks. Based on major element trends (Fig. 5), possible fractionating mineral assemblages are also likely to include feldspars, particularly plagioclase. Our attempts to model possible fractionation scenarios for Cathedral Peak rocks proved equivocal due to the lack of dacitic or finegrained granodiorite dykes or enclaves that might represent plausible low-silica parents or recharging liquids. The major element trends observed in the unit could result from a number fractionation processes, including fractionation of hornblende – biotite rich schlieren in conjunction with plagioclase from a granodioritic parent and, given the linearity of major elements at .67 wt% SiO2, simple bulk-segregation of granodiorite from granite. Similarly, magma mixing models suffer from not knowing the end-member magmas involved. It has been suggested that mixing of mafic to intermediate magmas that are similar to enclaveforming magmas within granitic magma can produce granodiorite compositions in the Tuolumne Batholith, including the Cathedral Peak unit

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

(cf. Reid et al. 1983). Production of the observed compositional variation in the Cathedral Peak granodiorites by this method is considered unlikely for the following reasons: (1) microgranular mafic/ intermediate enclaves are sparse within the Cathedral Peak Granodiorite, typically comprising 1% of the total outcrop; (2) mixing models we tested require a larger mass fraction of felsic relative to andesitic magma (approximately 60:40) to produce the hybrid granodiorite, which is unlikely because of thermal and rheological limits (Sparks & Marshall 1986; Frost & Mahood 1987); (3) a large compositional gap (c. 9 wt% SiO2) exists between andesitic enclave compositions and the lowest silica Cathedral Peak compositions (Fig. 5). The thermal and rheological limitations might be circumvented if andesitic enclave magma were hydrous, as seems likely (Sisson et al. 1996), and it is also possible that fractionation of enclave-forming magma deeper in the magma chamber produced melts that then mixed with granite magma to make the main granodiorite magmas (e.g. Bateman 1995; Wiebe 1996). If such hybrids exist then they are buried beneath the exposed structural levels, and our observation of enclave mineralogy and colour index in the field indicates that few are more felsic than the one we analysed in this study (BTL018). Additionally, there is little evidence at the mineral scale for large-scale hybridization among strongly contrasting magmas. Expected features such as complex zoning, mottled cores and reverse zoning of plagioclase that occur commonly as a result of magma mixing are not observed, and within a single sample, the An# is relatively restricted (see supplementary table 1). Finally, it should be noted that partial melting experiments on high-K mafic rocks that are thought to be representative of Sierran granitoid protoliths (cf. Sisson et al. 2005; Ratakjeski et al. 2005) also produce linear major element trends for experimental melt compositions over the range of likely SiO2 values in the Cathedral Peak granitoids. Thus it is also possible to have repeated injection of diverse felsic magmas from the source with very limited or no fractionation or mixing in the magma chamber and still produce compositional variation like that seen in the Cathedral Peak Granodiorite. It is clear from the above arguments that major element trends cannot definitively argue for or against fractionation, magma mixing or melt source-dominated chemical variation in the Cathedral Peak unit. These processes must therefore be examined using other data, including trace elements and isotopes. Trace element variation Trace elements that are generally compatible in mafic silicates (especially

227

hornblende) show scattered but generally decreasing trends with SiO2 (e.g. Sc, Y). These could be produced by fractional crystallization but similar trends would also be expected for compatible elements in variable-degree partial melts of mafic (e.g. amphibolitic) protoliths, which, as mentioned above, have been proposed to be the source of Sierran granitoids like Cathedral Peak (Sisson et al. 2005; Ratakeski et al. 2005). Feldspar-compatible elements (e.g. Sr and Ba), some high field strength elements (e.g. Zr, Nb) and the generally strongly incompatible elements (e.g. Th, U) typically show the greatest scatter, both with longitude and with SiO2 (Fig. 7). Scatter among incompatible elements, particularly when plotted against each other, is too large for simple binary mixing or single pulse closed system fractional crystallization to apply. If mixing and/or fractionation operated, then it must have involved recharge with multiple pulses of magma with variable incompatible element contents, and/or modification by fractionation of magma in the magma chamber between recharge events. Some of the observed scatter could be explained by limited mixing or fractionation of multiple inputs of magma, where initial trace element composition is controlled mainly by variations in the modal mineralogy of the melt source and/or degree of melting (i.e. the granitoids are mainly ‘imaging’ their melt sources; cf. Miller et al. 1988; Pressley & Brown 1999; Gray 2003; Clemens et al. 2005). Coupled Rb –Sr –Ba variation is typical for feldspar fractionation, although granodiorite melts formed by partial melting of a plagioclase-poor amphibolite protolith (Fig. 14a & b) could produce some of the scatter observed at lower SiO2. The high Sr contents observed in the granodiorites could not be reached without unrealistically large melt fractions for most melting scenarios involving substantial plagioclase in the melt source. Additionally, Sr concentrations for Cathedral Peak granodiorites are exceptionally high when compared with other Cretaceous Sierran granitoids. For example, granitoids from the Yosemite Valley Intrusive Suite just to the west of the Tuolumne batholith (Ratajeski et al. 2001) have a factor of 2 lower Sr. Thus either the melt source for the Cathedral Peak Granodiorite was plagioclase-poor (e.g. garnet amphibolite) or the high-Sr granodiorite melts were produced by fractionation of more mafic melts earlier in their history (e.g. amphibole or pyroxene fractionation at higher pressures; cf. Garrison & Davidson 2003). The subsequent evolution to lower Sr values in more evolved granitoid compositions would reflect fractionation of feldspar. Variation of Rb v. Ba (Fig. 14b) is especially difficult to

228

S. D. BURGESS & J. S. MILLER

Fig. 14. (a) Strontium v. barium for Cathedral Peak rocks including compositional outliers. (b) Barium v. rubidium for Cathedral Peak rocks including compositional outliers. Mineral fraction vectors (note log versus linear scale on plates (a) and (b). Batch melting curves representing melt fractions from 100 to 50% for Plagioclase and K-feldspar compositions are shown on both (a) & (b). Plot (b) shows partial melting curves for a low D (DBa ¼ 0.25, DRb ¼ 0.03) and high D (DBa ¼ 0.6, DRb ¼ 0.3) Ba– Rb. Initial Ba concentration is 400 ppm and initial Rb concentration is 25 ppm. Mineral/melt distribution coefficients for Ba, Rb and Sr were taken from http://earthref.org/GERM/.

explain by variations in partial melting because in all plausible partial-melting scenarios both elements should behave incompatibly. In Cathedral Peak rocks, Ba v. Rb shows a scattered but

generally negative correlation that is at a high angle to the partial-melting model curves (i.e. changing the Ba –Rb ratio by partial melting is difficult).

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

Again, not knowing the parent magma leaves no way of gauging the degree to which any granodiorite sample has lost melt, and so the degree to which any granodiorite is more or less cumulate (e.g. McCarthy & Groves 1979) is difficult to ascertain. The instantaneous mineral ‘cumulates’ will be correspondingly richer in their respective compatible elements (i.e. Sr in plagioclase and Ba and Sr in K-feldspar). Because K-feldspars typically have 64 –65 wt% silica, slight variations in the amount of accumulated K-feldspar would produce an almost imperceptible change in SiO2, but could produce fairly large differences in Ba concentrations. Also, if melt is lost during fractionation, mixing the partially accumulative magma with new batches of melt from the source could generate scatter in feldspar compatible elements. It is also important to try to determine if variations seen in REE patterns in the granitoids could result from variations in the degree of melting of the source or from fractionation during solidification. The combination of decreasing total REE and increasing Ce/YbN with increasing SiO2 is most easily explained by fractionation of the common accessories (titanite, apatite and zircon) and possibly also hornblende (cf. Frey et al. 1978). The most leucocratic samples (i.e. 75 wt% SiO2) show especially strong depletion of light and middle REEs, which can be attributed to fractionation of titanite and allanite, both common accessories in Cathedral Peak rocks. Although low-degree partial melt fractions of a mafic protolith would probably have higher Ce/ YbN (particularly if garnet or amphibole are present) relative to higher-degree partial melts, total REE abundances should also be higher (especially the LREEs) in low-degree partial melts. The presence in the melt source of early saturating REE-concentrating accessory minerals might circumvent this problem, but seems unlikely to be able to produce the necessary depletions in resultant melts. Apatite/melt and titanite/melt partition coefficients for the REEs are not generally large enough to overwhelm the effects of melting in the presence of abundant hornblende if the melt protolith is an amphibolite. Therefore, we conclude that their effect was probably minimal. Relatively low total Zr for the Cathedral Peak granodiorites also indicates that initial melts were probably undersaturated in zircon at the melt source at the likely anatectic temperatures (cf. Miller et al. 2007). We thus maintain that the changing REE patterns in the Cathedral Peak are controlled mainly by fractionation of accessory minerals in the solidifying pluton, once saturation in these phases was reached, rather than by residual accessories in the melt source.

229

The Eu anomaly problem The lack of negative Eu anomalies in REE patterns is a problem if the trace element variation observed in Cathedral Peak is caused by fractionation involving substantial plagioclase. Specifically, negative Eu anomalies would presumably be produced in silicic melts fractionated from plagioclase-rich feldspar cumulates. In addition, if the granitoids have lost extracted silicic melt on a large scale then we might also expect to see positive Eu anomalies (e.g. Glazner et al. 2006). It may be that the modest variation we observe in the feldspar-compatible trace elements is primarily caused by K-feldspar fractionation with more limited involvement of plagioclase, although if fractionation is at least partly responsible for the compositional variation observed in Cathedral Peak, then at some point plagioclase was almost certainly fractionating, and textural evidence suggests that plagioclase was early-saturating. Lipman (2007) also noted that sub-volcanic, caldera-related intrusions in the Southern Rocky Mountain volcanic field tend to lack Eu anomalies, even though large-volume (.103 km3) ignimbrites that were erupted from the co-genetic and related calderas often show distinct negative Eu anomalies. Lipman proposed (rhetorically) several possible ways in which this might arise: (1) crystal residue might be located at deeper structural levels; (2) magma chamber recharge with more primitive (and less fractionated) magma progressively buffers the trace element content of the magma chamber over time, and the resulting large-volume crystal mush does not therefore show the effects of fractionated melt loss; (3) the complementary cumulate is simply diluted in the volume of unerupted magma (the pluton) that may not have experienced melt loss. In light of our previous discussion, we consider the last two especially plausible scenarios for the Cathedral Peak Granodiorite. It is impossible to know to what extent crystals and melt were redistributed, and to what extent recharge may have obscured the REE patterns. We also concur with point (1) above in that, while negative Eu anomalies in felsic rocks on a broad scale require the existence of a complementary residue somewhere (possibly in the lower crust), the presence or absence of an Eu anomaly as it relates to a specific mid-crustal or upper-crustal subvolcanic pluton does not exclude it from being partially accumulative unless one can unambiguously relate the relative magnitude of the Eu depletion to a plausible parent. If, for example, parent melts had negative Eu anomalies then plagioclase accumulation would tend to reduce the Eu anomaly. We also raise the possibility that Eu anomalies in general are not observed in Cathedral Peak Granodiorite because of relatively high f O2,

230

S. D. BURGESS & J. S. MILLER

which resulted in low Eu2þ/Eu3þ in the melt, and therefore no appreciable Eu anomaly in either fractionated rocks or cumulates. The Cathedral Peak Granodiorite (and most of the TB) is characterized by the oxygen buffer assemblage titanite þ magnetite þ quartz, which is indicative of highly oxidizing conditions during crystallization (Wones 1989). High fO2 is also indicated by the low Fe/Fe þ Mg contents of hornblendes from the Tuolumne, which are among the lowest reported by Dodge et al. (1968) for Sierran plutons regionally. Gray (2003) analysed hornblende from the Cathedral Peak Granodiorite and also found low Fe/Fe þ Mg. This all points to oxidizing conditions, probably one to two log units above QFM over the temperature interval of crystallization (see Wones 1989 for discussion), and strong f O2-dependence of Eu partitioning into plagioclase has been well established experimentally (e.g. Wilke & Behrens 1999). AignerTorres et al. (2007) found that at QFM þ 2, the ratio of Eu2þ/Eu3þ was so low that crystallization of plagioclase will not generate any significant Eu anomaly. The schlieren effect and trace element heterogeneity Although we have discounted the schlieren as cumulates related to generation of the main Cathedral Peak rocks, they nevertheless may impact the geochemistry of the pluton (Reid et al. 1993). Disruption and disaggregation of schlieren is observed in the field and provides at least some very direct evidence for mixing involving crystal mush. The commensurate effect on trace elements is likely to be much more dramatic than simple liquid–liquid mixing, especially on a local scale. Sample 31A, for example, contains .1500 ppm Zr, .60 ppm Nb and Y, .400 ppm Ce and .100 ppm Th (and is correspondingly depleted in feldspar compatible elements and quartz). This particular schlieren layer is only a few centimetres thick but the kilometre-scale zones of schlieren like those discussed earlier constitute a potential ‘reservoir’ of considerable trace element heterogeneity. Given the impressive complement of trace elements contained in one schlieren, it seems unavoidable that if mixing with cumulate schlieren occurs on a broader scale during magma recharge then considerable trace element heterogeneity will be imparted to the final pluton. Sr and Nd isotopic variation In comparison to the entire Tuolumne batholith (Fig. 12) the Cathedral Peak unit generally has higher 87Sr/86Sr(i) and lower 1Nd(t) than all other units (except Johnson Granite) as was noted previously by Kistler et al. (1986), Reid et al. (1993) and Gray (2003). This clearly suggests a greater contribution of crustal

input with time, but as with the age variability, the greatest isotopic variability is observed in the outer units and the spread of isotopic composition with time diminishes. This observation was also noted by Bateman (1992). The generally narrow range of 87Sr/86Sr(i) and 1Nd(t) for the Cathedral Peak unit is consistent with a more or less closed-system fractionation process (e.g. Bateman 1992) (Fig. 12), or with very effective source isotopic homogenization of different melt sources. Over the entire silica range observed in the Cathedral Peak Granodiorite (48– 77 wt%), 1Nd(t) values overlap substantially within error (Fig. 12). If the trend of decreasing 87 Sr/86Sr(i) with SiO2 is indeed real, it suggests that periodic recharge into the Cathedral Peak magma chamber involved input of less radiogenic (with respect to Sr) melts over time as the pluton was constructed. Whether or not the decreasing amount of isotopic variation observed within more interior and progressively younger units signifies a greater role for closed system differentiation within the whole batholith or more effective isotopic homogenization of the melt source, it probably indicates a chemical and thermal maturation of the felsic magma system with time. The same trend toward more silicic, larger, and more crustal isotopic compositions, but more limited isotopic variability, is also observed in large, long-lived caldera complexes (De Silva et al. 2006). We suggest, following others (De Silva et al. 2006; Hildreth & Wilson 2007), that this represents the growth and maintenance of a large magma body toward the terminal stages of the growth of the entire Tuolumne Batholith.

Conclusions 1.

2.

Uranium–lead zircon geochronology from this study and from previous work suggests that, although the entire Tuolumne Batholith was constructed over an 8–10 Ma time period, the volumetrically dominant Cathedral Peak unit was constructed within the last 1–2 Ma of this interval. These age constraints allow for extensive magma interaction during construction. This hypothesis is supported by field evidence of distinct contact types in the study area, which probably indicate the presence of mush within the evolving magma system and the repeated separation of crystals from melt within this mush and reworking of this mush, either by late inputs and/or internal processes (e.g. transient convection) that destabilized and mobilized the mush.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE

3.

4.

5. 6.

Major element trends from the Cathedral Peak are potentially compatible with fractionation of mafic silicates, feldspars (particularly plagioclase) and accessory minerals. Major element variation is not likely to be the result of mixing andesite enclave-forming magma with granitic magma because (1) these enclaves are spare within the unit, (2) thermal and rheological limits preclude the possibility of attaining the appropriate felsic/mafic mass fraction and (3) there are no rocks at the exposed structural level with silica concentration between andesitic enclave and the most mafic granodiorites in the unit. Trace element variation can be imparted to the magma by either variations in the modal mineralogy of the melt source and/or degree of partial melting, or by modification of the magma in-situ by fractionation, mixing or a combination of the two. A source-dominated scenario is unlikely to produce the observed feldspar-compatible trace element scatter because (1) Sr concentration in the Cathedral Peak samples require unrealistically large melt fractions for models involving the likely plagioclase-rich parent, (2) Rb and Ba (which should behave incompatibly) show a generally negative correlation at a high angle to partial melting model curves. Scatter among these trace elements is too great for simple binary mixing or single-pulse closed-system fractional crystallization to apply. If these processes operated, they must have involved periodic recharge with multiple pulses of magma with variable trace element compositions. Additionally, feldspar-compatible element variation could be produced if fractionation resulted in melt loss and subsequent mixing of new melt with partially accumulative crystal mush. Variation in REE patterns in the Cathedral Peak unit are likely to be controlled by fractionation of accessory minerals primarily. The generally narrow range of 87Sr/86Sr(i) and 1Nd over the entire silica range observed in the Cathedral Peak unit is consistent with a more or less closed-system fractionation process or with effective isotopic homogenization (at the source) of different melt sources.

Many of the observations above suggest that the Cathedral Peak fits with a ‘silicic mush model’ (Hildreth & Wilson 2007), wherein inputs of felsic magma over time produce growth of a crystal mush body. Additionally, fractionation from the mush imparts considerable heterogeneity in trace elements, but isotopic compositions are relatively invariant.

231

The expanded and comprehensive data set of trace elements on pumices from the Bishop Tuff shows considerable heterogeneity, which Hildreth & Wilson (2007) argue is preserved in a zoned silicic cap above a large mush body. Such heterogeneity should also be preserved in the pluton beneath the silicic cap, and it may well be that the Cathedral Peak granodiorite represents the solidified plutonic roots of such a silicic mush body, although whether the pluton fed eruptions of evolved rhyolite is unknown. Maintaining appreciable melt probably requires considerable heat and/or volatile input, and implies that mafic/intermediate magmas were present in the roots of the Cathedral Peak during intrusion and solidification, and the widely dispersed mafic enclaves in the Cathedral Peak are evidence that mafic magmas probably helped to maintain the system above the solidus, even if they did not mix with the granodiorite on a large scale (e.g. Bachmann & Bergantz 2006). The extent to which any given domain in the Cathedral Peak was largely liquid at the same time will probably never be satisfactorily answered, but we recognize within the exposed geology the superimposed events of mafic (in the form of enclaves) and felsic recharge, fractionation and redistribution of melt and crystals. To develop these features along with the overall isotopic homogeneity and the overall chemical zoning, as modest and disorganized as it is, would seem to require fairly large quantities of melt during solidification (perhaps periodically on short timescales), and maintenance of the system in at least a melt present (mushy) state over a longer period of time. The authors would like to thank R. Trumbull and R. A. Wiebe for very thorough and constructive reviews. We thank D. Coleman for his support in the Isotope Geochemistry and Geochronology Laboratory at UNC and T. Sisson for his support on the electron microprobe at the United States Geological Survey in Menlo Park, CA, USA. We would also like to thank R. Miller, J. Matzel, E. Metzger, S. Paterson, A. Glazner, D. Coleman, T. Sisson, G. Bergantz, and V. Memeti for many helpful and stimulating discussions and D. Redell, A. Mlinarevic and R. Karpowicz for field assistance. This work was supported by EAR-0074099 from the National Science Foundation.

References A IGNER -T ORRES , M., B LUNDY , J., U LMER , P. & P ETTKE , T. 2007. Laser ablation ICPMS study of trace element partitioning between plagioclase and basaltic melts: an experimental approach. Contributions to Mineralogy and Petrology, 153, 647–667. B ACHMANN , O. & B ERGANTZ , G. W. 2006, Gas percolation in upper-crustal silicic crystal mushes as a

232

S. D. BURGESS & J. S. MILLER

mechanism for upward head advection and rejuvenation of near-solidus magma bodies. Journal of Volcanology and Geothermal Research, 149, 82– 102. B ARRIERE , M. 1981. On curved laminae, graded layers, convection currents and dynamic crystal sorting in the Plumanac’h (Brittany) subalkaline granite. Contributions to Mineralogy and Petrology, 77, 214–224. B ATEMAN , P. C. 1992. Plutonism in the Central Part of the Sierra Nevada Batholith, California. United States Geological Survey Professional Paper, 1483. B ATEMAN , P. C. 1995. The interplay between crystallization, replenishment and hybridization in large felsic magma chambers. Earth Science Reviews, 39, 91–106. B ATEMAN , P. C. & C HAPPELL , B. W. 1979. Crystallization, fractionation, and solidification of the Tuolumne intrusive Series, Yosemite National Park, California. Geological Society of America Bulletin, 90, 465–482. B ATEMAN , P. C. & N OKELBERG , W. J. 1978. Solidification of the Mount Givens Granodiorite, Sierra Nevada, California. Journal of Geology, 86, 563–579. B ATEMAN , P. C., K ISTLER , W. R., P ECK , D. L. & B USACCA , A. J. 1988. Tuolumne Meadows quadrangle, California–analytic data. United States Geological Survey Bulletin, 1819. B ERGANTZ , G. W. 2000. On the dynamics of magma mixing by reintrusion: implications for pluton assembly processes. Journal of Structural Geology, 22, 1297–1309. C HAPPELL , B. W. & W HITE , A. J. R. 1974. Two contrasting granite types. Pacific Geology, 8, 173–174. C LEMENS , J. D., H ELPS , P. A. & S TEVENS , G. 2005. Chemical variations in granitic magmas: source-inherited or products of magmatic processes? EOS Transactions of the AGU, 86, Joint Assembly Supplemental, Abstract V22A-05. C OLEMAN , D. S. & G LAZNER , A. F. 1997. The Sierra Crest magmatic event: rapid formation of juvenile crust during the Late Cretaceous in California. International Geology Review, 39, 768 –787. C OLEMAN , D. S., G RAY , W. & G LAZNER , A. F. 2004. Rethinking the emplacement and evolution of zoned plutons: geochronologic evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology, 32, 433–436. C OMPTON , R. R. 1955. Trondhjemite batholith near Bidwell Bar, California. Geological Society of America Bulletin, 66, 9 –44. D E S ILVA , S. L., Z ANDT , G., T RUMBULL , R. & V IRAMONTE , J. 2006, Large scale silicic volcanism – the result of thermal maturation of the crust. In: C HEN , Y. T. (ed.) Advances in Geosciences. World Scientific Press, 215– 230. D ODGE , F. C. W., P APIKE , J. J. & M AYS , R. E. 1968. Hornblendes from granitic rocks of Central Sierra Nevada Batholith, California. Journal of Petrology, 9, 378– 410. F LECK , R. J., K ISTLER , R. W. & W OODEN , J. L. 1996. Geochronologic complexities related to multiple emplacement history of the Tuolumne Intrusive Suite, Yosemite National Park, California. [Abstract.] Geological Society of America Abstracts with Program, 28, 65–66. F REY , F. A., C HAPPELL , B. W. & R OY , S. D. 1978. Fractionation of rare-earth elements in the Tuolumne

Intrusive Series, Sierra Nevada Batholith, California. Geology, 6, 239–242. F ROST , T. P. & M AHOOD , G. A. 1987. Field, chemical, and physical constraints on mafic-felsic interaction in the Lamarck Granodiorite, Sierra Nevada, California. Geological Society of America Bulletin, 99, 272– 291. G ARRISON , J. M. & D AVIDSON , J. P. 2003. Dubious case for slab melting in the Northern volcanic zone of the Andes. Geology, 31, 565–568. G LAZNER , A. F., B ARTLEY , J. M., C OLEMAN , D. S., G RAY , W. & T AYLOR , R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today, 14(4/5), 4– 10. G LAZNER , A. F., C OLEMAN , D. S., B ARTLEY , J. M. & W OHLETZ , K. H. 2006. What plutons have to say about big eruptions. Eos Transactions of the AGU, 87, Fall Meeting Supplement, Abstract V23G-05. G RAY , W. 2003. Chemical and Thermal evolution of the Late Cretaceous Tuolumne Intrusive Suite, Yosemite National Park, California, PhD Dissertation, Chapel Hill, NC, University of North Carolina, 202. H ALLIDAY , A. N., S TEPHENS , W. E. & H ARMON , R. S. 1980. Rb–Sr and O isotopic relationships in three zoned Caledonian granitic plutons, Southern Uplands, Scotland: evidence for varied sources and hybridization of magmas. Journal of the Geological Society of London, 137, 329 –348. H ILDRETH , W. & W ILSON , C. J. N. 2007. Compositional zoning of the Bishop Tuff. Journal of Petrology, 48, 951–999. H ILL , R. I. 1988. San Jacinto intrusive complex 1. Geology and mineral chemistry, and a model for intermittent recharge of tonalitic magma chambers. Journal of Geophysical Research, 93, 10 325– 10 348. H ILL , M., O’N EIL , J. R., N OYES , H., F REY , F. A. & W ONES , D. R. 1988. Sr, Nd and O isotope variations in compositionally zoned and unzoned plutons in the central Sierra Nevada Batholith. American Journal of Science, 288-A, 213– 241. K ISTLER , R. W. & F LECK , R. J. 1994. Field Guide for a Transect of the Central Sierra Nevada, California. Geochronology and isotope geology: United States Geological Society open file report, 94–267. K ISTLER , R. W., C HAPPELL , B. W., P ECK , D. L. & B ATEMAN , P. C. 1986. Isotopic variation in the Tuolumne Intrusive Suite, central Sierra Nevada, California. Contributions to Mineralogy and Petrology, 94, 205–220. L IPMAN , P. W. 2007. Incremental assembly and prolonged consolidation of Cordilleran magma chambers: evidence from the Southern Rocky Mountain volcanic field. Geosphere, 3, 27–70. M ATZEL , J. E. P., M ILLER , J. S., M UNDIL , R. & P ATERSON , S. R. 2006. Zircon saturation and the growth of the Cathedral Peak pluton, CA. 16th annual V.M. Goldschmidt Conference, abstract, ,www.goldschmidt2006.org/cd/goldschmidt/ 0608025_matzel01528.html.. M ATZEL , J., M UNDIL , R., P ATERSON , S., R ENNE , P. & N OMADE , S. 2005, Evaluating pluton growth models using high-resolution geochronology: Tuolumne Intrusive Suite, Sierra Nevada, CA. Geological Society of America Abstracts with Programs, 37, 131.

EVOLUTION OF CATHEDRAL PEAK GRANODIORITE M C C ARTHY , T. S. & G ROVES , D. I. 1979. The Blue Tier Batholith, Northeastern Tasmania. Contributions to Mineralogy and Petrology, 71, 193–209. M ILLER , J. S., M ATZEL , J. E. P., M ILLER , C. F., B URGESS , S. D. & M ILLER , R. B. 2007. Zircon growth and recycling during the assembly of large, composite arc plutons. Journal of Volcanology and Geothermal Research, DOI: 10.1016/ j.jvolgeores.2007.04.019. M ILLER , C. F., W ATSON , E. B. & H ARRISON , T. M. 1988. Perspectives on the source, segregation, and transport of granitoid magmas. Transactions of the Royal Society of Edinburgh, 79, 135–156. N ABELEK , P. I. 1986. Trace-element modeling of the petrogenesis of granophyres and aplites in the Notch Peak granitic stock, Utah. American Mineralogist, 71, 460–471. P RESSLEY , R. A. & B ROWN , M. 1999. The Phillips Pluton, Maine, USA: evidence of heterogeneous crustal sources, and implications for granite ascent and emplacement mechanisms in convergent orogens. Lithos, 46, 335–366. R AGLAND , P. C. & B UTLER , J. R. 1972. Crystallization of the West Farrington Pluton, North Carolina, USA. Journal of Petrology, 13, 381–404. R ATAJESKI , K., G LAZNER , A. F. & M ILLER , B. V. 2001. Geology and geochemistry of mafic to felsic plutonic rocks in the Cretaceous intrusive suite of Yosemite Valley, California. Geological Society of America Bulletin, 113, 1486– 1502. R ATAJESKI , K., S ISSON , T. W. & G LAZNER , A. F. 2005. Experimental and geochemical evidence for derivation of the El Capitan Granite, California by partial melting of hydrous gabbroic lower crust. Contributions to Mineralogy and Petrology, 149, 713–734. R EID , J. B., E VANS , O. C. & F ATES , D. G. 1983. magma mixing in granitic rocks of the central Sierra Nevada, California. Earth and Planetary Science Letters, 66, 243–261. R EID , J. B., M URRAY , D. P., H ERMES , O. D. & S TEIG , E. J. 1993. Fractional crystallization in granites of the Sierra Nevada: How important is it? Geology, 21, 587–590. S AWKA , W. N., C HAPPELL , B. W. & K ISTLER , R. W. 1990, Granitoid compositional zoning by side-wall boundary layer differentiation: evidence from the Palisade Crest Intrusive Suite, Central Sierra Nevada, California. Journal of Petrology, 31, 519– 553. S ISSON , T. W., G ROVE , T. L. & C OLEMAN , D. S. 1996. Hornblende gabbro sill complex at Onion Valley, California, and a mixing origin for the Sierra Nevada batholith. Contributions to Mineralogy and Petrology, 126, 81–108. S ISSON , T. W., R ATAJESKI , K., H ANKINS , W. B. & G LAZNER , A. F. 2005. Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology, 148, 635–661.

233

S PARKS , R. S. J. & M ARSHALL , L. A. 1986. Thermal and mechanical constraints on mixing between mafic and silicic magmas. Journal of Volcanology and Geothermal Research, 29, 99– 124. S TACEY , J. S. & K RAMERS , J. D. 1975. Approximation of terrestrial lead isotope evolution by a two-stage mode. Earth and Planetary Science Letters, 26, 207– 221. S TEIGER , R. H. & J AGER , E. 1977. Subcommission on geochronology: convention on the use of decay constants in geo-and cosmo-chronology. Earth and Planetary Science Letters, 36, 359–362. S TERN , T. W., B ATEMAN , P. C., M ORGAN , B. A., N EWELL , M. F. & P ECK , D. L. 1981. Isotopic U–Pb Ages of Zircon form the Granitoids of the Central Sierra Nevada, California. United States Geological Survey Professional Papers, 1185. W ALAWENDER , M. J., G ASTIL , R. G. ET AL . 1990. Origin and evolution of the zoned La Posta-Type plutons, eastern Peninsular Ranges batholith, southern and Baja California. In: A NDERSON , J. L. (ed.) The Geology of North America, The Nature and Origin of Cordilleran Magmatism. Geological Society of America Memoir, 174, 1 –18. W EINBERG , R. F., S IAN , A. N. & P ESSOA , R. R. 2001. Magma flow within the Tavares pluton, northeastern Brazil: compositional and thermal convection. Geological Society of America Bulletin, 113, 508–520. W IEBE , R. A. 1996. Mafic-silicic layered intrusions: the role of basaltic injections in the magmatic process and evolution in silicic magma chambers. Transactions of the Royal Society of Edinburgh: Earth Sciences, 87, 233– 242. W IEBE , R. A., J ELLINEK , M., M ARKLEY , M. J., H AWKINS , D. P. & S NYDER , D. 2007. Steep schlieren and associated enclaves in the Vinalhaven granite, Maine: possible indicators for granite rheology. Contributions to Mineralogy and Petrology, 153, 121– 138. W ILKE , M. & B EHRENS , H. 1999. The dependence of the partitioning of iron and europium between plagioclase and hydrous tonalitic melt on oxygen fugacity. Contributions to Mineralogy and Petrology, 137, 102– 114. W ONES , D. R. 1989. Significance of the assemblage titanite þ magnetite þ quartz in granitic rocks. American Mineralogist, 74, 744– 749. V ANCE , J. A. 1961. Zoned granitic intrusions-an alternative. Geological Society of America Bulletin, 72, 1732–1738. Z AK , K. & P ATERSON , S. R. 2005. Characteristics of internal contacts in the Tuolumne Batholith, central Sierra Nevada, California (USA): implications for episodic emplacement and physical processes in a continental arc magma chamber. Geological Society of America Bulletin, 117, 1242– 1255.

Snake River Plain – Yellowstone silicic volcanism: implications for magma genesis and magma fluxes WILLIAM P. LEEMAN1, CATHERINE ANNEN2 & JOSEF DUFEK3 1

National Science Foundation, Earth Sciences Division, 4201 Wilson Blvd, Arlington, VA 22330, USA (e-mail: [email protected]) 2

De´partment de Mine´ralogie, University of Geneva, 1205 Geneva, Switzerland

3

Department of Earth and Planetary Science, University of California, Berkeley, CA 94720, USA Abstract: The origin of large-volume, high-temperature silicic volcanism associated with onset of the Snake River Plain – Yellowstone (SRPY) hotspot track is addressed based on evolution of the well-characterized Miocene Bruneau– Jarbidge (BJ) eruptive centre. Although O –Sr –Pb isotopic and bulk compositions of BJ rhyolites exhibit strong crustal affinity, including strong 18 O-depletion, Nd isotopic data preclude wholesale melting of ancient basement rocks and implicate involvement of a juvenile component – possibly derived from contemporaneous basaltic magmas. Several lines of evidence, including limits on 18O-depletion of the rhyolite source rocks due to influx of meteoric/hydrothermal fluids, constrain rhyolite generation to depths shallower than mid-upper crust (,20 km depth). For crustal melting driven by basaltic intrusions, sustenance of temperatures exceeding 900 8C at such depths over the life of the BJ eruptive centre requires incremental intrusion of approximately 16 km of basalt into the crust. This minimum basaltic flux (c. 4 mm year21) is about one-tenth that at Kilauea. Nevertheless, emplacement of such volumes of magma in the crust creates a serious room problem, requiring that the crust must undergo significant extensional deformation – seemingly exceeding present estimates of extensional strain for the SRPY province.

This paper addresses two related issues – each of which carries important broader implications. The first concerns the origin of large-volume, hightemperature silicic magmas associated with the Yellowstone hotspot track. Assuming they form largely in response to intrusion of voluminous basaltic magma into the crust, we are interested in knowing how much basalt is required, and at what depths that magma is emplaced. Our approach is to evaluate the conditions required to elevate crustal temperatures to magmatic values over volumes that are sufficiently large to generate the quantities of magma produced. Also, many rhyolites from this province exhibit 18O-depletion, and this characteristic is widely attributed to involvement of low-d18O meteoric waters. Assuming that this anomaly is source-related, the depths to which surface fluids can plausibly circulate and effectively alter the oxygen isotopic composition of large volumes of crust provide an important constraint on maximum depth of crustal melting. Thus, a second issue concerns how and where the magmas acquire this signal, and what significance this holds regarding both rhyolite petrogenesis and fluid flow in the crust. Although we focus mainly on the physical processes involved, an important side issue concerns the relative contributions of pre-existing

crustal rocks v. basalt-derived components in forming the rhyolite magmas; this, in turn, bears on how much older crust is melted. As a framework for the problem, discussion is centred mainly on the relatively well-characterized Miocene Bruneau –Jarbidge (BJ) volcanic field in the central Snake River Plain (cf. Bonnichsen et al. 2008). This area is examined because, unlike Yellowstone itself, (1) the entire cycle of BJ silicic magmatism has run to completion, and (2) all BJ rhyolites exhibit strong 18O-depletion – collectively comprising the most voluminous occurrence of such magmas known in the world (Boroughs et al. 2005; Cathey et al. 2007). We envisage that the processes underlying BJ magmatism were repeated in a diachronous fashion as magmatism migrated laterally over time. Two investigative approaches are combined to constrain the depth of rhyolite genesis. On one hand, the capacity of meteoric/hydrothermal fluids to modify a substantial volume of crust diminishes exponentially with depth (and decreasing crustal permeability) such that 18O-depleted sources for BJ rhyolites are probably confined to the upper crust. On the other hand, even under optimal conditions, production of just the minimal estimated volume of BJ rhyolite requires rather extreme inputs of basaltic

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 235–259. DOI: 10.1144/SP304.12 0305-8719/08/$15.00 # The Geological Society of London 2008.

236

W. P. LEEMAN ET AL.

magma into the crust in order to maintain sufficiently high temperatures at those depths. The consequences of our models regarding crustal/lithospheric modification are profound and require confirmation by additional approaches.

Background: fundamental characteristics of the SRPY magmatic province Volcano-tectonic overview The Neogene Snake River Plain – Yellowstone (SRPY) province is one of the most prolific volcanic systems in the western United States. Volcanism can be generalized in terms of migration of a series of volcanic centres across southern Idaho (Fig. 1) starting in north-central Nevada (at c. 16.5 Ma) and progressing to the currently active Yellowstone Plateau (at c. 2 Ma) in northwestern

Wyoming (Armstrong et al. 1975; Pierce & Morgan 1992). All eruptive centres in this province exhibit a two-stage history: first (1) an initial phase (lasting c. 2 to as much as 4 Ma) that produced voluminous high-silica rhyolite lavas and ignimbrites; then (2) a culminating phase of predominantly basaltic activity. Although onset of basaltic volcanism was delayed relative to appearance of the earliest rhyolites at any given centre, basaltic magmatism is inferred to be coeval with (or to have even preceded) the earliest rhyolites (cf. Hildreth et al. 1991). Bonnichsen et al. (2008) also inferred from widespread (over some 400 km) lateral distribution of Miocene (11 – 10 Ma) rhyolites that basaltic intrusions must have been similarly distributed in the crust by that time. Moreover, once initiated, basaltic eruptions continued intermittently – up to Quaternary time across much of the province. For example, the Yellowstone Plateau volcanic field is considered to be in

Fig. 1. Map showing general loci and ages of initial eruptions for silicic eruptive centres ascribed to activity of the Yellowstone ‘hotspot’ (after Pierce & Morgan 1991). These include (from oldest to youngest): McDermitt (McD), Owyhee-Humboldt (OH), Bruneau– Jarbidge (BJ), Twin Falls (TF), Picabo (P), Heise (H) and Yellowstone Plateau (YP). The Shaded line connects centroids of the eruptive centres; changes in azimuth relative to absolute motion of North America (arrow labelled NA; Gripp & Gordon, 2002) are probably artefacts of Neogene extensional deformation. Most of the Snake River Plain –Yellowstone province is underlain by Precambrian cratonic continental lithosphere that extends as far west as the 87Sr/86Sr ¼ 0.706 isopleth (Leeman et al. 1992); the region west of this line is underlain by accreted oceanic terranes of Phanerozoic age. The southern lobe of the Idaho batholith and surface exposures of Precambrian basement are shown in shaded fields. Inset shows loci of the BJ and YP centres superimposed on the physiographic area of the Snake River Plain province (WSRP and ESRP refer to western and eastern parts of the Snake River Plain). Nash et al. (2006) discuss possible implications of spacing of the volcanic centres.

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

the early part of its second phase, with most basaltic lavas confined to the last 0.5 Ma (Christiansen 2001). However, basaltic magmatism in this age range also occurred locally across the eastern and central SRP and as far west as the Boise area. Simply put, SRP basaltic magmatism did not follow a simple age – space progression. Despite this, it is widely considered that the SRPY province could be a ‘hotspot track’ produced by migration of North America over an ascending mantle plume or melting anomaly that presently underlies the Yellowstone area (cf. Pierce & Morgan 1991). In support of this view, recent seismic tomographic studies document the presence beneath Yellowstone of a NW-dipping plume-like velocity anomaly that extends some 500 km into the upper mantle (Yuan & Duecker 2005; Waite et al. 2006). Nash et al. (2006) attempt to relate spacing of the volcanic centres to the presence of a plume-tail like feature with a diameter on the order of 70–100 km. However, the style of SRPY volcanism differs dramatically from that associated with oceanic ‘hotspots’, where magmatism generally follows simple age – distance progressions and is fundamentally basaltic in character. Magma generation processes in this setting probably are complicated by the presence of a thick continental lithospheric lid and cratonic crust that are inferred to extend as far west as the 87Sr/86Sr ¼ 0.706 isopleth as measured in Mesozoic to Neogene igneous rocks (cf. Leeman et al. 1992). These authors note that silicic volcanic rocks from centres (e.g. McDermitt) associated with the Snake River Plain trend, but that are situated west of the ‘0.706 line’, have lower 87Sr/86Sr; such compositional variations presumably reflect lateral heterogeneities in the crustal source rocks. As a step toward developing a comprehensive physical model for the SRPY magmatic province, here we evaluate the crustal-level processes responsible for the voluminous silicic magmatism and thereby indirectly estimate the required flux of basaltic magma driving the system.

Volume and scale of the rhyolite systems A principle constraint is the volume of silicic magma that was produced over time. Each of the SRPY eruptive centres produced several large rhyolite ignimbrite eruptions of ‘super volcano’ class (1000 km3) along with multiple smaller ignimbrites and numerous lavas (Perkins et al. 1995; Morgan & McIntosh 2005). Currently, volumes are best known in the vicinity of Yellowstone and in the central Snake River Plain (Fig. 2). Yellowstone Plateau (YP) At Yellowstone, it is clear that most rhyolitic eruptions were confined to

237

a well-mapped set of nested calderas, the largest of which is approximately 75 km across in its longest dimension (Christiansen 2001). A variety of geophysical observations indicate the presence of a large magma body at depths between 6 and 10 km below the surface (Eaton et al. 1975; Smith & Braile 1994). Peripheral eruptions of rhyolite lavas and of minor Pleistocene and younger basalts suggest that the areal footprint of the overall magma system is roughly circular with a diameter on the order of 100 km. The cumulative volume of rhyolite extruded in the past 2 Ma at Yellowstone is estimated to be at least 6000 km3 (Hildreth et al. 1991), not including widely dispersed airfall tephra or intracaldera fill. More than half of the total volume is represented by the large Huckleberry Ridge (at least 2500 km3) and Lava Creek (c. 1000 km3) tuffs, erupted at 2.03 and 0.64 Ma, respectively. Central Snake River Plain (CSRP) Source regions for CSRP (acronym used to denote geographic area) silicic magmas are poorly exposed owing to subsidence and burial beneath younger sediments and extensive basalt flows. However, most ignimbrites in this area can be traced back to the so-called Bruneau Jarbidge (BJ; acronym used in referring to these specific volcanic rocks) volcanic centre (Fig. 1) – an area comparable in size and geometry to the Yellowstone Plateau volcanic field. In the eastern part of the map area, a number of younger (,10.2 Ma) ignimbrites overlap the BJ sequence; these presumably erupted from vents within the so-called Twin Falls volcanic field. Bonnichsen et al. (2008) suggest that, between 12.7 and 10.5 Ma, the BJ eruptive centre produced nine significant ignimbrites collectively comprising at least 7000 km3 (based on estimated thickness of outflow deposits integrated over their known distribution area). By analogy with Yellowstone, this early stage is considered to be related to caldera-style eruptions. BJ ignimbrite episodes were generally more frequent than at Yellowstone, but separated by irregular lulls of 200–400 ka duration. Between 10.5 and 8.0 Ma, a change in eruptive style was attended by production of predominantly rhyolitic lavas with individual volumes ranging between 10 and 200 km3. The first basaltic lavas (c. 9.5 Ma) of the BJ field further mark a distinctive developmental change in this eruptive centre. According to current interpretations (Pierce & Morgan 1991), the waning stage of the BJ centre was partly coeval with ignimbrite eruptions from the Twin Falls centre. However, it seems likely that there was a time-transgressive shift in the focus of magmatism. Based on regional tephrochronology studies, Perkins & Nash (2002) estimate that the total volume of rhyolite produced from the CSRP

238

W. P. LEEMAN ET AL.

Fig. 2. Detailed map showing the main focal area of this paper in the western and central Snake River Plain. Location of this area with respect to the broader Snake River Plain (SRP) is indicated in the inset (labels as in Fig. 1). We emphasize relations in the Bruneau– Jarbidge (BJ) volcanic centre that is representative of the greater central Snake River Plain (CSRP) region. We also discuss data from the northeastern flank of the Owyhee Mountains. Shaded circles indicate principal towns. The topographically low SRP is surrounded by highlands shown as shaded pattern.

(between 12.7 and 8.0 Ma) could have exceeded 10 000 km3. The total volume of silicic magma produced should also include an unknown quantity of equivalent intrusive material; by analogy with other treatments (cf. Crisp 1984) this could amount to several times the extrusive volume. Here, we consider alternative volume production scenarios in which the intrusive:extrusive magma ratio is set conservatively at 2:1. Using estimates of extruded volume (7000–10 000 km3), this approach implies a total magma volume on the order of 21 000– 30 000 km3. For comparison, the equivalent volume for the Yellowstone Plateau volcanic field could exceed 10 000 km3 (Hildreth et al. 1991). Minimal magma production estimates for the CSRP are summarized in Table 1 and illustrated in Figure 3. Magma production clearly was non-linear over time. The early ignimbrites and related deposits comprise about 75% (.5000–7500 km3) of the total effusive products, whereas the younger lavas comprise the remainder. The ignimbrite episodes also appear to have varied greatly in volume output, with more than half associated with the three largest units (Cougar Point tuffs VII, XI and XIII, all with volumes exceeding 1000 km3).

Rhyolite petrology and temperature CSRP rhyolites are almost exclusively metaluminous in composition, with SiO2 contents between 71 and 76% for most. With few exceptions, they carry distinctively anhydrous phenocryst assemblages, including plagioclase + sanidine + quartz þ magnetite + ilmenite þ pigeonite þ augite + orthopyroxene + fayalite with accessory zircon and apatite. Amphibole and biotite are conspicuously lacking. A variety of mineral thermometers (two-pyroxene, two-feldspar, two-oxide and Ti in quartz) demonstrate that magmatic temperatures typically exceeded 900 8C and in some cases reached 1000 8C (Honjo et al. 1992; Cathey & Nash 2004; Leeman, unpublished data). Yellowstone rhyolites are generally similar. Although mineral temperatures tend to be somewhat lower (800 –900 8C) for most units, a few approach 1000 8C (Hildreth et al. 1984; Leeman, unpublished data). Rare amphibole (e.g. in member A of the Lava Creek tuff) is consistent with slightly higher volatile content and/or lower temperature for very few units, but otherwise SRPY rhyolite magmas are significantly hotter than those from other

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

239

Table 1. Volume of rhyolite erupted from the CSRP (Bruneau –Jarbidge volcanic centre) CAT Group

Main BJ units

Main product

Age range (Ma)

Volume (km3)

Total volume (%)

Cumulative volume (%)

Effusion rate (km3/Ma)

1 2 3 4 5 6 7 8 9 10 11 12 13

CP III CP V CP VII CP IX CP XI CP XII CP XIII CP XV CT LD-BJ SC DC SF

ign ign ign ign ign ign ign ign ign lf lf lf lf

12.8–12.4 12.4–11.9 11.9–11.7 11.7–11.5 11.5–11.2 11.2–11.0 11.0–10.7 10.7–10.4 10.4–10.0 10.0–9.5 9.5–9.0 9.0–7.5 7.5–5.5

100 300 1200 300 1800 200 1800 200 100 200 400 300 100

1.4 4.3 17.1 4.3 25.7 2.9 25.7 2.9 1.4 2.9 5.7 4.3 1.4

1.4 5.7 22.9 27.1 52.9 55.7 81.4 84.3 85.7 88.6 94.3 98.6 100.0

250 600 6000 1500 6000 1000 6000 667 250 400 800 200 60

Note: volumes are based on stratigraphic data of Bonnichsen et al. (2008) and exclude distal tephras and intracaldera fill; volume is given as a percentage of total, which is estimated to be between 7000 and 10 000 km3. CAT groups include chemically similar units erupted within the indicated time intervals. Principal eruptive units include Cougar Point Tuffs (CP) and large rhyolite lavas: Cedar Tree (CT), Long Draw (LD), Bruneau Jasper (BJ), Sheep Creek (SC), Dorsey Creek (DC) and Shoshone Falls (SF); basaltic lavas first appear between CAT Groups 10 and 11. Products: ignimbrite (ign), lava flow (lf).

settings in the Cordilleran United (Christiansen & McCurry 2008).

States.

Rhyolite compositional variation with time Here we emphasize compositional variations in rhyolitic products of the CSRP based on detailed studies of selected stratigraphic sections, extensive chemical analyses, and new 39Ar/40Ar dating of critical units as synthesized by Bonnichsen et al. (2008). Data from three distal areas were compared on the basis of averaged analyses for individual units, and stratigraphic correlations based on all available data. The temporal evolution of the magmatic system is exemplified by data from the BJ volcanic centre proper. Figure 4 portrays temporal variation in selected compositional parameters. All element concentration or ratio data represent unit averages; isotopic data are individual sample analyses. Notably FeOT, TiO2 and 143Nd/144Nd (also 87Sr/86Sr, Sr, Nb, Y – not shown) display fairly systematic increases with time whereas highly incompatible elements and related ratios (e.g. Rb, Rb/Sr) decrease. The observed trends are antithetical to those expected as a result of fractional crystallization control, albeit minor reversals (e.g. near 11.8, 11.1, 9.5 and around 8 Ma) could signify magmatic differentiation during interludes in eruptive activity. In essence, the rhyolites evolved to less differentiated compositions with time. The overall pattern could be attributed to progressive melting of a common crustal reservoir except that

Nd isotopic data require some degree of open system behaviour that causes rhyolite 143Nd/144Nd values to shift toward the observed basalt range over time. This shift is unlikely to involve simple mixing of basaltic and rhyolitic magmas because initial 87 Sr/86Sr ratios are around 0.709 in the earliest rhyolites and increase to . 0.712 in the youngest lavas, becoming progressively more radiogenic than the basalt range (c. 0.706–0.707; cf. Menzies et al. 1984) over time. Nevertheless, mixing between pure crustal melts and derivative liquids produced from basaltic magma emplaced in the crust is likely to influence compositions of the erupted rhyolites (Hildreth et al. 1991; Annen & Sparks 2002). This topic is revisited shortly. As a final comment, the ‘reverse fractionation’ trend documented for the BJ magma system appears to be unique compared with other magmatic centres associated with the Yellowstone melting anomaly (Bonnichsen et al. 2008). In many respects it resembles the bottom-to-top compositional zoning commonly observed in individual ignimbrites (Hildreth 1981) but, because this trend developed over an order of magnitude longer timescale, it is unlikely to result from evolution within a single long-lived magma chamber (Vazquez & Reid 2002).

Rhyolite sources: evidence for a dominant crustal component Although protoliths for SRPY rhyolite magmas are uncertain, diverse evidence leaves little doubt that

240

W. P. LEEMAN ET AL.

Fig. 3. Temporal variation in relative (a) and cumulative volume (b) of rhyolitic magma extruded from the Bruneau– Jarbidge eruptive centre; these are shown as percentages of the total volume excluding intracaldera fill and distal airfall deposits (for which volumes are unknown). Bonnichsen et al. (2008) estimate minimum volume to be at least 7000 km3; Perkins & Nash (2002) suggest that total volume may have exceeded 10 000 km3. Eruption rates and volumes were clearly highest between 12 and 10 Ma (‘caldera-forming’ stage), with notable spikes corresponding to three ‘supervolcano scale’ ignimbrite events, each having a volume in excess of 1000 km3 (CPT units VII, XI and XIII). Appearance of the first basaltic eruptions (indicated by vertical line) closely coincides with a shift in eruption style to predominantly rhyolite lavas after 10 Ma (‘rifting’ stage).

these melts were largely derived by crustal melting. In most regards, SRPY rhyolites exhibit strong similarities to so-called A-type (or ‘anorogenic’) granitoids. Origins of such magmas have been discussed in detail by Patin˜o-Douce (1997), who noted that their high magmatic temperatures, metaluminous compositions and other geochemical characteristics are consistent with melting of

relatively water-poor calcalkalic igneous rocks at low pressures (,5 kbar) within the shallow crust. He also presented analyses of partial melts of tonalite (30–40% melting) and granodiorite (c. 20% melting) bulk compositions produced at 950 8C and 4 kbar that closely resemble compositions of SRPY rhyolites. A critical factor appears to be low pressure because melts produced at 8 kbar

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

241

Fig. 4. Compositional variations in BJ rhyolite tuffs and lavas as a function of time. Unit averages are plotted from Bonnichsen et al. (2008), with additional Nd isotopic data for BJ tuffs from Cathey & Nash (2004); error bars reflect uncertainties in radiometric ages and standard deviations on unit averages. Nd isotopic data for SRP basalts are from Menzies et al. (1984) and Leeman (unpublished data).

have quite distinctive compositions owing to different residual mineral assemblages. Below, we evaluate model parameters required to consistently sustain the requisite high temperatures within shallow domains of the crust. Differentiation of basaltic magmas also could generate some rhyolitic magma (McCurry et al. 2008), but it is unlikely that SRPY rhyolites could be produced principally by this process. In addition to geochemical inconsistencies noted below, there are problems with the large amounts of basalt required and the paucity of intermediate

composition magmas as discussed by Bonnichsen et al. (2008). Trace element constraints Bonnichsen et al. (2008) demonstrate that strongly to moderately incompatible elements in SRPY rhyolites exhibit similar enrichment levels when normalized to the average crustal composition of Taylor & McLennan (1985). Ignoring elements that are readily influenced by crystallization of observed phenocrysts or accessory phases (e.g. Sr, Ba, Eu), enrichment levels in BJ rhyolites are consistent with 15–25%

242

W. P. LEEMAN ET AL.

partial melting of sources similar to average crust. Data for Yellowstone and other SRP eruptive centres are similar (Hildreth et al. 1984, 1991; Hughes & McCurry 2002; Wright et al. 2002). For purposes of our modelling, we consider this range realistic for melt fraction (F) in the crust, and inferred source volumes may be calculated as estimated eruptive volume (see Table 1) divided by F. For simplicity, an average F value of 0.25 was adopted in estimating source volume. Larger F values imply smaller source volume estimates, and the effects of changing this parameter are illustrated later. Radiogenic isotope constraints The Sr–Nd–Pb isotope geochemistry of SRPY rhyolites is distinctive in several respects (Doe et al. 1982; Hildreth et al. 1991; Leeman et al. 1992; Nash et al. 2006; Bonnichsen et al. 2008). Virtually all are characterized by elevated 87Sr/86Sr (.0.709) and low 143 Nd/144Nd (,0.5123) compared with associated basalts. Pb isotopic ratios are varied, with 206 Pb/204Pb ranging from unradiogenic (,17.5) near Yellowstone to progressively more radiogenic (c. 19.0) in the CSRP area. 206Pb/204Pb– 207Pb/204Pb data define a near-linear array, the slope of which conforms to an Archaean (c. 2.5 Ga) secondary isochron (Leeman, unpublished data). Whether or not this array has strict age significance, the Pb data are consistent with incorporation of significant amounts of old craton-derived crustal Pb in the rhyolite source domain (e.g. Leeman et al. 1985; Wooden & Mueller 1988). On the other hand, Nd isotopic data for SRPY rhyolites are more radiogenic (typically, 1Nd .210) than compositions of Archaean basement rocks (1Nd ,230) in the region (Leeman et al. 1985). Much the same was observed at Yellowstone, although some extracaldera rhyolites have 1Nd values as low as –18 (Hildreth et al. 1991). Overall, these data preclude wholesale melting of Archaean crust and seemingly require involvement of a more juvenile (i.e. recently mantle-derived) Nd component in most of the rhyolites (Nash et al. 2006). Oxygen isotope constraints Oxygen isotopic compositions of SRPY rhyolites vary significantly and clearly point to open system modification of crustal source materials. Based on analyses of quartz, feldspar and zircon phenocrysts, most Yellowstone rhyolites have normal or only slightly depleted magmatic d18O (c. 5–8‰), and some are extremely depleted (with d18O , 2‰). The latter typically followed major caldera collapse events (Hildreth et al. 1984); in these cases, the magmatic oxygen isotopic composition is thought to reflect assimilation (Hildreth et al. 1984; Balsley & Gregory 1998) or remelting (Bindeman & Valley 2001a) of low-d18O hydrothermally altered older

volcanic rocks that subsided catastrophically into the underlying magma reservoir; Hildreth et al. (1984) also consider direct ingress of low-d18O hydrothermal brines into silicic magmas as a plausible alternative. However, recent oxygen isotopic analyses and U – Pb dating of zircons indicates that the assimilated/melted material is mainly cannibalized or recycled from early parts of the Yellowstone magmatic system rather than much older crustal rocks (Bindeman & Valley 2001b; Vazquez & Reid 2002). A similar picture has emerged for the slightly older Heise volcanic centre, except that there the earliest ignimbrites have fairly normal d18O values and low d18O rhyolites are restricted to the youngest, albeit voluminous, Kilgore ignimbrite as well as subsequent post-caldera rhyolite lavas (Bindeman et al. 2007). In dramatic contrast, nearly all rhyolites, including the earliest ones, from the BJ and related eruptive centres in the CSRP, are characterized by d18O-depletion (magmatic values near 21.4 to þ3.8‰, based on feldspar and quartz analyses; Boroughs et al. 2005). Interestingly, partly contemporaneous rhyolites from the Owyhee Front (some 100 km to the NW) have normal d18O (7–9‰). The BJ results are confirmed by analyses of zircons from the Cougar Point tuffs (Cathey et al. 2007). These authors discount assimilation of low-d18O crustal materials by rhyolite magmas as the primary cause for 18O-depletion; rather, their oxygen composition must be inherited from the respective magma sources. Boroughs et al. (2005) attribute CSRP rhyolite compositions to melting of Idaho batholith granitic rocks that were hydrothermally altered during the Eocene (Criss & Fleck 1987), and suggest that the Owyhee Front rhyolites are melts of similar lithologies that somehow escaped alteration. In any case, this work implies that much of the oxygen in these rhyolites is crustal in origin, and that very large volumes of the underlying crust have been modified by interaction with low-d18O waters, presumably of nearsurface origin. Although subsequent events, like those at Yellowstone or Heise, could have contributed to the observed 18O-depletion effects, extensive crustal modification must have preceded generation of the earliest rhyolite magmas. This critical observation dictates that extensive crustal melting occurred at shallow depths, within reach of infiltrating meteoric/hydrothermal waters.

Relative crustal and mantle (i.e. basaltic) contributions to BJ rhyolites Of critical petrogenetic concern is the relative contribution of crust v. mantle components to SRPY rhyolite magmas. Specifically, what fraction of the

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

erupted magma derives from remelting of old pre-existing crust? To address this question we consider variations in O and Nd isotopic and selected major element data for BJ rhyolites in Figure 5. The composition of the ‘purest’ crustal-derived rhyolite (‘C’) is approximated by the most 18O-depleted samples: d18O ¼ c. 0‰, Nd ¼ 65 ppm, 1Nd ¼ 28.4. Major element composition for ‘C’ was based on partial melt of granodiorite at 4 kbar (Patin˜o-Douce 1997): SiO2 ¼ 74.1%, FeOT ¼ 2.2%. BJ area basalt compositions vary somewhat, but a representative analysis (‘B’) is as follows: d18O ¼ 5.5‰, Nd ¼ 33 ppm, 1Nd ¼ 23.7, SiO2 ¼ 47%, FeOT ¼ 14% (Leeman, unpublished data). Simple mixing models between these end member compositions (Fig. 5) provide a gauge for estimating crustal pedigree. The rhyolite data define broad trends that deviate from simple end member mixing. As discussed earlier, part of the skewness in the d18O plots relates to inherent variations in oxygen composition of the rhyolite sources, but the Nd data most likely reflect open system mixing processes. Low FeOT and high SiO2 as well as lower 1Nd in

243

the BJ ignimbrites (tuffs) preclude a significant basalt contribution (perhaps ,10%) in the most voluminous rhyolites, whereas compositions of the relatively small volume rhyolite lavas are shifted toward expected differentiation products (or remelts) of the basaltic end member. Although inferred mixing proportions depend strongly on assumed end member composition(s), basaltderived melt components plausibly could contribute as much as half the mass of the more iron-rich rhyolites (i.e. the late stage lavas). As seen in Figure 4, temporal increases in FeOT and 1Nd suggest that basaltic contributions increased as the BJ system matured. This observation is in accord with increased availability of basaltic magma (and its derivative liquids) for mixing with crust-derived melts over time. Considering the volumetric dominance of early ignimbrites (5900 km3 or c. 85% erupted before 10.5 Ma; Table 1) over the younger rhyolite lavas (1100 km3 or c. 15% after 10.5 Ma), it is possible to evaluate these contributions. Assuming roughly 10% (or c. 590 km3/ basalt-derived melt 2.2 Ma ¼ 268 km3/Ma) contribution during the early stage and 50%

Fig. 5. Isotope systematics in rhyolites from the Bruneau– Jarbidge (BJ) area (Bonnichsen et al. 2008). Trajectories are shown for mixing between representative SRP basalt (‘B’) and crustal-derived liquids (‘C’) as represented by the most 18O-depleted BJ rhyolite. Large circles represent BJ rhyolites: ignimbrites (shaded), lavas (white). Comparative data are shown for rhyolites from the WSRP (þ) and Owyhee Front (), and for intra- (small circles) and extra-caldera (small squares) rhyolites from Yellowstone (data from Hildreth et al. 1991; Boroughs et al. 2005). Dashed arrows schematically show expected trends for differentiation of primitive basaltic magmas and/or point toward compositions of silicic melts produced by remelting basaltic intrusive rocks.

244

W. P. LEEMAN ET AL.

(or c. 550 km3/2.5 Ma ¼ 220 km3/Ma) during the late stage suggests that absolute contribution was nearly constant over time. Thus, temporal changes in BJ rhyolite chemistry could plausibly reflect variations in crustal melt production superimposed on a steady ‘background’ input of basalt-derived melts. In comparison, data for Yellowstone and Owyhee Front rhyolites (Fig. 5) seemingly do not require large basaltic contributions. However, Hildreth et al. (1991) suggest that significantly higher 1Nd values in intracaldera v. extracaldera rhyolites from Yellowstone probably reflect greater basaltic inputs proximal to the centre of the volcanic field.

Evidence for basalt intrusion in the middle – upper crust Although basaltic volcanism is always posterior to the main episode of rhyolite volcanism, mafic magma needs to be involved in the generation of the rhyolites either as a parent or as a heat source to induce partial melting of the crust. Geophysical studies suggest the presence of a mafic body in the crust. A roughly 10 km thick, high-Vp lens has been seismically imaged in the middle crust between c. 9 and 19 km depth beneath the eastern SRP (Sparlin et al. 1982; Peng & Humphreys 1998); recent tomographic studies (Dueker et al. 2007) revise its depth downward to 15–25 km. This feature has been interpreted as a basaltic sill complex (Smith & Braile 1994; Shervais et al. 2006); there are insufficient seismic data to define its lateral extent. Phase equilibria experiments on SRP basalts (Thompson 1975) and petrographic observations support the idea that these magmas commonly evolved at such shallow depths. Specifically, SRP basalts carry phenocryst assemblages of olivine + plagioclase, consistent with crystallization at pressures less than c. 8 kbar (Leeman & Vitaliano 1976). Notably, the basalts conspicuously lack phenocrysts of clinopyroxene, whereas experimentally this mineral is found to be the primary liquidus phase in such magmas at pressures greater than 8–10 kbar, corresponding to the lower crust or uppermost mantle. Although geochemical evidence allows that many SRP basalts could have experienced ‘cryptic’ clinopyroxene crystallization at greater depths (Hildreth et al. 1991), the prevailing petrographic evidence indicates that most SRP basalts last segregated from storage reservoirs shallower than 25 + 3 km. Although this does not preclude storage of some basalt in the deeper crust, no basalt originating directly from such depths has been identified so far. Positive Bouguer gravity and magnetic intensity anomalies associated with the SRP (Mabey 1982;

Smith & Braile 1994) support the concept that relatively shallow basaltic intrusion has significantly modified parts of the underlying crust. This idea is supported by long-wavelength isostatic residual gravity and magnetic potential maps (figs 8 & 18 of Mankinen et al. 2004) that effectively illuminate positive mass and magnetic anomalies at midcrustal depths beneath much of the SRP and show that these anomalies are largely confined to the physiographic province (c. 100 km wide by 600 km long). Magmatic densification of the crust is also supported by available seismic refraction studies (Hill & Pakiser 1967; Pakiser 1991). Available data indicate a westward thickening of the higher-Vp lower crust (at the expense of the upper crustal layer), whereas overall crustal thickness remains nearly constant (c. 40 km beneath the entire province (Leeman 1982; Smith & Braile 1994). These relations are consistent with a greater relative proportion of high velocity and high density material in the crust beneath the western SRP, and we propose this to be an artefact of basaltic intrusion and extraction of voluminous rhyolite. Unfortunately, crustal structure in that region remains to be investigated in detail.

Physical models of magma generation To generate rhyolite magmas, basalt must intrude, heat, and partially melt the crust. Furthermore, we assume that the source domain must have been altered by meteoric water. The infiltration of nearsurface fluids is probably limited by decreasing crustal permeability with depth (Ingebritsen & Manning 1999). However, as we will see, it is difficult to sustain elevated temperatures (.900 8C) equivalent to the hottest rhyolites at shallow crustal depths. Thus, thermal and oxygen isotopic data together provide potentially powerful constraints on the possible depths of rhyolite magma generation. How these two factors might realistically be reconciled is addressed below.

A conceptual magma reservoir model The observed transition from rhyolitic to basaltic volcanism in the SRPY province may be explained in terms of the role of the underlying continental crust as a low-density barrier that effectively traps ascending basaltic magmas during early stages of magmatism. Although Kavanagh et al. (2006) showed that density is not the only parameter that controls formation of sills into an elastic medium, buoyancy clearly is a driving force for the ascent of magmas and, on a crustal scale, must influence the final level where magmas stall and crystallize. A buoyancy model is considered here in which

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

representative densities are assumed for upper (2.6 kg dm23) and lower (2.9 kg dm23) crust, mantle (3.2 kg dm23), basaltic magma (2.7 kg dm23) and equivalent gabbro (2.9 kg dm23). In the simplest view, basaltic magma detached from a mantle source would be expected to ascend buoyantly into the crust and form sill-like lenses at depths where neutral buoyancy is favoured (cf. Ryan 1987), such as near the boundary between upper and lower crust. However, because cooling and solidification will produce denser gabbroic rock, subsequent batches of basaltic magma would tend to be emplaced above the growing sill complex (Fig. 6a). Not only will this process increase the overall density of the crust, but the proportion of crust having lower-crust density also will increase at the expense of the original upper crust domain. Although depths of magma stagnation (Z) in the crust could differ in detail from the above scenario for a variety of reasons, the proposed intrusive scenario provides a useful conceptual framework for thermal modelling presented below.

245

Based on the petrologic and experimental constraints summarized earlier, it is clear that most erupted SRPY basalts last evolved at depths shallower than 25 + 3 km. Consideration of rhyolite generation (discussed below) and revised position of the seismically imaged sill complex both point to basaltic intrusion as shallow as 10–15 km depth. Given their density contrast relative to upper crustal rocks, basaltic magmas trapped at such midcrustal depths would have little buoyancy incentive to rise much higher or to erupt, and would ultimately crystallize and heat the surrounding wall rocks. As an interesting corollary, the onset of basaltic eruptions some 2 Ma after initial magmatism is consistent with time-progressive densification of normal continental crust. In its simplest form, the problem concerns how to sustain a magma column height (H ) such that it exceeds the depth of the magma reservoir (Z) below the surface (Fig. 6b). H is computed simply as a function of magma density and lithostatic pressure at depth Z. It can be shown (Fig. 6c) that positive (H–Z) (i.e. eruption) is favoured by increasing reservoir depth or thinning

Fig. 6. Schematic diagram relating magma storage depths and ascent scenarios. (a) Overaccretion concept: basaltic magma ascends from the mantle and is emplaced as successive sills in the crust; density considerations favour emplacement of new magma above earlier, now denser intrusions emplaced in order of numbers. (b) Schematic terminology: height (H ) of a magma column originating from a reservoir at depth (Z) will vary depending on the density structure of the overlying crustal rocks; positive H– Z ¼ eruption. (c) Model: H– Z relations for basaltic magma as functions of reservoir depth and thickness of overlying rocks with upper crustal density. With sustained mafic intrusion, the thickness of effective upper crust density layer will decrease with time, causing the relation between relative column height and reservoir depth to shift upward (indicated by vertical arrow). The dashed rectangle outlines maximum depth range for erupted SRPY basalts that is compatible with the pressure stability of observed olivine þ plagioclase phenocryst assemblage. The absence of clinopyroxene phenocrysts limits Z values to shallower than 25 + 3 km (see text for details).

246

W. P. LEEMAN ET AL.

of the effective upper crust density domain (hereafter, simply called ‘upper crust’). However, if SRPY basalts are constrained to rise from reservoirs shallower than 25 km, they can only reach the surface if the effective density of the upper crust is increased sufficiently to overcompensate the mass of the magma column. This implies that some basalt is intruded within the shallow upper crust and/or mass is redistributed internally within the upper crust by some other means. An important process to increase upper crustal density could be extraction and upward migration of silicic magmas. Also, as discussed by Hildreth (1981), the mere presence of voluminous silicic magma bodies or partial melt zones in the upper crust (e.g. as inferred for Yellowstone today) would probably impede ascent of basaltic magma to the surface; solidification of such intercepted magmas would increase upper crustal density. While over-pressurized magma chamber conditions at depth could modify these calculations, the general relations should be valid regardless of the chamber pressure conditions. For basaltic magmas to erupt from reservoirs as shallow as 15 km (top of the geophysically defined sill complex), the low-density upper crust domain must be less than c. 10 km thick to provide sufficient buoyancy lift. This intrusion model would consistently produce the observed phenocryst assemblage in SRPY basalts as long as new magma was emplaced near the top of the sill complex (i.e. where neutral buoyancy is most probable). Unfortunately, little is known regarding the nature of the earliest basaltic magmatism, or the overall volume production of such magmas, during the c. 2 Ma interval when crustal density structure prevented their eruption. As an indirect means of constraining the scale and energetics of the entire magma system, the remainder of this paper considers the processes required to produce the large volumes of associated rhyolitic magma. It is widely accepted that the energy needed to produce the rhyolites is derived from inputs of basaltic magma into the crust (Lachenbruch et al. 1976; Younker & Vogel 1976; Huppert & Sparks 1988; Bergantz 1989; Bittner & Schmeling 1995; Petford & Gallagher 2001; Annen & Sparks 2002; Dufek & Bergantz 2005; Annen et al. 2006). Our approach is to consider the volume production, as well as critical physical and chemical properties, of SRPY silicic magmas to constrain physical models for melt generation.

Thermal models for rhyolite generation Our analysis emphasizes the CSRP rhyolites because their eruptive volumes and temperatures

are reasonably well known and their d18O-depletion provides limits on the possible depth of the rhyolite crustal protolith. Realistic physical models for rhyolite generation are constrained to satisfy the following criteria based on the geological record and on the geochemical and petrological data. The cumulative amount of CSRP rhyolite ranges from a minimum of 7000 (stratigraphic estimate) to perhaps as much as 30 000 km3 (tephra estimate coupled with an intrusive:extrusive ratio of 2). Average temperatures above 900 8C are required within the source volumes. Although speculative, it is illustrative to consider the scale of magma production by estimating source volume. Assuming a 50 km radius footprint (i.e. width of the SRP) for the source, and degree of melting near 25%, the above volumes of magma could be produced from cylindrical sources at least 3.6 km thick (these relations and the effect of varying melt fraction, F, are illustrated in Fig. 7). Estimates of source thickness may be exaggerated to the extent that an unknown fraction of the overall magma volume is derived from the large basaltic inputs required to drive crustal melting. Even with conservative volume estimates, the scale of the problem is formidable. The primary purpose of the thermal models presented here is to understand the conditions that can warm such thicknesses of crust to the range of liquidus temperatures estimated for SRPY rhyolites and, in addition, simulate the geologically constrained

Fig. 7. Diagram showing relation between total magma volume v. thickness of the corresponding source domain as a function of degree of melting or melt fraction (F ). Source volume is assumed to be cylindrical with a diameter of 100 km (i.e. the width of the SRP). Dashed boxes indicate scenarios based (a) solely on erupted volumes (I:E ¼ 0) and (b) on combined eruptive and intrusive volumes (I:E ¼ 2) for the BJ centre. For F ¼ 0.25, source thickness ranges between approximately 3.5 and 15 km to produce limiting volumes (7000– 30 000 km3) of BJ rhyolite. The thickness of the source volume decreases with increasing melt fraction, but even for F ¼ 0.5, must exceed at least 2 km.

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

magma production. We present here the results of one-dimensional heat transfer calculations assuming that melting is driven by repeated injection of hot mafic magmas into the crust. This concept conforms closely with the magma ascent processes discussed earlier. The computation of heat balance between basaltic intrusions and the surrounding crust takes into account the conductive heat transfer and the latent heat released by crystallization of the basaltic magma:

r cp

@T @X @2T þ rL ¼k 2 @t @T @x

(1)

where r is density, cp is specific heat capacity, L is latent heat released by magma crystallization, T is temperature, t is time and x is distance. The equation is solved for the basalt intrusion and country rock systems using forward finite differences. The initial condition is a geotherm of 30 8C km21. The boundary conditions are a fixed temperature of 0 8C at the Earth’s surface and the initial geotherm at depth. More details of the simulation method can be found in Annen et al. (2006). Computation of the latent heat released by crystallization requires knowledge of the relationship between basalt temperatures and melt fractions (Fig. 8). Using MELTS (Ghiorso & Sack 1995; Asimow & Ghiorso 1998), this relationship was modelled for the composition of representative SRP basalt, assumed to contain a small amount (c. 0.1 wt%) of H2O. The effect of increasing H2O content to 0.4% (typical for many SRP olivinehosted melt inclusions; C. Stefano, pers. comm.) is also shown in Figure 8; this results in a lowering of temperature by c. 50 8C at a given pressure.

Fig. 8. Relationship between melt fraction and temperature for a SRP primitive basalt (McKinney basalt sample 72-20; Leeman & Vitaliano, 1976) calculated with MELTS at pressures of 1, 3.5, 7 and 15 kbar, FMQ buffer and initial H2O near 0.1 wt%. Note that water-free solidus temperatures exceed 1050 8C at all pressures. The dashed line shows the effect of increasing H2O to 0.4% at 7 kbar.

247

A latent heat of 5.3  105 J kg21 was used for the basalt (Kojitani & Akaogi 1994). It should be noted that the small amount of water in this magma results in significant lowering of the solidus at low melt fractions compared with the equivalent dry magma (solidus T near 1050 8C). The composition and melting behaviour of the crustal protolith is not known. For this reason, instead of calculating the melting degree of the crust (which is strongly dependent on assumed bulk composition), we calculated the thickness of crust for which temperature exceeds 900 8C. With this approach, no assumption is required concerning the relationship between temperature and melting degree for the crust. However, because we neglect the latent heat absorbed by crust during partial melting, temperatures are slightly overestimated. We compared the results obtained when neglecting latent heat with results obtained by allowing a granodioritic crust to melt (using the granodiorite melttemperature relationship as in fig. 1 of Annen & Sparks 2002, granodiorite latent heat of 3.5  105 J kg21, and degree of melting up to 50% at the contact with basalt but decreasing rapidly away from the contact); the maximum error in temperature is 15 8C and the difference in thickness of crust heated above 900 8C is 150 m or less. We model the emplacement geometry of the growing mafic body by sill over-accretion; i.e. each sill is emplaced above the former one at the top of the mafic body and in contact with the upper crust. This emplacement geometry maximizes temperatures in the shallow crust as well as the degree of crustal partial melting (Annen et al. 2007). For convenience, the thickness of individual sills is assumed to be 50 m and models were run with up to 20 km of basalt injected into the crust. Annen & Sparks (2002) show that, as long as the time interval between successive injections is much shorter than the total duration of the simulation, the long-term thermal evolution of the system is independent of the exact thickness of the sills and, rather, depends primarily on the average emplacement rate; this is obtained by dividing the sill thickness by the time interval between sill emplacements. Repeated infusion of basalt into the crust causes the mafic body to grow and heat the surrounding country rocks, such that temperatures in both the crust and the basalt intrusive complex increase with time. If melt is extracted periodically from the partially molten crust, the residual crust will become more mafic and refractory with time. Also, if crustal partial melts and basalt residual melts reside in the crust for protracted time periods, increasing temperature could result in an increasing melt fraction overall (Annen et al.

248

W. P. LEEMAN ET AL.

2006). Both processes are consistent with compositional evolution of the BJ rhyolites, for which SiO2 content decreases and FeOT and TiO2 increase with time. The evolution of temperature in the crust is controlled by the balance between the heat that is advected by the basalt sills and the heat that is conducted through the crust. The emplacement rate of the basalt controls the distribution of temperature within the crust and its ability to partially melt. We have simulated the repeated injection of basalt sills with different emplacement rates and at different depths over 4 Ma, which is similar to the duration of BJ rhyolitic magmatism. The volume of crust above 900 8C steadily increases with time as long as basalt injection and growth of the composite mafic intrusive complex continue at a sufficient rate (Fig. 9). However, once basalt injection abates or ceases (e.g. due to the source moving along the hot spot track), crustal temperatures and volumes of partial melt begin to decrease. As an example of the timescale for this process, following basalt injection over 4 Ma at a rate of 4 mm year21 (i.e. 16 km of basalt added to the crust), some 1 Ma is required for hot zones in the crust to cool down below 900 8C once basalt input stops.

Fig. 9. Temporal evolution of a crustal hot zone. In this model, basalt is repeatedly emplaced at a constant depth of 15 km over 4 Ma with an influx rate of 4 mm year21 (total thickness ¼ 16 km), after which basalt influx terminates and the system begins to cool. The graph illustrates the thickness of upper crust above 900 8C as a function of time; the maximum is less than 3 km for this intrusion rate. In reality, the crust must cool off at a slower pace owing to prolonged basaltic volcanism following the initial pulse of silicic volcanism. Also, note that an incubation period of c. 1 Ma is required before a significant volume of crust reaches 900 8C.

The maximum thickness of crust heated above 900 8C increases with increasing basalt emplacement rate and with depth of the mafic body (Fig. 10a & b). If the emplacement rate is less than 2 mm year21, the crustal temperatures cannot exceed 900 8C on the timescale of 4 Ma except for emplacement depths in the lower crust. Emplacement rates greater than 5 mm year21, correspond to intrusion of more than 20 km of basalt over 4 Ma, which we consider unlikely (although the upper limit remains unconstrained). To heat more than 2 km of crust above 900 8C, emplacement of a total basalt thickness of at least 12 km is required (equivalent to a minimum basalt intrusion rate of 3 mm year21 over 4 Ma). However, the thickness of basalt that is above 900 8C is at least twice the thickness of old crust heated above 900 8C, and the proportion of basalt becomes greater with increasing emplacement rate (Fig. 10c). The proportion of crust-derived v. basalt-derived melts depends on the relative fertility of these sources. Our results indicate that only with melting degrees of 40% or more (e.g. tonalite source of Patin˜oDouce 1997) would the amount of crustal melts exceed the quantity of residual melt generated within the basalt. Despite this, the dominantly crustal geochemical signature of voluminous early BJ rhyolites suggests that, during this stage, crustalderived melts were preferentially extracted with only moderate dilution by residual basalt-derived liquids. Conversely, during the waning later stage, contributions from residual basaltic liquids increased in proportion. Our thermal model is not dynamic and the withdrawal and eruption of melt is not explicitly simulated. Although Jackson et al. (2003) showed that melt composition evolves during compaction, our knowledge of the system and of the physical parameters that control its dynamic behaviour is insufficient to warrant complex simulations that include compaction, extraction and possibly injection at shallower depths. Extraction of melt will not modify the temperature of the rock from which it was removed but transfer or reinjection of such melts elsewhere within the crustal hot zone could alter the temperature profile in the melting zone. This process is unlikely to significantly affect the thicknesses of heated crust (shown in Figs 9 & 10) as long as the volume of mobile melt involved is small compared with the volume of solid heated rocks. In our model system, where mafic magma is incrementally emplaced as sill intrusions, each successive sill cools down on timescales of a few hundreds to a few thousand years depending on its exact thickness (figs 7 and 19 of Annen et al. 2006). Following thermal equilibration with the surrounding wall rocks, temperatures of the intrusions,

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

Fig. 10. Thickness of upper crust and basalt with temperatures above 900 8C after 4 Ma of repeated basalt emplacement as functions of mafic magma emplacement rate (i.e. total thickness of emplaced basalt) and emplacement depth. Magma emplacement at depths shallower than 5 km, even at high emplacement rate, cannot bring the crustal wall rocks to 900 8C. (a) Thickness of pre-existing upper crust heated to T  900 8C. (b) Thickness of basalt plus old upper crust heated to T  900 8C. (c) Percentage of old upper crust heated above 900 8C relative to the total (upper crust þ basalt) that is above 900 8C.

249

hence also compositions of the residual melts, evolve slowly as the whole system progressively warms up. Mafic and possibly intermediate composition melts are probably preserved in late-injected sills that had insufficient time to equilibrate with the remaining crust. Because of the thermal gradient within the basalt and intruded crust, spatial diversity in melting degree and in melt composition is predicted. For anhydrous basaltic intrusions, the maximum model temperature (c. 1140 8C) corresponds to a melt fraction of less than 0.25, which is likely to have a relatively silicic composition. For more hydrous basalt (cf. Fig. 8), melting degrees would be higher at comparable temperatures and liquids of intermediate composition would be produced. The volume of melt produced can be scaled to match BJ eruptive volumes. Figure 11 shows the predicted evolution of the temperature distribution within the crust when basalt sills are emplaced at an average rate of 4 mm year21. After 4 Ma, this scenario results in heating a volume of crust (c. 2.8 km thick) that is insufficient to match the minimal BJ extrusive volume (7000 km3) if the rhyolites are pure crustal melts and F smaller than 0.33. To heat the necessary thickness of crust (3.6 km for F ¼ 0.25) requires longer time of intrusion (i.e. longer incubation time), higher basalt input rate or a combination of these factors. To produce greater (and more realistic) total melt volumes (.10 000 km3) strictly from the crust requires even more extreme adjustments of one or more of these parameters and/or, greater lateral dimension of the source volume. Within limits provided by geochemical constraints, the total volume of magma in the system could also be enhanced by liquids produced by differentiation of the intruded basalt or remelting of portions of the mafic intrusive complex. For the BJ centre basalt-derived melt contributions are inferred to increase as the system evolved and became warmer in response to progressive intrusion of basaltic magma into the shallow crust. Our modelling suggests that a composite mafic body less than 20 km thick, emplaced over intervals of 2–4 Ma, can elevate temperatures above 900 8C within only a few kilometres of proximal pre-existing crust (Fig. 10a), particularly if intrusion depths are shallower than 20 km. It is difficult to maintain such high temperatures at depths shallower than 5 –10 km owing to initially low wall rock temperatures and considerable heat losses to the surface. We also note that relatively fertile source rocks considered previously by Annen and coworkers are not well suited as protoliths in this setting because they produce a considerable amount of melt at temperatures well below those of SRPY silicic magmas, and

250

W. P. LEEMAN ET AL.

estimate is about twice that based only on a simple energy balance (c. 8–10 km over 4 Ma; Bonnichsen et al. 2008). This difference reflects the fact that the latter approach ignores substantial dissipation and refraction losses of heat associated with incremental intrusion of basalt. For a circular footprint with 100 km diameter, the time-averaged basalt flux is between 0.024 and 0.031 km3year21; i.e. about an order of magnitude smaller than the magma flux for Kilauea volcano (c. 0.2 km3year21; Crisp 1984).

Oxygen isotopic constraints on rhyolite genesis The oxygen isotope problem

Fig. 11. Profiles of temperature v. depth in the crust after 2 and 4 Ma of basalt emplacement at a sustained rate of 4 mm year21 In this model, successive basalt sills are emplaced at 15 km depth. Over this time interval, the thickness of a crustal hot zone (T . 900 8C) varies from about 4.5 to 14 km. Note: a and b are thicknesses of upper crust and intruded basalt heated above 900 8C after 2 Ma; c and d are thicknesses of upper crust and intruded basalt heated above 900 8C after 4 Ma.

there is no surface manifestation of such magmatism. Assuming that depths of pervasive 18 O-depletion are probably restricted to the uppermost portions of the crust, and that such modified rocks are protoliths for the BJ rhyolites, it appears inescapable that sizable mafic intrusions must be concentrated in the shallow crust (probably within 15 km of the surface) to promote extensive melting of these distinctive domains. For such conditions our models for the BJ magmatic system require a larger addition (c. 16 km thickness over 4 Ma) of basaltic magma to the crust. This new

Because of its importance in constraining melting depths in the crust, we now return to the issue of how 18O-depletion in the BJ rhyolite source could be related to infiltration of waters controlled by near surface processes preceding and possibly attending magmatism in the region. It is well known that hydrothermal circulation can lead to chemical and isotopic alteration of aureoles surrounding epizonal plutons, but it is not clear how deep such effects can extend or to what extent O isotopic compositions of crustal rocks can be shifted as a function of depth. It is clear that modest 18O-depletion extends to magma chamber depths in many places. Bindeman et al. (2004) present extensive data for Kamchatka suggesting that 18O-depletion is related to deep circulation of meteoric waters, and suggest that magnitude of the 18O-shift depends on latitude and climate influences on water compositions. Moreover, these authors show that 18O-depletion is more pronounced in caldera-related silicic magmas than in smaller stratovolcanoes. The difference could be related to tectonic and structural controls that influence fluid circulation, leading to heterogeneous crustal modification. These are critical issues in understanding origins of SRPY silicic magmas. Given the outcomes of our thermal models, it seems virtually impossible to get crustal temperatures above 900 8C at depths shallower than 5–10 km for volumes of crust approaching even the minimum estimate for the rhyolite source. This implies that fluid infiltration and significant O-exchange must reach greater depths to explain the chemistry of BJ rhyolites. If this condition was established in the crust prior to the earliest BJ eruptions, the modification could have occurred over a relatively long time period; it could also have been a partly syn-magmatic process. We consider the ‘open ended’ questions: given unlimited time, and that Miocene precipitation in the

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

region had d18O near – 20‰ (cf. Hearn et al. 1989; Horton et al. 2004), what is the maximum lowering of d18O that can be achieved in the crust and how deep could the modification extend? Specifically, can this process realistically create large volumes of crustal rocks (e.g. a 100 km diameter cylindrical disk at least 3.6 km thick) with d18O in the range observed in BJ rhyolites, and at appropriate depths where thermal conditions are suitable for generation of silicic magmas in volumes matching the estimated magmatic outputs? And, assuming that the underlying crust started with normal d18O (c. 10‰), what processes and/or conditions are required to modify the magma source to values as low as 22‰?

Theoretical considerations The rate of 18O-depletion of the crust is controlled by both the supply of low d18O meteoric waters at depth and by the kinetics of either oxygen exchange between the water and the permeable crystal framework or growth of new phases. The kinetics of oxygen exchange is highly dependent on the temperature of the reaction. A detailed calculation of the kinetics of the oxygen exchange process in porous media will be presented elsewhere, but insight can be gained from the observation that oxygen diffusivity drops four orders of magnitude between 800 and 400 8C in plagioclase (from c. 1  10216 to 1  10220 m2 s21) and is more than 10 orders of magnitude less than the equivalent thermal diffusivities (Cole & Ohmoto 1986). However, provided the crust is sufficiently permeable, advection of hydrothermal waters allows diffusion to operate at the grain-scale. The thermal gradient in the crust can be combined with a grain-diffusion model and appropriate temperature dependent oxygen diffusivities and fractionation factors to estimate the maximum possible d18O depletion in the crust. We also consider an end member in which separate phases have reached equilibrium with the circulating hydrothermal waters; this end member can also be thought of as a proxy for new phases that may grow in the presence of the hydrothermal waters. In these calculations we implicitly assume that the fluid d18O is constant and equivalent to 213‰, a median value for d18O measured in geothermal waters in southwestern Idaho (Rightmire et al. 1976). We use a d18O higher than the meteoric value (c. 220‰) because, at depth, the composition of water percolating through regions of low flux likely will be altered by oxygen exchange with crustal rocks. Values of d18O even greater than 213‰ are likely at depth as the measured surface geothermal fluids probably contain a component of near surface, unaltered water (Allan & Yardley 2007).

251

Hence the following calculations should be viewed as indicating maximum depletions. In delta notation, an adapted analytical expression for diffusion in a spherical grain is given as (Carslaw & Jaeger 1959):

d18 Or (r; t)  d18 O0r ¼1 d18 OBr  d18 O0r  2 2  1 2a X 1n pnr p n Dt sin þ : exp pr n¼1 n a2 a

(2)

Here and below, a denotes the crystal radius, r is the radial distance, D is the diffusivity, d18O0r is the initial rock composition (taken as 10‰), d18O0w is the water composition (taken as 213 ‰), d18Or(r, t) is the time and radially dependent d18O in the mineral, t is time, n is the dummy variable for the summation calculation, and d18OBr is the mineral equilibrium composition at the grain boundary. The grain boundary condition here is given as a function of the water composition, the equilibrium fractionation factor, and the mass and composition of water and rock.

d18 OBr ¼ Dwr þ d18 O0w þ d18 O0w þ Dwr þ d18 O0r   cW mW  exp : c R mR


(3)

Here mW and mR are the mass of water and rock, respectively, and cW and cR are the elemental concentrations of oxygen in water and rock, and Dw2r (T ) is the temperature dependent equilibrium fractionation factor. Note that if the flux of water is high (the mass of water to rock approaches infinity), the grain boundary composition approaches the equilibrium value of:

d18 OBr ¼ d18 Oiw  Dwr:

(4)

If the flux of water is limited at the grain boundary due the permeability of the matrix, the boundary will be modified by the exponential term in Equation 3. Ingebritsen & Manning (1999) suggest a permeability profile based on geothermal and fluid flux data: z

k ¼ 10143:2 log

(5)

where z is in kilometres. Near the brittle – ductile transition, fluid pressures may become greater than hydrostatic pressure, inhibiting the deeper circulation of water at depth (Huenges et al. 1997; Brown 2007). Although this transition probably

252

W. P. LEEMAN ET AL.

Table 2. Diffusivity and fractionation factor coefficients Mineral Anorthite Albite K-Feldspar Quartz Hornblende Biotite

A

B

C

A0 (m2 s21)

Ea (J mol21)

4.12 4.33 4.32 4.48 3.89 3.84

27.5 26.15 26.27 24.77 28.56 28.76

2.24 1.98 2.0 1.71 2.43 2.46

1.39  10211 9.8  10210 3.95  10212 2  1029 1  10211 9.1  10210

109,600 139,800 109,700 184,000 171,600 142,300

Equilibrium fractionation factor coefficients (used in Equation 7) are from Zheng (1993a, b). The diffusion coefficients for the specific phases (used in Equation 8) are from experiments (Giletti et al. 1978; Yund et al. 1981; Giletti & Yund, 1984; Farver & Giletti, 1985; Cole & Ohmoto, 1986; Fortier & Giletti, 1991; Freer et al. 1997; Cole & Chakraborty, 2001).

occurs over a range of temperatures depending on the lithology, we use 400 8C as the transition where the crystalline rock becomes ductile over long timescales (Manning & Ingebritsen 1999). The mass of water that passes through a unit volume of rock is then: mW ¼ q  t ¼ 

10143:2 log z @[Drwater gz]  t: (6) mwater @z

Here q is the water flux, mwater is the water viscosity, r is density, and g is the gravitational acceleration. The equilibrium fractionation factors (in D notation where Dw2r ¼ 103 ln a, and a is the equilibrium fractionation water-rock ratio) are given as D¼

A  106 B  103 þC þ T2 T

(7)

where A, B, C for the specific mineral phases are taken from Zheng (1993a, b; cf. Table 2). Oxygen diffusivities for specific phases are calculated from an Arrhenius expression, D ¼ A0 exp



Ea RT



(8)

where the pre-exponential (A0) and activation energy Ea terms are calibrated from appropriate experiments (cf. Table 2). Solving Equation 2 for d18Or(r, t), performing the spherical integration and dividing by the volume gives the mean oxygen isotopic compo¯ r(t), of crystals that we take as a proxy sition, d18O for the crust at a particular time and depth. Assuming a geothermal gradient of 25 8C km21, 2.5 mm crystal radii, and using Equations 2–8 gives depth-composition profiles for feldspars, quartz, hornblende and biotite after 1 Ma of interaction (Fig. 12). Profiles are shown for two extreme scenarios assuming either ‘infinite’ access of fluid (i.e.

very large water –rock ratios), or a more realistic water-flux limited case wherein fluid flow is governed by Darcy’s law, the permeability given in Equation 5, and a lithostatic pressure gradient. In the second case, if permeability was substantially enhanced, the profiles in Figure 12c would approach but never exceed those in Figure 12b. We again note that these profiles represent maximum depletions as we assume a constant fluid composition. A second scenario is considered in which a thermal anomaly from prior intrusions has warmed the mid-upper crust (Fig. 13a). Similar oxygen depletion calculations are performed for this thermal condition. The maximum depletion profiles (Fig. 13b) have proceeded to equilibrium with the circulating water. This end member can also be thought of as a proxy for recrystallizing phases in the presence of the hydrothermal fluid. The minimum depletion calculation is constrained by the amount of water that passes through the permeable rock (Fig. 13c). Although degrees of 18 O-depletion are similar for both scenarios, the maximum depletion is shifted to shallower depths and the thickness of the zone of depletion is also diminished for warmer crust. The assumption of a constant boundary condition of meteoric water d18O breaks down where permeability significantly reduces the fluid flux. Although temperature increases with depth in the crust, the reduction in mean permeability limits the degree of oxygen exchange (Manning 1981). In general, d18O depletion in the upper crust is kinetically (thermally) limited, and d18O depletion in the deep crust is water flux (permeability) limited. Finally, we note that isotopic shifts are more pronounced for feldspars and biotite than for quartz or hornblende. Thus, 18O-depletion for the bulk crust will depend on lithology and mineral proportions. The oxygen isotopic composition of crustal melts could then differ depending on specific lithologies and oxygen exchange histories of their source rocks.

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

253

Fig. 12. Calculated effects of infiltrating surface waters on oxygen isotopic composition of crustal minerals as a function of depth. Models include temperature-dependent diffusion kinetics and isotopic fractionation effects and assume input of constant composition of meteoric water; see text for details. Two scenarios are presented, both for 1 Ma time interval: (a) shows the thermal profile used in this calculation; (b) shows the effect of having an infinite supply of surface fluid, but with fluid circulation inhibited at the transition to ductile crust; (c) incorporates permeability-limited fluid infiltration using the permeability – depth relation of Ingebritsen & Manning (1999), and is considered more realistic for diffusional exchange. In both cases, kinetic effects associated with lower temperatures limit isotopic exchange at shallow crustal depths. These models, and (c) in particular, suggest that 18O-depletion varies with depth and may display a maximum effect in a specific depth range – in this case, near 15 km.

Results The first order result of these models is that 18 O-depletion in the shallow to mid crust appears to be of sufficient magnitude to match that required

for the BJ rhyolite source volume (i.e. d18O as low as 22‰). Maximum effects are localized in the depth range between 10 and 15 km depending on crustal lithology. Our models for thermally perturbed syn-magmatic crust essentially limit

Fig. 13. Calculated effects of infiltrating surface waters on oxygen isotopic composition of crustal minerals as a function of depth, as in previous figure but with a syn-volcanic scenario in which the geothermal gradient is perturbed by repeated magma injections. Again, two scenarios are presented, both for 1 Ma time interval: (a) shows the perturbed thermal profile used; (b) shows effect of having an infinite supply of water, and profiles indicate maximum d18O-depletion for diffusion exchange or recrystallization; (c) incorporates permeability-limited fluid infiltration. In both cases, we assume that fluid pressures exceed hydrostatic pressure and inhibit circulation of meteoric waters at depths below the brittle – ductile transition (i.e. at temperatures above 400 8C).

254

W. P. LEEMAN ET AL.

18

O-depletion to depths shallower than 10 km where it is extremely difficult to heat significant volumes of crust to 900 8C or more. This result is consistent with the notion that crustal modification largely predated onset of CSRP magmatism. The fact that early rhyolites from the Heise and Yellowstone volcanic fields have ‘normal’ d18O implies that meteoric fluids did not significantly modify their source rocks.

Discussion and summary This work provides new insights into the origin of silicic magmas associated with the Yellowstone melting anomaly. Observations of CSRP rhyolites provide critical constraints on the melting processes by which they formed. Foremost are the large magma volumes, the consistently elevated magmatic temperatures, the long duration of silicic magmatism, and the bimodal basalt – rhyolite productivity (with sparse intermediate composition magmas). Also, the transition from dominantly silicic to dominantly basaltic volcanism over time provides insight regarding the influence of the continental crust on magmatic processes (and vice versa). Our interpretation of how SRPY magmatism affected the underlying crust is summarized in Figure 14. We view CSRP silicic magmatism as being broadly representative of that for the entire province. However, owing to diachronous development, the Yellowstone volcanic centre presently is at a relatively early evolutionary stage compared with older centres to the west that have essentially run their course in terms of silicic volcanism. Furthermore, lateral or vertical heterogeneities in the underlying crust may impose important differences in magmatic response across the province. For example, 18O-depletion in crustal sources appears to be more extensive beneath the CSRP, and suggests that fluid-related alteration of the crust was more extensive in that area. The unusual temporal variation in composition of BJ rhyolites, that is both highly systematic yet opposite from normal magmatic differentiation trends, also implies that melting processes varied in detail across the province. Key findings from this study are highlighted below. Combined constraints of thermal modelling and oxygen isotopic evidence for an 18O-depleted source strongly limit depths of magma formation in the CSRP. The latter observation provides compelling evidence that these rhyolites were predominantly crustal melts because (a) low-d18O is unlikely to be inherited from the mantle and (b) oxygen comprises more than 50% of the magma by weight. The strongly bimodal (basalt – rhyolite)

Fig. 14. Schematic diagram for crust beneath the CSRP. Major injections of basalt occur at depths as shallow as c. 15 km by an overaccretion process that results in downward displacement of the intrusive sill complex. Rhyolite generation occurs largely in a crustal partial melt zone (PMZ) above the sill complex and may be augmented by incorporation of remobilized evolved melts derived from the basaltic sills (via normal differentiation and/or remelting of intrusive equivalents). Prior to eruption, rhyolitic magmas may coalesce in a plexus of shallow reservoir lenses as inferred for Yellowstone. The crustal section must deform to accommodate the volume of basaltic intrusions, either by extension or some form of crustal flow. It is suggested that attenuation of the crust is approximately balanced by new basaltic inputs so as to maintain near-constant crustal thickness. We assume that such a configuration extends beneath most of the SRPY province, and that peak basaltic intrusion into the crust migrated northeastward with time as reflected by migration of silicic volcanic centres in Figure 1. However, continued basaltic inputs sustained the thermal anomaly and locally led to renewed silicic magmatism (McCurry et al. 2008). Basaltic inputs to the deep crust remain uncertain, but no erupted lavas are thought to originate from such depths.

compositions of erupted magmas, and the restricted compositional range for the rhyolites, suggest that direct mixing between basaltic magma and silicic crustal melts is limited (,c. 10%, based on mass balance considerations). Relatively radiogenic Nd compositions of BJ rhyolites preclude wholesale melting of old crust, and are consistent with contributions of evolved liquids of basalt-derivation (e.g. by differentiation or remelting) to direct crustal melts; such contributions appear to be

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES

more substantial (perhaps up to 50%) in the younger rhyolite lavas than in the early ignimbrites (,20%). Owing to decreasing permeability with depth, it is difficult to produce an 18O-depleted crustal volume capable of generating the estimated volume of CSRP rhyolites. Assuming optimal conditions, theoretical simulations of downward circulation of surficial fluids suggest that maximum 18 O-depletion is likely to be concentrated at depths near 15 km for a normal crustal geotherm (c. 25 8C km21), but as shallow as 8–10 km for warmer, syn-magmatic geotherms. The latter scenario is least favoured owing to (a) the difficulty in reaching crustal melting conditions at such shallow depths, and (b) the fact that 18O-depletion must have largely predated the earliest (12.7 Ma) BJ silicic volcanism. Causes for the crustal modification remain uncertain. In addition to processes suggested by Boroughs et al. (2005), access of sufficient fluid to produce the required volume of low-d18O crust may require enhanced fluid permeability due to crustal deformation. Also, the inferred source area for the most 18O-depleted BJ rhyolites roughly coincides with the presence of a large lacustrine environment (Lake Idaho) in topographically low parts of the west-central SRP during Neogene time (fig. 1 of Beranek et al. 2006); thus, proximity of lake waters might have enhanced fluid infiltration and 18O-depletion of the underlying crust. Control by such factors as well as heterogeneous crustal permeability could possibly account for variable 18O-depletion in rhyolite magmas from the CSRP. Our thermal models indicate that a large volume of basalt (equivalent to a thickness as great as 16 km) must be injected incrementally into the crust over a period of several Ma to raise temperatures to those of the rhyolite magmas (900 8C) within a thickness of crust large enough to generate the estimated volume of CSRP rhyolitic magmas. It is increasingly difficult to achieve this condition as depth of basaltic intrusion decreases, and virtually impossible at depths shallower than 5– 10 km. Because these results are predicated on minimal to modest estimates of source volume thickness, the amounts of intruded basalt could be larger. In summary, the ‘sweet spot’ for magma generation that ideally satisfies both the thermal and oxygen isotope constraints appears to centre near a depth of 15 km. Partial melting of crustal rocks near this depth can satisfy experimental constraints that A-type magmas probably form at low pressures (c. 4 kbar). Intrusion and storage of basaltic magmas at this depth can also explain the persistent occurrence of a low-pressure phenocryst assemblage and absence of clinopyroxene in erupted SRPY basalts. Emplacement of large volumes of

255

basalt into the crust inevitably leads to a preponderance of this material relative to the thickness of preexisting crust that will remain for long times at temperatures above 900 8C. Given that crustal thickness is nearly uniform (c. 40 km) beneath the entire province, it appears that added volumes of basalt must somehow be accommodated by extensional deformation of the crust and lithosphere (Rodgers et al. 2002; Wood & Clemens 2002). Bonnichsen et al. (2008) present a simple two-dimensional model using geologically constrained extension rates (2–3% Ma; Rodgers et al. 2002) wherein predicted attenuation and thickening by basalt intrusion combine to maintain near-constant crustal thickness. The higher flux estimates from our study exacerbate the ‘room problem’ and seemingly require more dramatic extension along the province. A time-averaged extension rate near 5%/Ma could accommodate the upwardly revised basalt influx while still preserving a constant crustal thickness, but at present there is little quantitative evidence for this amount of extension. Thus, we believe that true basaltic flux for the SRPY province is unlikely to be larger than the maximum considered in this study, i.e. about one-tenth that estimated for Kilauea volcano. Further work is required to better understand the history of crustal deformation, its relation to regional Basin and Range extension, and how the predicted large volumes of basalt can be accommodated within the crust. Crustal deformation may not have been uniform in space or time, and could have been more extensive in the CSRP than in the eastern part of the province. Although there is evidence for development of a deformation parabola ahead of the migrating Yellowstone hotspot (cf. Pierce & Morgan 1991), there is only limited evidence for syn-volcanic extensional deformation of the SRP and its margins (cf. Bonnichsen et al. 2008). Finally, based on the respective eruptive volumes of rhyolite, apparent basalt flux for the BJ volcanic centre is seemingly higher than that for the Yellowstone area. In both cases, the inferred basalt flux is significantly lower than that estimated for Hawaii – thus, raising doubt as to whether SRPY magmatism is dominantly hotspot-driven. Given the evidence for concurrent lithospheric extension, it is possible that widespread basaltic magmatism associated with the province could be related in part to extensional tectonism. Better understanding of lateral variations in the basalt flux is needed to evaluate underlying causes for SRPY magmatism. Leeman acknowledges support from the National Science Foundation in the form of research grants and for the time spent on this research since moving to NSF. Dufek acknowledges support from the Miller Institute,

256

W. P. LEEMAN ET AL.

University of California, Berkeley. We also thank D. Geist, J. L. Vigneresse, and M. Jackson for helpful reviews and G. Zellmer for editorial suggestions that greatly improved our presentation.

References A LLAN , M. N. & Y ARDLEY , B. W. D. 2007. Tracking meteoric infiltration into a magmatic – hydrothermal system: a cathodoluminescence, oxygen isotope and trace element study of quartz from Mt. Leyshon, Australia, Chemical Geology, 240, 343–360. A NNEN , C. & S PARKS , R. S. J. 2002. Effects of repetitive emplacement of basaltic intrusions on thermal evolution and melt generation in the crust. Earth and Planetary Science Letters, 203, 937– 955. A NNEN , C., B LUNDY , J. D. & S PARKS , R. S. J. 2006. The genesis of calcalkaline intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology, 47, 505–539, DOI: 10.1093/petrology/egi084. A NNEN , C., B LUNDY , J. & S PARKS , R. S. J. 2008. The sources of granitic melt in deep hot zones. Transactions of the Royal Society of Edinburgh: Earth Sciences, 97, In Press. A RMSTRONG , R. L., L EEMAN , W. P. & M ALDE , H. E. 1975. K –Ar dating of Quaternary and Neogene volcanic rocks of the Snake River Plain, Idaho. American Journal of Science, 275, 225– 251. A SIMOW , P. D. & G HIORSO , M. S. 1998. Algorithmic modifications extending MELTS to calculate subsolidus phase relations. American Mineralogist, 83, 1127–1132. B ALSLEY , S. D. & G REGORY , R. J. 1998. Low-d18O silicic magmas: Why are they so rare? Earth and Planetary Science Letters, 162, 123– 136. B ERANEK , L. B., L INK , P. K. & F ANNING , C. M. 2006. Miocene to Holocene landscape evolution of the western Snake River Plain region, Idaho: using the SHRIMP detrital zircon provenance record to track eastward migration of the Yellowstone hotspot. Geological Society of America Bulletin, 118, 1027– 1050. B ERGANTZ , G. W. 1989. Underplating and partial melting: implications for melt generation and extraction. Science, 245, 1093– 1095. B INDEMAN , I. N. & V ALLEY , J. W. 2001a. Low-d18O rhyolites from Yellowstone: magmatic evolution based on analyses of zircons and individual phenocrysts. Journal of Petrology, 42, 1491–1517. B INDEMAN , I. N. & V ALLEY , J. W. 2001b. Post-caldera volcanism: In situ measurement of U-Pb age and oxygen isotope ratio in Pleistocene zircons from Yellowstone caldera. Earth and Planetary Science Letters, 189, 197 –206. B INDEMAN , I. N., P ONOMAREVA , V. V., B AILEY , J. C. & V ALLEY , J. W. 2004. Volcanic arc of Kamchatka: A province with high-d18O magma sources and large-scale 18O/16O depletion of the upper crust. Geochimica et Cosmochimica Acta, 68, 841–865. B INDEMAN , I. N., W ATTS , K. E., S CHMITT , A. K., M ORGAN , L. A. & S HANKS , P. W. C. 2007. Voluminous low d18O magmas in the late Miocene Heise volcanic field, Idaho: implications for the fate of Yellowstone hotspot calderas. Geology, 35, 1019–1022, DOI: 10:1130/G24141A.1.

B ITTNER , D. & S CHMELING , H. 1995. Numerical modeling of melting processes and induced diapirisms in the lower crust. Geophysical Journal International, 123, 59–70. B ONNICHSEN , B., L EEMAN , W. P., H ONJO , N., M C I NTOSH , W. C. & G ODCHAUX , M. M. 2008. Miocene silicic volcanism in southwestern Idaho: Geochronology, geochemistry, and evolution of the central Snake River Plain. Bulletin of Volcanology, 70, 315–342, DOI: 10.1007/s00445-007-0141-6. B OROUGHS , S., W OLFF , J., B ONNICHSEN , B., G ODCHAUX , M. & L ARSON , P. 2005. Large volume, low-d18O rhyolites of the central Snake River Plain, Idaho, USA. Geology, 33, 821– 824. B ROWN , M. 2007. Crustal melting and melt extraction, ascent and emplacement in orogens: mechanisms and consequences. Journal of the Geological Society of London, 164, 709– 730. C ARSLAW , H. S. & J AEGER , J. C. 1959. Conduction of Heat in Solids. Clarendon Press, Oxford. C ATHEY , H. E. & N ASH , B. P. 2004. The Cougar Point Tuff: Implications for thermochemical zonation and longevity of high-temperature, large-volume silicic magmas of the Miocene Yellowstone hotspot. Journal of Petrology, 45, 27–58. C ATHEY , H. E., N ASH , B. P., V ALLEY , J. W., K ITA , N., U SHIKUBO , T. & S PICUZZA , M. 2007. Pervasive and persistent large-volume, low delta 18O silicic magma generation at the Yellowstone hotspot, 12.7– 10.5 Ma: Ion microprobe analyses of zircon in the Cougar Point Tuff. Transactions of American Geophysical Union, 88, V51C-0708. C HRISTIANSEN , E. H. & M C C URRY , M. 2008. Contrasting origins of Cenozoic silicic volcanic rocks from the western Cordillera of the United States. Bulletin of Volcanology, 70, 251– 267, DOI: 10.1007/s00445007-0138-1. C HRISTIANSEN , R. L. 2001. The Quaternary and Pliocene Yellowstone Plateau Volcanic Field of Wyoming, Idaho, and Montana. US Geological Survey Professional Papers, 729-G, 1– 145. C OLE , D. R. & C HAKRABORTY , S. 2001. Rates and mechanisms of isotopic exchange. Reviews in Mineralogy and Geochemistry, 43, 83–223. C OLE , D. R. & O HMOTO , H. 1986. Kinetics of isotopic exchange at elevated temperatures and pressures. Reviews in Mineralogy and Geochemistry, 16, 41–90. C RISP , J. A. 1984. Rates of magma emplacements and volcanic output. Journal of Volcanology and Geothermal Research, 20, 177– 211. C RISS , R. E. & F LECK , R. 1987. Petrogenesis, Geochronology, and Hydrothermal Systems of the Northern Idaho Batholith and Adjacent Areas Based on 18O/16O, D/H, 87 Sr/86Sr, K–Ar, and 40Ar– 39Ar studies. US Geological Survey Professional Papers, 1436, 95–137. D OE , B. R., L EEMAN , W. P., C HRISTIANSEN , R. L. & H EDGE , C. E. 1982. Lead and strontium isotopes and related trace elements as genetic tracers in the Upper Cenozoic rhyolite-basalt association of the Yellowstone Plateau volcanic field. Journal of Geophysical Research, 87, 4785– 4806. D UEKER , K., S TACHNIK , J., Y UAN , H. & S CHUTT , D. 2007. Image of Yellowstone magmatic history and

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES lower crustal outflow. Abstracts with Programs, Geological Society of America, 39, 202. D UFEK , J. & B ERGANTZ , G. W. 2005. Lower crustal magma genesis and preservation: a stochastic framework for the evaluation of basalt-crust interaction. Journal of Petrology, 46, 2167– 2195. E ATON , G. P., C HRISTIANSEN , R. L., I YER , H. M., M ABEY , D. R., B LANK , H. R., Z IETZ , I. & G ETTINGS , M. E. 1975. Magma beneath Yellowstone National Park. Science, 188, 787– 796. F ARVER , J. R. & G ILETTI , B. J. 1985. Oxygen diffusion in amphiboles. Geochimica et Cosmochimica Acta, 49, 1403–1411. F ORTIER , S. M. & G ILETTI , B. J. 1991. Volume of selfdiffusion of oxygen in biotite, muscovite and phlogopite micas. Geochimica et Cosmochimica Acta, 55, 1319–1330. F REER , R., W RIGHT , K., K ROLL , H. & G OTTLICHER , J. 1997. Oxygen diffusion in sanidine feldspar and a critical appraisal of oxygen isotope-mass-effect measurements in non-cubic materials. Philosophical Magazine, A75, 485–503. G HIORSO , M. S. & S ACK , R. O. 1995. Chemical mass-transfer in magmatic processes. 4. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid– solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology, 119, 197– 212. G ILETTI , B. J. & Y UND , R. A. 1984. Oxygen diffusion in quartz. Journal of Geophysical Research, 89, 4039–4046. G ILETTI , B. J., S EMET , M. P. & Y UND , R. A. 1978. Studies in diffusion – III. Oxygen in feldspars: an ion microprobe determination. Geochimica et Cosmochimica Acta, 42, 45–57. G RIPP , A. E. & G ORDON , R. G. 2002. Young tracks of hotspots and current plate velocities. Geophysical Journal International, 150, 321– 361. H EARN , P. P., J R , S TEINKAMPF , W. C., H ORTON , D. G., S OLOMON , G. C., W HITE , L. D. & E VANS , J. R. 1989. Oxygen-isotope composition of ground water and secondary minerals in Columbia Plateau basalts: implications for the paleohydrology of the Pasco Basin. Geology, 17, 606– 610. H ILDRETH , W. 1981. Gradients in silicic magma chambers: implications for lithospheric magmatism. Journal of Geophysical Research, 86, 10,153– 10,192. H ILDRETH , W., C HRISTIANSEN , R. L. & O’N EIL , J. R. 1984. Catastrophic isotopic modification of rhyolitic magma at times of caldera subsidence, Yellowstone Plateau volcanic field. Journal of Geophysical Research, 89, 8339– 8369. H ILDRETH , W., H ALLIDAY , A. N. & C HRISTIANSEN , R. L. 1991. Isotopic and chemical evidence concerning the genesis and contamination of basaltic and rhyolitic magma beneath Yellowstone Plateau volcanic field. Journal of Petrology, 32, 63–138. H ILL , D. P. & P AKISER , L. C. 1967. Seismic-refraction study of crustal structure between the Nevada Test Site and Boise, Idaho. Geological Society of America Bulletin, 78, 685–704. H ONJO , N., B ONNICHSEN , B., L EEMAN , W. P. & S TORMER , J. C., J R . 1992. High-temperature rhyolites

257

from the central and western Snake River Plain. Bulletin of Volcanology, 54, 220– 237. H ORTON , T. W., S JOSTROM , D. J., A BRUZZESE , M. J., P OAGE , M. A., W ALDBAUER , J. R., H REN , M., W OODEN , J. & C HAMBERLAIN , C. P. 2004. Spatial and temporal variation of Cenozoic surface elevation in the Great Basin and Sierra Nevada. American Journal of Science, 304, 862– 888. H UENGES , E., E RZINGER , J., K UCK , J., E NGESER , B. & K ESSELS , W. 1997. The permeable crust: geohydraulic properties down to 9101 m depth. Journal of Geophysical Research, 102, 18,255–18,265. H UGHES , S. S. & M C C URRY , M. 2002. Bulk major and trace element evidence for a time – space evolution of Snake River Plain rhyolites, Idaho. In: B ONNICHSEN , B., M C C URRY , M. & W HITE , C. M. (eds) Tectonic and Magmatic Evolution of the Snake River Plain volcanic province. Idaho Geological Survey Bulletin, 30, 161– 176. H UPPERT , H. E. & S PARKS , R. S. J. 1988. The generation of granitic magmas by intrusion of basalt into continental crust. Journal of Petrology, 29, 599–624. I NGEBRITSEN , S. E. & M ANNING , C. E. 1999. Geological implications of a permeability – depth curve for the continental crust. Geology, 27, 1107– 1110. J ACKSON , M. D., C HEADLE , M. J. & A THERTON , M. P. 2003. Quantitative modeling of granitic melt generation and segregation in the continental crust. Journal of Geophysical Research, 108, 2332, DOI: 10.1029/2001JB001050. K AVANAGH , J. L., M ENAND , T. & S PARKS , R. S. J. 2006. An experimental investigation of sill formation and propagation in layered elastic media. Earth and Planetary Science Letters, 245, 799– 813. K OJITANI , H. & A KAOJI , M. 1994. Calorimetric study of olivine solid solution in the system Mg2SiO4 – Fe2SiO4. Physics and Chemistry of Minerals, 20, 536– 540. L ACHENBRUCH , A. H., S ASS , J. H., M ONROE , R. J. & M OSES , T. H., J R . 1976. Geothermal setting and simple magmatic models for Long Valley caldera. Journal of Geophysical Research, 81, 769– 784. L EEMAN , W. P. 1982. Development of the Snake River Plain – Yellowstone Plateau province, Idaho and Wyoming: an overview and petrologic model. In: B ONNICHSEN , B. & B RECKENRIDGE , R. M. (eds) Cenozoic Geology of Idaho. Idaho Bureau of Mines and Geology Bulletin, 26, 155–177. L EEMAN , W. P. & V ITALIANO , C. J. 1976. Petrology of Mckinney – Basalt, Snake-River – Plain, Idaho. Geological Society of America Bulletin, 87, 1777–1792. L EEMAN , W. P., M ENZIES , M. A., M ATTY , D. J. & E MBREE , G. F. 1985. Strontium, neodymium, and lead isotopic compositions of deep crustal xenoliths from the Snake River Plain: evidence for Archean basement. Earth and Planetary Science Letters, 75, 354– 368. L EEMAN , W. P., O LDOW , J. S. & H ART , W. K. 1992. Lithosphere-scale thrusting in the western U.S. Cordillera as constrained by Sr and Nd isotopic transitions in Neogene volcanic rocks. Geology, 20, 63–66. M ABEY , D. R. 1982. Geophysics and tectonics of the Snake River Plain, Idaho. In: B ONNICHSEN , B. &

258

W. P. LEEMAN ET AL.

B RECKENRIDGE , R. M. (eds) Cenozoic Geology of Idaho. Idaho Bureau of Mines and Geology Bulletin, 26, 139 –153. M ANKINEN , E., H ILDEBRAND , T. G., Z IENTEK , M. L., B OX , S. E., B OOKSTROM , A. A., C ARLSON , M. H. & L ARSEN , J. C. 2004. Guide to Geophysical Data for the Northern Rocky Mountains and Adjacent Areas, Idaho, Montana, Washington, Oregon, and Wyoming. U.S. Geological Survey Open-File Reports, 2004-1413. M ANNING , C. E. & I NGEBRITSEN , S. E. 1999. Permeability of the continental crust: implications of geothermal data and metamorphic systems. Reviews in Geophysics, 27, 127– 150. M ANNING , D. A. C. 1981. The effect of fluorine on liquidus phase relationships in the system qz-ab-or with excess water at 1 kb. Contributions to Mineralogy and Petrology, 76, 206– 215. M C C URRY , M., H AYDEN , K. P., M ORSE , L. H. & M ERTZMAN , S. 2008. Genesis of post-hotspot A-type rhyolite of the Eastern Snake River Plain volcanic field by extreme fractional crystallization of olivine tholeiite. Bulletin of Volcanology, 70, 361– 368, DOI: 10.1007/s0445-007-0143-4. M ENZIES , M. A., L EEMAN , W. P. & H AWKESWORTH , C. J. 1984. Geochemical and isotopic evidence for the origin of continental flood basalts with particular reference to the Snake River Plain, Idaho, USA. Transactions of the Royal Society of London, A310, 643– 660. M ORGAN , L. A. & M C I NTOSH , W. C. 2005. Timing and development of the Heise volcanic field, Snake River Plain, Idaho, western USA. Geological Society of America Bulletin, 117, 288–306. N ASH , B. P., P ERKINS , M. E., C HRISTENSEN , J. N., L EE , D.-C. & H ALLIDAY , A. N. 2006. The Yellowstone hotspot in space and time: Nd and Hf isotopes in silicic magmas. Earth and Planetary Science Letters, 247, 143–156. P AKISER , L. C. 1989. Geophysics of the Intermontane system. In:P AKISER , L. C. & M OONEY , W. D. (eds) Geophysical Framework of the Continental United States. Geological Society of America Memoirs, 172, 235– 247. P ATIN˜ O D OUCE , A. E. 1997. Generation of metaluminous A-type granites by low-pressure melting of calc-alkaline granitoids. Geology, 25, 743 –746. P ENG , X. & H UMPHREYS , E. D. 1998. Crustal velocity structure across the eastern Snake River Plain and the Yellowstone swell. Journal of Geophysical Research, 103, 7171– 7186. P ERKINS , M. E. & N ASH , W. P. 2002. Explosive silicic volcanism of the Yellowstone hotspot: The ash fall tuff record. Geological Society of America Bulletin, 114, 367–381. P ERKINS , M. E., N ASH , W. P., B ROWN , F. H. & F LECK , R. J. 1995. Fallout tuffs of Trapper Creek, Idaho – a record of Miocene explosive volcanism in the Snake River Plain volcanic province. Geological Society of America Bulletin, 107, 1484– 1506. P ETFORD , N. & G ALLAGHER , K. 2001. Partial melting of mafic (amphibolitic) lower crust by periodic influx of basaltic magma. Earth and Planetary Science Letters, 193, 483 –499.

P IERCE , K. L. & M ORGAN , L. A. 1992. The track of the Yellowstone hot spot: Volcanism, faulting, and uplift. In: L INK , P., K UNTZ , M. A. & P LATT , L. B. (eds) Regional Geology of Eastern Idaho and Western Wyoming. Geological Society of America Memoirs, 179, 1 –53. R IGHTMIRE , C. T., Y OUNG , H. W. & W HITEHEAD , R. L. 1976. Geothermal Investigations in Idaho. US Geological Survey and Idaho Department of Water Resources, Boise, ID, Water Information Bulletins, 30, 1– 28. R ODGERS , D. W., O RE , H. T., B OBO , R. T., M C Q UARRIE , N. & Z ENTNER , N. 2002. Extension and subsidence of the eastern Snake River Plain, Idaho. In: B ONNICHSEN , B., M C C URRY , M. & W HITE , C. M. (eds) Tectonic and Magmatic Evolution of the Snake River Plain Volcanic Province. Idaho Geological Survey Bulletins, 30, 121– 155. R YAN , M. P. 1987. Elasticity and contractancy of hawaiian olivine tholeiite and its role in the stability and structural evolution of subcaldera magma reservoirs and rift systems. In: D ECKER , R. W., W IGHT , T. L. & S TAUFFER , P. H. (eds) Volcanism in Hawaii. US Geological Survey Professional Papers, 1350, 1395– 1447. S HERVAIS , J. W., V ETTER , S. K. & H ANAN , B. B. 2006. Layered mafic sill complex beneath the eastern Snake River Plain: evidence from cyclic geochemical variations in basalt. Geology, 34, 365–368. S MITH , R. B. & B RAILE , L. W. 1994. The Yellowstone hotspot. Journal of Volcanology and Geothermal Research, 61, 121– 187. S PARLIN , M. A., B RAILE , L. W. & S MITH , R. B. 1982. Crustal structure of the eastern Snake River Plain determined from ray trace modeling of seismic refraction data. Journal of Geophysical Research, 87, 2619– 2633. T AYLOR , S. R. & M C L ENNAN , S. M. 1985. The Continental Crust: its Composition and Evolution. Blackwell, Oxford. T HOMPSON , R. N. 1975. Primary basalts and magma genesis II. Snake River Plain, Idaho, U.S.A. Contributions to Mineralogy and Petrology, 52, 213– 232. V AZQUEZ , J. A. & R EID , M. R. 2002. Time scales of magma storage and differentiation of voluminous high-silica rhyolites at Yellowstone caldera, Wyoming. Contributions to Mineralogy and Petrology, 144, 274 –285. W AITE , G. P., S MITH , R. B. & A LLEN , R. M. 2006. VP and VS structure of the Yellowstone hot spot from teleseismic tomography: evidence for an upper mantle plume. Journal of Geophysical Research, 111, B04303, DOI: 10.1029/2005JB003867. W OOD , S. H. & C LEMENS , D. M. 2002. Geologic and tectonic history of the western Snake River Plain, Idaho and Oregon. In: B ONNICHSEN , B., M C C URRY , M. & W HITE , C. M. (eds) Tectonic and Magmatic Evolution of the Snake River Plain Volcanic Province. Idaho Geological Survey Bulletins, 30, 69– 103. W OODEN , J. L. & M UELLER , P. A. 1988. Pb, Sr, and Nd isotopic compositions of late Archean igneous rocks, eastern Beartooth Mountains: implications for crustmantle evolution. Earth and Planetary Science Letters, 87, 59– 72.

SNAKE RIVER PLAIN – YELLOWSTONE RHYOLITES W RIGHT , K. E., M C C URRY , M. & H UGHES , S. S. 2002. Petrology and geochemistry of the Miocene tuff of McMullen Creek, central Snake River Plain. In: B ONNICHSEN , B., M C C URRY , M. & W HITE , C. M. (eds) Tectonic and Magmatic Evolution of the Snake River Plain Volcanic Province. Idaho Geological Survey Bulletins, 30, 177–194. Y OUNKER , L. W. & V OGEL , T. A. 1976. Plutonism and plate tectonics: the origin of circumPacific batholiths. Canadian Mineralogist, 14, 238–244.

259

Y UAN , H. & D UEKER , K. 2005. P-wave tomogram of the Yellowstone plume. Geophysical Research Letters, 32, L07304, DOI: 10.1029/2004GL022056. Y UND , R. A., S MITH , B. M. & T ULLIS , J. 1981. Dislocation-assisted diffusion of oxygen in albite. Physics and Chemistry of Minerals, 7, 185–189. Z HENG , Y. 1993a. Calculation of oxygen isotope fractionation in anhydrous silicate minerals. Geochimica et Cosmochimica Acta, 57, 1079–1091. Z HENG , Y. 1993b. Calculation of oxygen isotope fractionation in hydroxyl-bearing silicates. Earth and Planetary Science Letters, 120, 247– 263.

Uniform processes of melt differentiation in the central Izu Bonin volcanic arc (NW Pacific) SUSANNE M. STRAUB1,2 1

Lamont Doherty Earth Observatory at the Columbia University, 61 Route 9W, Palisades, NY 10964, USA (e-mail: [email protected])

2

Institute of Earth Sciences, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 11529, Taiwan, ROC

Abstract: The intra-oceanic Izu Bonin arc (NW Pacific) has produced a bimodal spectrum of melts with maxima in the basaltic andesitic (c. 53– 54 wt% SiO2) and rhyolitic range (c. 70– 72 wt% SiO2) since arc inception c. 48– 49 million years ago. Composition of phenocrysts and accessory minerals from 21 contemporaneous fallout tephras from ODP Site 782A confirm the bimodality and uniformity of the erupted melts. The basaltic andesite melts equilibrated with calcic plagioclase (c. An70 – 95), high-Mg# clino- and orthopyroxene and low-Ti titanomagnetite. Dacitic and rhyolitic melts crystallized sodic plagioclase (c. An40 – 60), low-Mg# clino- and orthopyroxene, apatite, Ti-rich titanomagnetite in addition to occasional ilmenite and amphibole. The Izu melts are inferred to crystallize at oxygen fugacities between c. 0 to þ2.5 log10 units relative to FMQ, at temperatures between c. 7758 and 1100 8C and at pressures between c. 300 and c. 1100 MPa, corresponding to c. 5 –35 km lithospheric depth. The compositional uniformity of the tephra layers, which are spaced on average 230 + 380 ka apart, suggest uniform processes of differentiation since at least c. 42 Ma ago. The tephra record shows no indication of periodic or progressive crustal growth that might correlate with the alternate periods of arc formation, arc rifting or backarc spreading, or would suggest an increasingly efficient ‘crustal filter’ with time. The tephra data tentatively conform to a model where crust grows steadily through intrusions of mafic and evolved melt body batches whereby buoyancy controls the level of solidification. While the tephra compositions demonstrate the uniformity of the processes of melt formation and differentiation through time, the data do not permit the differentiation processes themselves to be constrained. These may comprise fractional crystallization, crustal fusion, fusion of non-peridotitic sub-crustal lithologies, or any combination of these processes.

In recent years, dacitic to rhyolitic rocks (c. 65– 75 wt% SiO2) have become recognized as common and volumetrically significant components of intra-oceanic island arcs. In the extrusive series, high-silica magmas form another, if weaker, major mode next to the predominant basaltic to andesite series (c. 50–60 wt% SiO2; e.g. Lee et al. 1995; Tamura & Tatsumi 2002; Leat et al. 2003; Smith et al. 2003). Moreover, low seismic P wave velocities (6.0–6.5 km s21) suggest that high-silica intrusive rocks (granites and tonalites) may comprise up to 25 vol% of the arc crust (e.g. Izu Bonin, Mariana, South Sandwich arcs, Suheyiro et al. 1996; Leat et al. 2003; Takahashi et al. 2007). Because intra-oceanic arcs initiate on oceanic crust, the high-silica melts cannot be supplied from an external component, but must have been generated by indigenous processes. Assuming a basaltic flux through the Moho, high-silica melts can form by (i) fractional crystallization of basaltic parental melts, or (ii) by partial fusion of fusible components in the crust, such as hydrated, amphibolized lower crust (e.g. Smith et al. 2003) or hydrated intrusive andesite (e.g. Tamura &

Tatsumi 2002). In either model, the oceanic crust acts as filter that impedes the ascent of primitive mantle melts and causes them to stagnate and differentiate. As the addition of mantle melt and melt differentiation continues through time, the crust should then thicken and evolve towards more siliceous compositions (Suheyiro et al. 1996; Tatsumi & Kogiso 2003; Takahashi et al. 2007). Such progressive crustal evolution must be visible in the chemistry of arc melts that are considered sensitive to crustal thickness and composition (e.g. Gill 1981; Leeman 1983; Tamura & Tatsumi 2002). The intra-oceanic Izu Bonin arc is an excellent setting to investigate long-term trends of arc melts for several reasons: (i) the arc was initiated on oceanic crust about c. 49 million years ago, (ii) there is abundant evidence of high-silica magmas that may have contributed as much as a third of the arc’s total magma production (Suheyiro et al. 1996; Tamura & Tatsumi 2002; Bryant et al. 2003; Haraguchi et al. 2003; Straub 2003) and (iii) a temporally precise and highly resolved rock records exist of the last 42 million years of

From: ANNEN , C. & ZELLMER , G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications, 304, 261–283. DOI: 10.1144/SP304.13 0305-8719/08/$15.00 # The Geological Society of London 2008.

262

S. M. STRAUB

arc evolution by means of the distal tephra fallout (Bryant et al. 1999; Straub 2003). Previous investigation of the tephra record showed a notable absence of any temporal trends of major elements that are sensitive to crustal thickness and differentiation (Straub 2003). To further investigate these findings, the composition of phenocrysts from selected tephra melts was analysed. These data confirm that mafic and evolved melts have formed through the life of the arc by very similar processes of melt differentiation, and cast doubt on models that predict crustal evolution by progressively stronger crustal processing of mantle derived melts.

Geological setting Detailed summaries of the geology and the evolution of the Izu Bonin volcanic arc can be found elsewhere (e.g. Taylor 1992; Arculus et al. 1995; Stern et al. 2003), and only the most relevant facts are re-iterated here. The Izu Bonin volcanic arc is the northern half of the intra-oceanic Izu Bonin–Mariana arc/backarc system in the northwestern Pacific that extends c. 2500 km south of Honshu, Japan (Fig. 1). Arc volcanism began in the middle Eocene (c. 48–49 Ma) with the subduction of the Mesozoic Pacific plate beneath the Philippine plate in westerly direction. The Izu Bonin–Mariana (IBM) evolution is divided into two structurally and compositionally different periods of arc formation. The first, a short period in the middle Eocene (c. 49 to c. 45 Ma, Ishizuka et al. 2006), is characterized by voluminous boninitic and tholeiitic volcanism in an extensional, rift-like setting that formed a coherent terrane of c. 3000 km length and c. 300 km width. After a drastic decrease of magma production in the late Eocene (transitional period), a classical arc with stratovolcanic chains emerged in the early Oligocene. Since then, a dynamic and mobile arc/backarc system was established that is still active today. A first episode of along-arc rifting and backarc spreading (c. 31–15 Ma) created the Shikoku backarc basin. At this time, arc volcanicity was at a minimum and may have even ceased in the early Miocene. The second arc rifting episode began c. 2–3 million years ago in the central Izu arc (Ishizuka et al. 2003b). The Eocene tholeiites and boninites are low-K basalt to rhyolites with very low contents of high-field-strength elements (HFSE) and rare-earth elements (REE) (e.g. Dobson 1986; Dobson & O’Neil 1987; Arculus et al. 1992; Bloomer et al. 1995; Pearce et al. 1999; Ishizuka et al. 2006). Low-K basalts to rhyolites with higher REE and HFSE abundances similar to N-type MORB have erupted at the central Izu arc volcanic front since

the Eocene. During the Oligocene, medium-K, high-K and shoshonitic melts also emerged, which most likely erupted in the rear-arc region (e.g. Hochstaedter et al. 2000, 2001; Schmidt 2001). Current models assume that the nascent Izu Bonin arc was constructed on mafic oceanic crust of c. 6 km thickness (e.g. Stern et al. 2003). Today, the arc crust is c. 20 km thick, and structured into upper crust (c. 45% of volume), middle crust (c. 25 vol%, at c. 7 –12 km depth) and lower crust (c. 30 vol%, at 12– 20 km depth; Suheyiro et al. 1996; Kodaira et al. 2007). The low seismic P-wave velocity of the middle crust (6.0– 6.5 km s21) suggests granitic to tonalitic compositions. These should be similar to Eocene tonalites from the Kyushu–Palau Ridge (remnant Izu arc; Haraguchi et al. 2003), or to the now exposed Miocene Tanzawa intrusive complex (Izu Peninsula) that may be a northern extension of the middle Izu crust (Kawate & Arima 1998).

Samples and analytical methods The fallout tephra investigated was drilled at ODP Site 782A (30851.660 N, 141818.850 E), which is located on the outer Izu forearc c. 120 km west of the Quaternary arc volcanic front (Fig. 1). At this location, a 70 m thick, middle Eocene to late Oligocene sequence of nannofossil chalk is separated by a major hiatus of 7.5 Ma (lower Miocene) from an upper 330 m thick sequence of Neogene nannofossil chalk (Xu & Wise 1992). The entire sequence contains abundant fallout tephra (.150 discrete layers) that range from Pleistocene to middle Eocene (c. 44.4 Ma) in age (Xu & Wise 1992; Schmidt 2001). Evidence of post-depositional transport of tephra layers is rare, and tephra beds are not significantly disturbed by bioturbation, or syn- and post-depositional erosion. Thus, the emplacement of tephra fallout are single, timediscrete events, that provide a time-precise record of arc evolution with a temporal resolution of ,1 million years. The 782A fallout tephra is mostly brown to blackish ash composed of variable amounts of juvenile lithics, scoria particles, volcanic glass and phenocrysts. Light-coloured tephra layers, composed of ash-sized pumice, are less frequent. Phenocrysts (a few 100 mm to c. 1–2 mm in size) are recognized by their euhedral shapes. They comprise plagioclase, followed by clino- and orthopyroxene, titanomagnetite and +amphibole in decreasing order of abundance. Apatite, ilmenite, magmatic quartz (very rare) and sulphides (as globules in plagioclase phenocrysts) are accessory phases. Olivine was never observed despite its common presence in the proximal mafic Izu

UNIFORM PROCESSES OF DIFFERENTIATION

263

Fig. 1. Geological setting of the Izu–Bonin arc/backarc system with DSDP and ODP drill sites. From East to West: Pacific Plate, Izu Trench, Izu Fore-arc, Izu Volcanic Front (stippled line), Sumisu rift (grabens), rear-arc with cross-chains (hatched with diagonal thick lines that denote rear-arc volcanic chains), and the inactive Shikoku Backarc Basin (double lines, spreading axis). Depth contours are in metres. ZR Zenisu Ridge; subaerial volcanoes (black triangles) unless labelled are: Myk, Miyake-jima; Mkr, Mikura-jima; A, Aoga-shima; Nis, Nishinoshima.

Bonin lavas. Fresh tephra particles, including fresh phenocrysts, occur throughout the sequence, although alteration generally increases with increasing age. In Oligocene and Eocene tephra, fresh volcanic glass is mostly confined to melt inclusions.

Sample preparation and analytical methods The tephra samples (c. 10 cm3 or less) were freezedried, stirred in ultrasound and wet-sieved through a 32 mm polyester screen using de-mineralized water. The coarse fraction of all samples was

used to prepare standard polished thin sections for petrographic and microbeam studies. Because the phenocrysts, lithics and glasses were very likely fractionated during transport, modes are unlikely to be representative, or comparable to each other, and were therefore not obtained. Glass analyses (matrix shards and melt inclusions in plagioclase and pyroxene) by electron microprobe from a total of 76 tephra samples have previously been reported by Straub (2003) and Bryant et al. (2003). From the Straub (2003) samples, a subset was chosen for the analyses of plagioclase, pyroxenes, amphibole, Fe –Ti-oxides and apatite. The analyses were

264

S. M. STRAUB

performed with a Cameca SX-50 microprobe at the GEOMAR Research Center (Kiel, Germany) equipped with four wavelength-dispersive spectrometers. Minerals were analysed at an accelerating voltage of 15 keV, a focused beam (c. 2–3 mm), 10–20 nA of beam current and peak counting times ranging between 10 and 20 s. The elemental abundances of silicates and apatite were calibrated by Smithonian mineral standards USNM 111312/ 444 (San Carlos Olivine); USNM 122142 (Kakanui Augite); USNM 104021 (Durango Apatite) (Jarosewich et al. 1980), and by Cameca mineral standards (orthoclase, albite, wollastonite, rutile and rhodonite). The Fe –Ti-oxides were principally calibrated by means of spinel 52NL-11 (Cr-spinel) and USNM 96189 (ilmenite) (Jarosewich et al. 1980). ZAF data reductions were carried out by means of the in-built PAP routines. Table 1 lists precision and accuracy of electron microprobe analyses. Full analytical data are presented in Tables 2–8. (Tables 1–8 are available online at: http://www.geolsoc.org.uk/SUP18319). For convenience, the ODP samples numbers have been coded with numbers between 1 and 118 with increasing age. The code numbers are also listed in Tables 2–8.

Results Glass compositions (glass shards, melt inclusions) Classification of the site 782 tephra glasses The glass chemistry of the Site 782A tephra glasses has been presented elsewhere (Bryant et al. 2003; Straub 2003; Straub et al. 2004), and only pertinent data are discussed here. Figure 2 shows that tephra fallout glasses range from basalt to rhyolite in composition while being bimodally distributed with maxima in the basaltic andesitic (c. 53– 54 wt% SiO2) and rhyolitic (c. 70 –72 wt% SiO2) range. A similar bimodal distribution has been reported from the proximal volcanics at the Quaternary Izu arc front (Tamura & Tatsumi 2002). The basaltic andesitic peaks of the distal fallout and proximal volcanics reasonably overlap, although the tephra glasses show a slight displacement to higher SiO2 contents as expected from multiply saturated liquids. Moreover, the low-silica alkaline basalts that are more abundant in the rear-arc region are unlikely to contribute to tephra bed formation at the outer forearc. Relative to the tephra, the siliceous mode of the proximal volcanics is far less prominent, but the true proportion of the siliceous volcanics may not be reflected in the data. Most of the high-silica volcanics are pyroclastic rocks (pumice lapilli and blocks) from

submarine siliceous calderas (Ikeda & Yuasa 1989; Yuasa & Nohara 1992; Tamura & Tatsumi, 2002) that are prone to rapid erosion and may not be as well preserved as the effusive mafic series. On the other hand, siliceous melts are more likely to prevail in the explosive, tephra-bed forming eruptions. Regardless of their true proportions, however, both distal and proximal volcanics document the ubiquity of high-silica rocks in space and time. The majority of the Site 782A tephra are low-K basalts to rhyolites (Fig. 3). Medium-K and high-K dacitic and rhyolitic glasses are subordinate (c. 9% of total glass population), and mafic medium-K and high-K glasses are largely absent. Boninitic glasses were never observed, which is consistent with the termination of boninitic volcanism prior to the emplacement of the oldest Site 782A tephra layer (c. 42 Ma, Ishizuka et al. 2006; Xu & Wise 1992). The low-K glasses have the typical characteristics of the Quaternary magmas erupting from the northern to central Izu VF (c. 34.78N to c. 28.38N). These are the flat, depleted REE patterns similar to N-MORB that are paired with very low HFSE (Nb ,2 ppm) abundances and relative enrichments in fluid-mobile large-ion lithophile elements (LILE). The origin of the medium-K and high-K glasses is more ambiguous. The most likely source of these is the Izu rear-arc region that has produced a wider range of low-K to high-K magmas (e.g. Hochstaedter et al. 2000, 2001; Taylor & Nesbitt 1998). Trace element patterns, and, in two cases, the isotope compositions (Schmidt 2001) are consistent with a rear-arc origin. Tephra zoning Results from previous work (Bryant et al. 2003; Straub 2003; Straub et al. 2004) show that some of the Site 782A tephra layers are homogenous, while the majority (about two-thirds) have a large range in composition that is linear on variation diagrams. In medium-K and high-K tephras, SiO2 may differ by up to 10 wt% within individual layers. In low-K tephras, glasses of single layers may differ to up to c. 23 wt% in SiO2, or c. 6 wt% in MgO. Compositional zoning may result from different causes, such as (i) pre-eruptive heterogeneity of co-genetic melts, (ii) simultaneous tephra input from several erupting volcanoes or (iii) postdepositional mixing (re-deposition, bioturbation). Many strongly zoned tephra layers are bimodal, which may argue for physical mixing of melts of different origin. However, physical mixing should lead to random zoning patterns, but not to patterns that are highly systematic throughout time. In site 782A tephra layers, highly incompatible elements, e.g. K2O, follow linear straight trends against an

UNIFORM PROCESSES OF DIFFERENTIATION

265

Fig. 2. SiO2 histogram of Neogene and Eocene–Oligocene tephra glasses from Site 782A (a) compared with lavas from the northern to central Izu arc and the adjacent rear region (b). Note that low-silica basalts (,50 wt% SiO2) are mostly from the rear-arc that is unlikely to be represented by the tephra. The thick line in (b) is the outline of volume-weighted SiO2 distribution of the Quaternary Izu arc after Tamura & Tatsumi (2002). Data sources for Izu arc front volcanic rocks: Langmuir et al. (2008), Hamuro et al. (1983), Aramaki et al. (1986), Yokoyama et al. (2003), Taylor & Nesbitt (1998), Tamura et al. (2005), Kuritani et al. (2003), Ishizuka et al. (2007), Hochstaedter et al. (2000), Amma-Miyasaka & Nakagawa (2002), Ikeda & Yuasa (1989), Yuasa & Nohara (1992), Notsu et al. (1983). Data sources for Izu rear-arc volcanic rocks: Machida & Ishii (2003), Ishizuka et al. (2003a), Hochstaedter et al. (2000, 2001), Taylor & Nesbitt (1998).

index of melt differentiation [e.g. SiO2, MgO or Mg#, where Mg# is the ratio of molar Mg/ (Mg þ Fe2þ) in melt, calculated assuming 15% ferric iron; Bryant et al. 2003; Straub et al. 2004]. In the K2O v. Mg# diagram (Fig. 4), glasses from layers with variable compositions form arrays of subparallel trends that differ by up to a factor of

c. 2 in K2O at a given Mg# where in single layers K2O may increase by up to a factor of 3. No ‘cross-trends’ are apparent that would display clearly more positive or more negative slopes than the average trend (e.g. Straub et al. 2004). This systematic diversity within and between tephra layers points to pre-eruptive melt

266

S. M. STRAUB

Fig. 3. K2O v. SiO2 of tephra glasses (glass shards, melt inclusions) (left) compared with lavas from the northern to central Izu arc front between 34.78N to 28.38N, and the adjacent Izu reararc (right). Note strong linear trends of K2O with increasing SiO2 in low-K tephras. For Izu arc data sources see Figure 2.

heterogeneity generated by melt differentiation processes (e.g. partial melting, fractional crystallization, melt mixing). Based on these systematics and a broader range of incompatible elements and radiogenic isotopes (Sr–Nd–Pb), a previous study by Straub et al. (2004), argued that each zoned fallout tephra represents the eruption of a single, zoned magma. Straub et al. (2004) observed that, at a given index of differentiation, only the fluid mobile LILE vary significantly, while the variations of non-fluid mobile LILE (e.g. Nb) are negligible. Because .90% of most of the fluid-mobile LILE (including K2O) must be derived from the subducting slab, the variations of the fluid mobile LILE can be attributed to their origin from mantle sources that are impregnated by slightly variable amounts of slab-derived fluids (Bryant et al. 2003; Straub et al. 2004). In contrast, nonfluid mobile LILE are mainly mantle-derived. While this model satisfactorily accounts for the variable K2O levels in different melt batches, it does not explain the regular within-layer variation of K2O. This latter range must reflect the processes of melt differentiation.

Mineral chemistry Figure 5 indicates the stratigraphic position of 21 tephra layers that have been analysed for

phenocrysts and accessory minerals. Both low-K and medium-K to high-K tephras were selected. The choice was made from standard petrographic thin sections and aimed to represent all phenocrysts and minerals of the entire Site 782A tephra spectrum through time. An important advantage of fallout tephra phenocryst is that the magma which produces the fallout tephra is instantaneously chilled upon eruption by either air or water. Relative to slow cooling lavas, which are probably further modified during post-eruptive crystallization, the chemistry of tephra phenocrysts and minerals records the pre-eruptive conditions. Plagioclase Plagioclase is the most abundant phenocryst. Plagioclase phenocrysts are commonly zoned whereby the chemical range and the complexity of the zoning patterns of individual plagioclases increase with the chemical zoning of a given single tephra. In weakly zoned tephras, plagioclase is normally zoned with broad cores and only a few mol% decrease in An towards the rims. In strongly zoned tephras, individual phenocryst may range by 35 mol% in An whereby sieve textures are common as well as inverse zoning, oscillatory zoning and abrupt changes in the An content of adjacent layers that may differ by as much as 30 mol%. There is no indication that plagioclases are xenocrysts, because glass analyses (melt

UNIFORM PROCESSES OF DIFFERENTIATION

267

Fig. 4. K2O v. Mg# of tephra glasses in selected low-K Neogene tephras. Triangles are lavas from the northern to central Izu arc front between 34.78N and 28.38N. (For data sources see Fig. 2. See Table 2 for full sample numbers of Site 782A tephra.)

inclusion or matrix shards) next to the crystals show that melt and plagioclase composition are generally correlated. The plagioclases range from anorthites to andesines (An96 – 33), with low Or content (0– 1.6 mol%), but significant FeO* (up to 1.2 wt%; Table 2). The plagioclase distribution reproduces the fundamental bimodality of the tephra glasses with peaks at c. 50 –60 and c. 88 –94 An mol%, respectively (Fig. 6a). The composition of the plagioclase in equilibrium with glass compositions cannot be precisely calculated, since the plagioclase/melt equilibrium is strongly dependent on melt water content (e.g. Blatter & Carmichael 1998b; Moore et al. 1998; Sisson & Grove 1993a; Panjasawatwong et al. 1995). The tephra melts commonly have several wt% pre-eruptive H2O that may be variably lowered by degassing (Straub & Layne 2003). Figure 6b shows a range of An values calculated from tephra glass compositions, using experimentally – Na ) for determined exchange coefficients (KDCaplag=melt

melts with variable water contents at 2 kbar (Sisson & Grove 1993). The anhydrous KD ¼ 1.0 calculates an An of plagioclase too sodic, but the KDs of 1.7 and 3.4 for melts with 2 and 4 wt% H2O produce An contents similar to those measured. Such melt water contents are in agreement with those estimated for mafic Izu VF melts (Straub & Layne 2003). Moreover, these KDs also produce plagioclase compositions from the proximal Izu volcanics that overlap – as expected – with the calcic peak of the measured plagioclase population (Fig. 6c). Clino- and orthopyroxenes Next to plagioclases, pyroxenes are the most abundant phenocrysts, and most tephras contain both clino- and orthopyroxenes. Clinopyroxenes (En34Wo28Fs4 –En51 Wo25Fs45) are augites with Mg# ¼ 60– 92 [Mg# ¼ atomic ratio Mg/(Mg þ Fe2þ)]. Measurable amounts of minor elements include Al2O3 (0.8– 5.4 wt%), TiO2 (0.08–0.89 wt%), MnO (0.07 –0.46 wt%), Na2O (0.18–0.44 wt%) as well

268

S. M. STRAUB

Fig. 5. Temporal trends of K4.0 and Si4.0 (K2O and SiO2 normalized to 4 wt% MgO) and stratigraphic positions of the tephra layers analysed for composition of phenocrysts and accessory minerals. Numbers in right panel are code numbers as given in Tables 2–8. In the K4.0 and Si4.0 v. time diagram, each point represents one tephra layer (data modified from Straub 2003; Bryant et al. 2003; Arculus & Bloomfield 1992). The numerical ages are based on sediment biostratigraphic data of Xu & Wise (1992) and physical ages of Schmidt (2001). Evolution of Izu arc/backarc system after Taylor (1992).

as NiO (up to 0.1 wt%) and Cr2O3 (up to 0.13 wt%; Table 3). Clinopyroxene phenocrysts are normally and inversely zoned, and a range of up to 8 Mg# in individual crystals is common. Orthopyroxenes (En34Wo28Fs4 –En51Wo25Fs45) are enstatites with Mg# ¼ 49 –84 (Table 4). Minor elements include Al2O3 (0.34–2.0 wt%), CaO (0.69–2.54 wt%), TiO2 (0.07 –0.35 wt%), Na2O (0.1–0.19 wt%), NiO (up to 0.15 wt%), MnO (0.16 –2.25 wt%) and Cr2O3 (up to 0.33 wt%). Orthopyroxenes are usually normally zoned. Mg# values usually differ by less than 2, and larger differences (up to 13 Mg#) are rare. The pyroxene/melt exchange coefficient – Mg of Baker et al. (1994), Grove & Bryan KDFepx=melt (1983) and Grove et al. (1982) have been determined for dry melts at atmospheric pressures for Cascade arc volcanic rocks and mid-ocean ridge lavas. Assuming the effects of pressure, temperature and melt water on the KD to be negligible, the melt Mg# can then be calculated from the pyroxene composition. Based on these values, most of the tephra melts are saturated with pyroxenes (Fig. 7). The majority of the calculated melts in

equilibrium with clinopyroxene overlap with the mafic peak of the tephra melts and extend to slightly higher melt Mg# values. The calculated melts in equilibrium with orthopyroxene are more evenly distributed, and appear to mimic the bimodal distribution of their host melts. The melts calculated to be in equilibrium with the mafic pyroxenes have compositions that overlap with those of the Neogene Izu lavas, consistent with the dominance of mafic volcanics in the proximal rock record.

Amphibole Amphibole phenocrysts are easily recognized by their euhedral shape, the intense green colours, and strong green and brown pleochroism. According to the classification of Leake (1978), the Site 782A amphiboles belong to the calcic amphiboles, and are mostly magnesiohornblendes with subordinate edenite and edenitic hornblendes. The amphiboles have significant Al2O3 (5.5– 12.2 wt%, mostly 6–8 wt%), TiO2 (0.7–3.4 wt%) with comparatively low contents of Na2O (1.0– 2.6 wt%) and K2O (0.05–0.85 wt%) (Table 5).

UNIFORM PROCESSES OF DIFFERENTIATION

269

Fig. 6. (a) Anorthite content of plagioclases from 17 fallout tephras emplaced between 0.55 and 45.9 Ma, separated into Neogene and Eocene–Oligocene age groups. (b) Plagioclase compositions calculated from tephra glasses – Na for different water contents of tephra melts after Sisson & Grove (1993) [KD ¼ 1 (anhydrous), using variable KDCaplag=melt KD ¼ 1.7 (c. 2 wt% H2O); KD ¼ 3.4 (c. 4 wt% H2O)]. (c) Plagioclase compositions calculated from the Neogene lavas of the northern to central Izu VF and adjacent rear-arc (for Izu arc data sources see Fig. 2).

Amphibole is never observed in low-K tephra, but is confined to medium-K and high-K tephras. This is consistent with the assumed rear arc origin of the latter where amphiboles are present in contrast to the Izu volcanic front (e.g. Machida & Ishii 2003). Possibly, the increased Na content of the rear-arc magmas is conducive to amphibole

saturation (Sisson & Grove 1993). Petrographically, amphiboles are always associated with the dacite – rhyolite melts. For example, amphiboles are attached to rhyolite pumice shards or intergrown with more sodic plagioclase (An40 – 76), or have inclusions of rhyolite glass, apatite and Ti-rich magnetite. Because of the strong pressure dependence of

270

S. M. STRAUB

Fig. 7. Range of melt Mg# in equilibrium with (a) orthopyroxene and (b) clinopyroxene, compared with Site 782A tephra glasses (c) and Neogene Izu arc volcanics (d) (for Izu arc data sources see Fig. 2). Numbers in italics are – Mg exchange coefficients KDFepx=melt after Baker et al. (1994), Grove & Bryan (1983) and Grove et al. (1982) that have been determined in dry melts at atmospheric pressures for Cascade arc volcanic rocks and mid-ocean ridge lavas. Numbers in italics with ‘%’ in (c) and (d) are percentages of ferric iron assumed in melt.

UNIFORM PROCESSES OF DIFFERENTIATION

271

Fig. 8. Cl v. F of apatites in high-silica tephras (.65 wt% SiO2) of Site 782A. Fluor-apatites are phenocrysts; chlor-apatites are inclusions in other phenocrysts.

Al in amphibole, it is difficult to link amphibole to melt composition. It is noted, however, that amphibole with Altot p.f.u. (,1.6) (which implies shallow crystallization pressures) is in equilibrium with the dacitic to rhyolitic melt pole, based on the experimentally determined KDAl=Siamph=melt  0.94 at a pressure of 200 MPa (Sisson & Grove 1993). Apatite Apatite crystals are commonly included in phenocrysts of dacitic and rhyolitic tephras, except for one tephra layer that has phenocrysts of fluorapatite. The confinement to siliceous melts is corroborated by the sharp decrease in P2O5 of the glasses at c. 65 wt% SiO2, which coincides with the experimentally determined apatite saturation curve (Green & Watson 1982). Apatites are enriched in both chlorine (Cl ¼ 0.8–2.4 wt%) and fluorine (F ¼ 1.2–2.4 wt%) and plot on a mixing line between the chlor- and fluor-apatite endmembers (Table 6; Fig. 8). An end member (fluorapatite; F ¼ 3.3 wt%, Cl c. 100 ppm) was found in one tephra layer only. The high Cl contents very likely reflect the high Cl/F ratios (c. 8) typical of the Izu VF melts (Straub & Layne 2003). All Cl-rich apatite minerals have the typical colourless needle-shapes, whereas fluorapatite occurs as isolated, semi-opaque, browngolden, strongly pleochroic crystal flakes. Fe–Ti-oxides Fe –Ti-oxides are present in most tephras, except for the most mafic tephra with all glasses having Mg# .45. Fe–Ti-oxides are mostly titaniferous magnetite (Usp10 – 39), with TiO2 contents between 5.1 and 16.3 wt% (Table 7). Fe–Ti-oxides are commonly included in, or attached to pyroxenes or plagioclases, but may also occur as individual crystals of up to c. 100 mm in size. Measurable minor elements in titanomagnetite are Al2O3 (1.4–4.8 wt%), MgO

(0.5– 3.6 wt%), MnO (0.2–1.4 wt%) and Cr2O3 (0– 0.7 wt%). Ilmenite (Hem8 – 26) has only been found in six high-silica tephras. It contains measurable abundances of MgO (1.1–2.9 wt%; except one data point of 5.0 wt%) and MnO (0.4–2.0 wt%; except one data point of 7.0 wt%; Table 8). Fe –Ti-oxides that are included or attached to clino- and orthopyroxenes can be used to correlate the composition of titanomagnetites with melt Mg# that is calculated from the pyroxenes (see above). Figure 9a shows that the Ti content of titanomagnetites increases broadly and linearly with a decreasing melt Mg# between c. 50 and 20. At first view, this correlation appears to suggest that TiO2 loss in the melt was cause by titanomagnetite fractionation. However, this steady increase does not coincide with the distinct ‘kink’ in the TiO2 trend of the tephra melts, where TiO2 first increases until about 40 –45 Mg# and then rapidly decreases (Fig. 9b). Rather, the onset of titanomagnetite saturation precedes these ‘kinks’. Possibly, the amount of early crystallizing low-TiO2 titanomagnetite was too low for suppressing the TiO2 increase. However, the following observations raise additional doubts on a quantitative control of titanomagnetite on melt TiO2: (i) absence of coherent trends in single tephra layer that include the tell-tale ‘kink’; (ii) straight lines of TiO2 v. Mg# in the tephra that are not typical fractionation trends; and (iii) the different slopes of the tephra trends (some of them very steep, e.g. layer 54) and a wide range of TiO2 in the Neogene Izu volcanics that do not support a single control mechanisms on the TiO2 content (Fig. 9b & c). In all, these observations question the control of titanomagnetite crystallization on the melt TiO2, which plays a key role in models of fractional crystallization models of arc melts (e.g. Gill 1981). A common finding in model calculations

272

S. M. STRAUB

Fig. 9. (a) TiO2 of titanomagnetite v. melt Mg# calculated from average Mg# of clino- and orthopyroxenes phenocrysts to which titanomagnetite is attached or into which it is included. Data are from eight different – Mg ¼ 0.26 (clinopyroxene) and tephras deposited between 0.55 and 42 Ma. Exchange coefficients are KDFepx=melt – Mg ¼ 0.30 (orthopyroxene). (b) TiO2 v. Mg# of selected tephra glasses demonstrating trends and range of TiO2 in KDFepx=melt individual melt batches (see Tables 2– 8 for full ODP sample number of the tephra). (c) TiO2 v. Mg# of Neogene Izu arc volcanics that show no relationship to the systematic trends of the tephra (for Izu arc data sources see Fig. 2).

is to is in

that the amount of titanomagnetite that needs be extracted (typically around 15–25 wt%) far in excess of the amount actually observed associated cumulates (e.g. Conrey 1990;

Woodhead 1990; Straub 1995). Still, this discrepancy may be explainable by ‘cryptic’ or ‘in-situ’crystallization where titanomagnetite influences the melt whilst being hidden in a crystal mush

UNIFORM PROCESSES OF DIFFERENTIATION

that never erupts (e.g. Langmuir 1989). However, a simpler, alternative explanation is that the increasing TiO2 content of titanomagnetite may reflect the typical re-equilibration of Fe–Ti-oxides with the evolving melt (e.g. Luhr & Carmichael 1985; Clynne & Borg 1997). In this case, Fe –Ti-oxides should have negligible quantitative influence, which would imply different causes for the TiO2 variations and trends in the Izu melts (e.g. dilution by melt mixing, source depletion by repetitive melting). Temperature of crystallization The temperatures of crystallization of the tephra melts can be calculated from pairs of coexisting clino- and orthopyroxenes, magnetites and ilmenites, and amphiboles and plagioclases (Fig. 10). Pyroxene thermometry is based on the graphic method of Lindsley & Andersen (1983), and has a precision of approximately +30 8C. Oxide thermometry is based on the methods of Stormer (1983) and Spencer & Lindsley (1981) and has a similar precision of +30 8C. The amphibole/plagioclase geothermometer is based on Holland & Blundy (1994) and has an inherent uncertainty of

273

+75 8C. Temperatures of crystallization range from c. 775 to 1100 8C, with most data falling between c. 800 and 1000 8C. The lower temperatures obtained by the oxide and plagioclase/ amphibole thermometers are consistent with the observation that ilmenite and amphibole are confined to dacitic and rhyolitic lavas. The fallout tephras do not have olivine phenocrysts, which are present in the proximal basaltic and andesitic Izu melts and imply higher crystallization temperatures of c. 1100 –1220 8C for the most mafic Izu VF melts (Amma-Miyasaka & Nakagawa 2002; Tamura et al. 2005). Oxygen fugacity Coexisting pairs of magnetite – ilmenite were only found in low-temperature, highsilica melts, and do not represent the entire tephra melt spectrum. They indicate oxygen fugacities of þ0.5 to þ2.5 log10 units relative to FMQ (Fig. 10), with no systematic difference between low-K (68, 118) and medium-K tephras (87, 88, 102, 106). The oxygen fugacities corresponds to Fe3þ/SFe ratios of 0.15– 0.33 (average 0.25 + 0.06) in the melt, calculated after Sack et al. (1980) and using the regression coefficients

Fig. 10. (a) Oxygen fugacity and temperature of crystallization based on magnetite and ilmenite pairs in six high-silica tephras deposited between 9.8 and 42.1 Ma. Calculation performed after Stormer (1983) and Spencer & Lindsley (1981). Fe– Ti-oxide pairs were tested for compositional equilibrium after Bacon & Hirschmann (1988). (b) Summary of temperatures of crystallization based on magnetite/ilmenite pairs (average of samples shown on panel above), intergrown clino- and orthopyroxene pairs (11 samples; 0.55–42.1 Ma), and intergrown plagioclase/amphibole pairs (three samples; 23.6 –26.6 Ma; all plagioclases ,An92 and amphiboles ,7.8 Si atoms p.f.u.). Pyroxene thermometer after Lindsley & Anderson (1983), amphibole-plagioclase thermometer after Blundy & Holland (1990). Uncertainties are +30 8C for pyroxene and oxides geothermometers, and +75 8C for amphibole – plagioclase geothermometer.

274

S. M. STRAUB

of Kilinc et al. (1983). In addition, sulphides were observed in two more tephras layers, either interstitial in a rock fragment (sample 1H-3-129-130, 0.11 Ma), or as globules included in a calcic plagioclase phenocryst (An77 – 82) (29X-5-13-15, 12.1 Ma). The same plagioclase contains basaltic and basaltic andesitic melt inclusions with a melt Mg# of up to 50 that have up to 1400 ppm S (Straub, unpublished data). Sulphide stability points to oxygen fugacities of ,þ1.0 log10 units relative to FMQ, and probably near FMQ (Nilsson & Peach 1993; Metrich & Clocchiatti 1996). Thus, it is assumed that the tephra melts crystallized at oxygen fugacities between c. 0 and þ2.5 log10 units relative to FMQ. These oxygen fugacities are within the range obtained from mantle xenoliths in subduction zones (c. 0 to þ2.5 log10 units relative to FMQ; Brandon & Draper 1996; Blatter & Carmichael 1998a). Pressures of crystallization Because clinopyroxenes are equilibrated with most of the tephra melts, they are excellently suited to infer pressures of crystallization. The geobarometers of Grove et al. (1989) and Nimis & Ulmer (1998) can be applied, which are based solely on clinopyroxene compositions. The barometer of Grove et al. (1989), with an inherent precision of +100 MPa, is calibrated in quartz-normative basalt liquids and

is insensitive to temperature. The Nimis & Ulmer (1998) barometer is calibrated to a wide range of hydrous and anhydrous, alkaline and tholeiitic magmas, and has an inherent precision of +180 to +310 MPa. For hydrous compositions, the Nimis & Ulmer (1998) formulation is dependent on temperature. Because the tephra melts crystallize over a range of temperature and water contents, neither variable can be exactly determined for a given clinopyroxene composition. Thus, the Grove et al. (1989) barometer was used. Cross-checks showed that very similar results are obtained with the Nimis & Ulmer (1998) barometer within the feasible range of crystallization temperatures and melt water contents. Figure 11 shows that clinopyroxenes from 13 fallout tephra imply pressures of crystallization of c. 300 to c. 1100 MPa, which corresponds to c. 5–35 km depth. The distribution peaks at c. 500–800 MPa (c. 20 km) which suggests slowing, or transient stagnation, of melts at this level. The melt Mg#s calculated from clinopyroxenes correlate positively with the pressures of crystallization. Therefore, the mafic melts (black bars in Fig. 11) appear to crystallize preferentially at higher pressures than evolved melts (grey bars in Fig. 11; e.g. samples 19X-2-24-26 (45) and 41X-2-53-55 (112); Fig. 11). The level of crystallization may be confined to a few kilometres of depth

Fig. 11. Pressure of crystallization inferred from clinopyroxene compositions. Pressures of crystallization inferred after Grove et al. (1989) from clinopyroxenes of 13 different tephra layers, ranging in age from 0.55 to 43 million years. Data from four selected individual tephras (see Tables 2– 8 for full sample numbers) indicate variable levels of crystallization. Dark grey histogram bars: pressures derived from clinopyroxenes equilibrated with melt of Mg# ¼ 28– 50; black histogram bars: pressures derived from pyroxene equilibrated with melts of Mg# ¼ 50–75. Melt Mg# data are based on a pyroxene/melt exchange coefficient of KD ¼ 0.26. The Moho may not have been constant through time – dark grey horizontal layer, minimum crustal thickness, assuming that the earliest arc was constructed on oceanic crust; light grey horizontal layer, modern crustal thickness after Suheyiro et al. (1996). See text for discussion.

UNIFORM PROCESSES OF DIFFERENTIATION

Fig. 12. Comparison of tephra amphibole to amphiboles produced in hydrous high-pressure experiments (200– 1000 MPa) at temperatures between 800 and 1000 8C and at high oxygen fugacities (þ2.5 log10 unit above FQM) in Pinatubo dacites. Pinatubo dacites are comparable in composition to the siliceous tephra (Prouteau et al. 1999).

in homogenous tephras. In contrast, strongly zoned tephra with abundant mixing texture layers and a broader range of pyroxenes indicate crystallization over a larger depth range [e.g. samples 2H-4-113-114 (3) and 29X-5-13-15 (80), Fig. 11]. The pressures inferred from the clinopyroxenes also agree with estimates from amphibole compositions. Figure 12 compares the tephra amphiboles to amphiboles that were experimentally produced in hydrous high-pressure experiments (200– 1000 MPa) at temperatures between 800 and 1000 8C and at high oxygen fugacities (þ2.5 log10 units relative to FQM). The experimental melt (Pinatubo dacite) has a composition similar to the siliceous tephra (Prouteau et al. 1999). While most of the tephra amphiboles match those grown at shallow pressures, some extend to the highpressure field which point to high crystallization pressures between 400 and 1000 MPa for some siliceous melts.

Discussion Implications for crustal growth in the Izu Bonin Arc Temporal trends Major element oxides comprise .99% of the mass of arc magmas, and hence of new arc crust produced. Consequently, timedependent trends and variations of the major element oxides through the life of an arc have

275

significance for the processes of crustal growth. Prior studies of the 782A tephra showed that none of the crust-building major elements, including SiO2, Na2O or CaO, shows any temporal change after the early boninitic stage (Straub 2003). Figure 5 shows the average arc outflux for major elements of Si4.0 and K4.0 that is obtained by normalizing to 4 wt% MgO (see Straub 2003 for details of normalization). While Si4.0 remains unchanged through time, K4.0 is more variable due to medium-K and high-K tephra at c. 36, c. 29 – 25 and c. 12.3 –12.5 Ma. However, these excursions are unrelated to the silica trend and are – consistent with available trace element and isotope data (Schmidt 2001; Straub et al. 2004) – best explained as occasional input from the more distant rear-arc sources volcanoes rather than indicative of short-term temporal change. Moreover, because .90% of the K2O in the low-K arc front magmas derives from the subducting slab (Straub et al. 2004), the steady K4.0 trend indicates a steady slab flux but no sensitivity to crustal evolution and thickening. Phenocryst compositions are the image of the major composition of their equilibrium melts. They preserve melt composition and – by means of their zoning – the history of melt evolution (e.g. Kawamoto 1992; Davidson & Tepley 1997; Straub & Martin-Del Pozzo 2001). This study demonstrates that the Site 782A tephra phenocryst compositions clearly reflect the prominent bimodality of the Izu Bonin melts: the basaltic andesitic magmas crystallize calcic plagioclase (c. An70 – 95), high-Mg# clino- and orthopyroxene and low-Ti titanomagnetite. The dacitic to rhyolitic melts crystallize sodic plagioclase (c. An40 – 60), low-Mg# clino- and orthopyroxene, apatite, Ti-rich titanomagnetite in addition to occasional ilmenite and amphibole. The phenocrysts thus allow for detecting the presence of these melts in the Izu Bonin rock record, regardless of the availability of bulk rock or glass compositions. This confirms that high-silica melts are an inherent component of Izu arc volcanism and that both mafic and evolved melts were erupted since earliest stages of arc evolution in apparently roughly constant proportions through time (Figs 2 & 6). The uniformity of the melt spectrum reflected in the uniquely coherent tephra record of Site 782A is surprising in view of the suggested timedependent thickening of the Izu arc crust that also includes the formation of the evolved middle crust (Suheyiro et al. 1996; Tamura & Tatsumi 2002; Kodaira et al. 2007). In the tephra record, there is no indication of periodic crustal growth that might correlate with the alternate periods of arc formation, arc rifting or Shikoku backarc spreading. Moreover, there is no indication of a trend towards more

276

S. M. STRAUB

siliceous arc flux that would reflect an increasing efficiency as the ‘crustal filter’ thickens with time. Possible mechanisms of crustal growth The absence of long-term trends or fluctuations points to crustal growth mechanisms that are neither dependent on the tectonic evolution nor on the timeprogressive thickening of the crust. Several possibilities exist. For example, it cannot be excluded that the bulk of the Izu crust may have formed during the earliest boninitic phase that is characterized by exceptionally high crustal growth rates (e.g. Stern & Bloomer 1992). If the crustal thickness – and the processes of melt differentiation – remained fairly constant afterwards at a much lower melt production rate, no change may be evident in the rock record. Another possibility is that crustal thickness levelled early, and remained constant by periodic delamination of mafic cumulates (Tamura & Tatsumi 2002; Takahashi et al. 2007). Thirdly, if the crust indeed thickened progressively with time, a differentiation process may exist that produces a broad range of bimodal melts independently from crustal thickening. In this case, the arc crust could grow by the accumulation of intrusive and extrusive melts over time. Further insights on the mechanism of crustal growth can be gained by combining geochemical data with data on geobarometry and eruption frequency. Given the slightly differing source signatures of individual tephra layers (Bryant et al. 2003; Straub et al. 2004), each of the large, tephra bed-forming events must represent a separate batch of mantle melt. At Site 782A, the average spacing of large, tephra-bed forming eruptions is 230 + 380 ka, with about 50% of the tephra layers spaced ,100 ka apart. This number probably denotes the minimum of eruptions within the arc segment that contributes to tephra formation, since explosive eruptions will not always form discrete tephra beds in marine sediments. This eruption frequency of compositionally distinct magma batches suggests a fast ascent of mantle-derived melts through lithosphere and crust. If crystal growth rates are faster than melt ascent rates, the pressures recorded by the clinopyroxenes do not record prolonged melt stagnation. Instead, these pressures may simply record a transient halt (and possibly only slowing) of a crystallizing magma batch during ascent. Interestingly, the pressure data also suggest density stratification (Fig. 11), because the heavier, mafic melts (Mg# . 55) tend to crystallize at deeper levels than lighter, evolved melts. Thus, if a larger fraction of melts solidified at the level indicated by pyroxene crystallization owing to insufficient momentum to ascent, the crust may dominantly grow by continuous accumulation of

melts. An obvious problem with this simple model, however, is the depth of the Moho through the life of the arc. The levels of preferred melt crystallization are within the crust at the current depth of the Moho, but they are clearly below the Moho of an oceanic crust (c. 6 km) that may have existed at arc initiation. Therefore, only if the bulk of the crust had grown early, and has remained fairly constant since 42 million years, the melts would crystallize in the crust. On the other hand, if the Moho deepened from c. 6 –7 km depth to the current depth of c. 20 km, the melts would initially have crystallized in the oceanic lithosphere. Notably, the finding of relicts of low-temperature (800 –900 8C) of dacitic melts in rare mantle xenoliths from subduction settings is in agreement with the existence of low-temperature melts (mostly between 800 and 1000 8C) at lithospheric depths (Schiano et al. 1995; Prouteau et al. 1999). If the magmas (including the evolved magmas) crystallized in the subcrustal lithosphere, they would not have evolved within the crust itself. Thus, an essential factor in constraining the mechanisms of crustal growth is to identify the processes of melt formation and differentiation. This is all the more important since only the rock record is left from the pre-Quaternary Izu-Bonin evolution whereas all physical parameters (e.g. measured crustal thickness) have been destroyed.

Processes of melt differentiation Notably, no consensus exists on which petrogenetic processes generate the bimodal spectrum of melts despite the fundamental role of origin and differentiation of arc melts for mass balance studies. Where do the silica-rich melts come from? Why are mafic and siliceous melts so systematically related as shown by the trends of strongly zoned tephra layers? In general, two different concepts of arc melt formation exist. One concept assumes a basaltic flux (from peridotite mantle) through the Moho to the crust. The crust acts as a ‘filter’, where melts further differentiate by fractional crystallization and crustal assimilation/contamination. The second concept assumes that a broader spectrum of melts (basalt to dacite) might pass the Moho prior to additional differentiation in the crust (e.g. Defant & Drummond 1990; Kelemen et al. 2004). Evolved melts can be created by non-peridotitic source lithologies present below the Moho, such as the subducting slab (e.g. amphibolite, eclogite, Defant & Drummond 1990; Gomez-Tuena et al. 2007), or sources were created through melt-rock reaction processes in the subarc mantles (e.g. pyroxenites, Straub et al. 2008). Distinguishing among these models requires stringent quantitative testing of suitable rock

UNIFORM PROCESSES OF DIFFERENTIATION

series. It is a short fall of the tephras investigated here that – despite showing uniformity of melts through time – they provide insufficient data to constrain unambiguously the process of differentiation itself. The major reasons are: (i) uncertainty concerning true phase proportions, which are important parameters in quantitative models; (ii) strong gradients in zoned tephra with respect to pressure, temperature and water content, that make it impossible to reliably determine equilibrium crystal/liquid pairs; and (iii) most mafic Izu Bonin melts and phenocrysts (olivine) are not represented by the tephra so that the path of differentiation is nowhere fully preserved. For these reasons, only some general observations are discussed. Shallow crustal differentiation Fractional crystallization Fractional crystallization in its simplest form implies melt evolution under closed system conditions, and thus is a straightforward way to produce co-genetic suites of basalts to rhyolites from similar, but not identical parental melts. Because K is highly incompatible in the amphibole-free, low-K Izu melts, the minimum degree of crystallization (X ¼ 1/F) can be inferred from the equation: CO/CL ¼ F (D21)) after Shaw (1970) for a bulk DK ¼ 0, where CO is the K2O concentration in the most mafic glass, CL is the K2O concentration in the most evolved glass, F is the fraction of liquid remaining and DK is the bulk partition coefficient of K2O. For strongly zoned basaltic to rhyolitic tephras, the minimum degree of crystallization averages at c. 60 –65%. The mafic tephra melts are comparable to the basaltic to andesite proximal lavas, despite a slightly lower MgO content, which is expected from phenocryst bearing liquids. A previous loss of c. 10 –15% of olivine from either mafic tephras or lavas can account for the difference from partial melts of peridotite that have a minimum of 10 wt% MgO and Mg# ¼ 72 (Langmuir et al. 1992). Added to these numbers, the total extent of crystallization required would be around c. 70 –80%. However, several observations do not conform to a model of fractional crystallization: 1. The tephra glasses do not exhibit the curved, hyperbolic trends typical for fractional crystallization. Moreover, mixing textures are abundant. Possibly, crystallization trends have been wiped out by later melt mixing prior or during eruption. Thus, tephra glasses may be true liquids, but may not represent liquid lines of descent. 2. There is a dearth of SiO2-poor phases that are needed to expedite the increase of SiO2 in the derivate melts. For example, glasses of fallout tephra 19 (sample 782A-11X-3-0-1)

277

range from low-MgO basalts (51.7 wt% SiO2) to rhyolite (70.3 wt% SiO2) in composition. At a minimum degree of crystallization of c. 62% (based on the concomitant K2O increase from 0.20 to 0.52 wt% and assuming DK ¼ 0), the cumulate extracted must have an average value of 40.1 wt% SiO2. This value corresponds to the SiO2 of olivine, which has never been modally observed in the tephra. It is much lower that any of the co-existing silicate phases measured in this tephra [plagioclase (42.7–55.8 wt% SiO2), clinopyroxene (50.0–52.3 wt% SiO2) and orthopyroxene (51.2– 54.1 wt% SiO2)]. Crystallization of titanomagnetite may increase SiO2. However, the new data presented in this study casts doubts on a quantitative role of titanomagnetite. 3. There is no immediate explanation for the consistent compositional gap between mafic and evolved melts in most of the samples. These observations suggest that, if fractional crystallization was operative, more complex fractionation processes (e.g. in-situ crystallization (Langmuir 1989)) are required in addition to postfractionation processes (melt mixing). Crustal assimiliation/contamination. Partial melting of the crustal basement is a possible source of high-silica melts that could mix with mantle melt to form hybrid arc andesites (e.g. Eichelberger 1978; Schiano et al. 1995; Tamura & Tatsumi 2002). Strong support for such melt hybridization is provided by the bimodality of tephra melts, the abundant petrographic evidence of melt mixing, as well as the mineralogy of the evolved melts that resembles assemblages generated by hydrous melting at shallow pressures (Tamura & Tatsumi 2002). Tamura & Tatsumi (2002) proposed that the Quaternary Izu rhyolites (c. 60 –80% SiO2) were produced by 20–30% dehydration melting of intrusive, still hot (c. 800 8C) calc-alkaline andesites located in the upper to middle crust. Melting is thought be triggered by the heat of crystallization of later ascending basalts. The repetition over time should have led to accumulation of large volumes of tonalitic crust, while the mafic cumulate restites from crystallization are thought to have sunk back into the mantle (‘delamination’). Figure 13a shows that the dacites –rhyolite glass compositions overlap with data of experimental melts from Beard & Lofgren (1991) that are considered to be the best proxies of the assumed Izu crustal melts by Tamura & Tatsumi (2002). The siliceous glasses are separated from the mafic tephra melts, whereby the gap between the two fields is bridged by the trends of zoned tephras.

278

S. M. STRAUB

Fig. 13. SiO2 and K2O v. Mg# in melts. (a) Strongly zoned tephra layers bridge the gap between primitive arc melts (in dark grey field) and evolved arc melts that are similar to experimentally produced siliceous melts from andesite protoliths at pressures between 100 and 690 MPa (light grey field). Experimental melts after Beard & Lofgren (1991), MORB field after Niu & Batiza (1997). (b) Observed trends (grey symbols with stippled lines) compared with calculated mixing trends (assuming 15% ferric iron) for selected mafic and siliceous end members of the tephra spectrum. Note that most of the FeO in the evolved tephra melts must be ferrous to obtain a low Mg# in the tephra melt. See text for further discussion.

UNIFORM PROCESSES OF DIFFERENTIATION

However, the following observations do not fit with the Tamura & Tatsumi (2002) model: 1. In Figure 13, the trends of SiO2 and K2O v. Mg# between mafic and siliceous end members are straight. This is significant because mixing trends must be strongly curved, since the parameter Mg# is a ratio and the Mg# numbers of the end member melts are very different (Fig. 13b). This is also true if most of the Fe in the dacitic– rhyolitic melt was ferric. Moreover, only at low melting temperatures (,900 8C), the experimental melts have very low FeO and MgO, and a Mg# value (similar to the measured Izu glasses). Higher FeO and MgO abundance in the melts (up to 7 wt% FeO, and 3–5 wt% MgO) that would create shallower (but not straight) mixing curves would require unreasonably high melting temperatures (.900 8C, Beard & Lofgren 1991; Johnson & Plank 1999; Rapp et al. 1999). 2. In the 782A tephra profile, mafic and evolved melts are paired with respect to their K2O abundances, thus causing the striking array of ‘sub-parallel’ trends (Figs 4 & 13). Such consistent pairing of melts from two fundamentally different lithologies (crust v. mantle) is unexpected, since both crustal and mantle sources are significantly heterogeneous in K2O (Bryant et al. 2003; Straub 2003; Straub et al. 2004). A new mantle melt has no control on which parcel of crust it will melt, and hence ‘cross-trends’ are expected to result from random melting. Moreover, at the given crustal K2O contents of intermediate rocks (c. 0.2 –0.6 wt% K2O) any lowtemperature (,900 8C), low-degree crustal melts should have much higher K2O contents than observed in the siliceous tephra, since there is no major phase that retained K2O during melting. However, only few siliceous tephra melts from the rear-arc have K2O . 1.2 wt%, and if such high K2O partial melts ever existed at the arc front, they have been erased without trace. Possibly, a more complex combination of fractional crystallization and crustal melting might be successful in fitting the observed trends of differentiation. On the other hand, the more processes are involved, the more difficult it will be to create melts of such uniform compositions and regular differentiation trends over an extended period of time. A different source lithology? The simplest way to create co-genetic series in which K2O varies by factor of c. 3 would be melting of a single source.

279

However, neither the Izu crust nor a peridotitic subarc mantle can create a range of basaltic to rhyolitic melt. The crust does not appear suitable because there is no heat source to force large extents of melting in order to generate of mafic melts. The peridotite mantle does not appear suitable because partial melting cannot create highsilica melt with low MgO and low Mg#’s (Langmuir et al. 1992). On the other hand, suitable sources may be present below the Moho. Amphibolite, eclogite and pyroxenite lithologies produce siliceous melt at a low degree of melting (e.g. Spulber & Rutherford 1983; Beard & Lofgren 1991; Rapp & Watson 1995; Kogiso et al. 2004). In the Izu arc, a slab source (amphibolite or eclogite) of the highsilica melt (‘adakite model’) can be ruled out because there is no evidence for a garnet signature in the REE patterns of the Izu VF melts (Taylor & Nesbitt 1998; Bryant et al. 2003; Straub et al. 2004). However, recent results indicate that the subarc peridotite mantle could be partially transformed to pyroxenite prior to melting (Sobolev et al. 2005, 2007; Straub et al. 2006). This partial transformation may have been caused by addition of silica (as hydrous fluid or melt) from the slab, reacting with olivine to form orthopyroxene and impregnating the surrounding peridotite (Straub et al. 2008). Olivine-free pyroxenite mantle can create a much broader range of melts including andesite and dacite melts with low MgO and Mg# contents (e.g. Kogiso et al. 2004). If those mingled with co-eval melts from a peridotite source, a bimodal spectrum of melt could be produced. Moreover, low-temperature (800 –900 8C) dacitic melts have been found in rare mantle xenoliths from subduction settings that are in equilibrium with the mantle (Schiano et al. 1995; Prouteau et al. 1999; Eiler et al. 2007). While the origin of these melts is not yet clear, the possibility clearly exists that the subarc mantle may create a broader spectrum of melts that may have a bearing on the origin of high-silica rocks in oceanic arcs.

Conclusions The following are the conclusions of this study: 1. Volcanic glasses and phenocrysts of distal tephra fallout demonstrate that the interoceanic Izu Bonin arc has produced a bimodal spectrum of basaltic to rhyolitic melts of similar proportions for at least c. 42 million years. 2. The compositional uniformity of these melts points to uniform processes of melt differentiation throughout the arc history. There is no indication of periodic crustal growth that

280

3.

S. M. STRAUB

might correlate with the alternate periods of arc formation, arc rifting or backarc spreading. There is no apparent relationship between the trends of crust-forming major element oxides and the alleged thickening and differentiation of the evolving arc crust. While the tephra demonstrate the uniformity of the processes of melt differentiation through time, the data are inconclusive as to distinguishing among different melt evolution processes, such as fractional crystallization, crustal melting, a combination of these or possibly an origin from non-peridotitic (pyroxenitic?) sub-arc lithologies.

The 782A tephra was made available through the Ocean Drilling Program (College Station, TX, USA). Many thanks to A. Freundt for providing calculation software for structural formulas as well as various geothermometers and -barometers which include the Excel spreadsheet with the Nimis & Ulmer (1998) pyroxene barometer. Mineral standards and reference samples were provided by the Smithonian Institution (E. Jarosewich) and the Harvard Mineralogical Museum (C. Francis). P. Gloer is thanked for help with electron microprobe work. G. Prouteau made available the full compositional data from experimentally produced amphiboles. Constructive reviews by D. Peate and I. Smith, and editorial comments by G. Zellmer significantly improved the manuscript. The study was funded by the Deutsche Forschungsgemeinschaft. A visiting fellowship from the Institute of Earth Sciences, Academia Sinica/Taiwan enabled me to write up these results.

References A MMA -M IYASAKA , M. & N AKAGAWA , M. 2002. Origin of anorthite and olivine megacrysts in island-arc tholeites: petrological study of 1940 and 1962 ejecta from Miyake-jima volcano, Izu–Mariana arc. Journal Volcanology and Geothermal Research, 117, 263– 283. A RAMAKI , S., H AYAKAWA , Y., F UJI , T., N AKAMURA , K. & F UKUOKA , T. 1986. The October 1983 eruption of Miyakejima volcano. Journal Volcanology and Geothermal Research, 29, 203 –229. A RCULUS , R. J. & B LOOMFIELD , A. L. 1992. Majorelement chemistry of ashes from sites 782, 784, and 786 in the Bonin Forearc. In: F RYER , P., P EARCE , J. A., S TOKKING , L. B. ET AL . (eds), Proceedings of the ODP Scientific Results 125. Ocean Drilling Program, College Station TX, 277– 292. A RCULUS , R. J., G ILL , J. B., C AMBRAY , H., C HEN , W. & S TERN , R. J. 1995. Geochemical evolution of arc systems in the Western Pacific: the ash and turbidite record recovered by drilling. In: T AYLOR , B. & N ATLAND , J. (eds) Active Margins and Marginal Basins of the Western Pacific. American Geophysical Union, Washington, DC, 45– 65. A RCULUS , R. J., P EARCE , J. A., M URTON , B. J. & VANDER L AAN , S. R. 1992. Igneous stratigraphy and

major-element geochemistry of Holes 786A and 786B. In: F RYER , P., P EARCE , J. A., S TOKKING , L. B. ET AL . (eds) Proceedings of ODP Scientific Research 125. Ocean Drilling Program, College Station, TX, 168. B ACON , C. & H IRSCHMANN , M. 1988. Mg/Mn partioning as a test for equilibrium between coexisting Fe–Tioxides. American Mineralogy, 73, 57–61. B AKER , M. B., G ROVE , T. L. & P RICE , R. 1994. Primitive basalts and andesites from the Mt. Shasta region, N. California: products of varying melt fraction and water content. Contributions to Mineralogy and Petrology, 118, 111–129. B EARD , J. S. & L OFGREN , G. E. 1991. Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3 and 6.9 kbar. Journal of Petrology, 32, 365– 401. B LATTER , D. L. & C ARMICHAEL , I. S. E. 1998a. Hornblende peridotites xenoliths from central Mexico reveal the highly oxidised nature of the subarc upper mantle. Geology, 26, 1035– 1083. B LATTER , D. L. & C ARMICHAEL , I. S. E. 1998b. Plagioclase-free andesites from Zitacuaro (Michoacan), Mexico: petrology and experimental constraints. Contributions to Mineralogy and Petrology, 132, 121–138. B LOOMER , S. H. ET AL . 1995. Early arc volcanism and the ophiolite problem: a perspective from drilling in the Western Pacific. In: T AYLOR , B. & N ATLAND , J. (eds) Active Margins and Marginal Basins of the Western Pacific. American Geophysical Union, Washington, DC, 1– 30. B LUNDY , J. D. & H OLLAND , T. J. B. 1990. Calcic amphibole equilibria and a new amphibole –plagioclase geothermometer. Contributions to Mineralogy and Petrology, 104, 208–224. B RANDON , A. D. & D RAPER , D. S. 1996. Constraints on the origin of the oxidation state of mantle overlying subduction zones: an exemple from Simcoe, Washington, USA. Geochimica et Cosmochimica Acta, 60, 1739– 1749. B RYANT , C. J., A RCULUS , R. J. & E GGINS , S. M. 1999. Laser ablation-inductively coupled plasma-mass spectrometry and tephras: a new approach to understanding arc magma genesis. Geology, 27, 1119– 1122. B RYANT , C. J., A RCULUS , R. J. & E GGINS , S. M. 2003. The geochemical evolution of the Izu– Bonin arc system: a perspective from Tephras recovered by deep-sea drilling. Geochemistry Geophysics Geosystems, 4, 1094, DOI: 10.1029/2002GC000427. C LYNNE , M. A. & B ORG , L. E. 1997. Olivine and chromian spinel in primitive calc-alkaline and tholeiitic lavas from the southernmost Cascade Range, California: a reflection of relative fertility of the source. Canadian Mineralogy, 35, 453– 472. C ONREY , R. M. 1990. Comments on a paper by Jon D. Woodhead titled ‘The origin of geochemical variations in Mariana lavas: a general model for petrogenesis in intra-oceanic island arcs?’ Journal of Petrology, 31, 957–962. D AVIDSON , J. P. & T EPLEY , F. J. 1997. Recharge in volcanic systems: evidence from isotope profiles of phenocrysts. Science, 275, 826– 829, DOI: 10.1126/ science.275.5301.826.

UNIFORM PROCESSES OF DIFFERENTIATION D EFANT , M. & D RUMMOND , M. 1990. Derivation of some modern arc magmas by melting of young subducted lithosphere. Nature, 347, 662– 665. D ICK , H. J. B. & B ULLEN , T. 1984. Chromian spinel as a petrogenetic indicator in abyssal and alpine-type peridotites and spatially associated lavas. Contributions to Mineralogy and Petrology, 86, 54– 76. D OBSON , P. F. 1986. The petrogenesis of boninite: a field, petrologic, and geochemical study of the volcanic rocks of Chichi-jima, Bonin Islands, Japan. PhD Thesis, Stanford University. D OBSON , P. F. & O’N EIL , J. R. 1987. Stable isotope compositions and water contents from boninite series volcanic rocks from Chichi-jima, Bonin islands, Japan. Earth and Planetary Science Letters, 82, 75– 86. E ICHELBERGER , J. C. 1978. Andesitic volcanism and crustal evolution. Nature, 275, 21– 27. E ILER , J. M., S CHIANO , P., V ALLEY , J. W., K ITA , N. T. & S TOLPER , E. M. 2007. Oxygen-isotope and traceelement constraints on the origins of silica-rich melts in the sub-arc mantle. Geochemistry Geophysics Geosystems, 8, Q09012, DOI:10.1029/2006GC001503. G ILL , J. 1981. Orogenic Andesites and Plate Tectonics. Springer, Berlin. G OMEZ -T UENA , A., L ANGMUIR , C. H., G OLDSTEIN , S. L., S TRAUB , S. M. & O RTEGA -G UTIERREZ , F. 2007. Geochemical Evidence for Slab Melting in the Trans-Mexican Volcanic Belt. Journal of Petrology, 48, 537–562, DOI: 10.1093/petrology/egl071. G REEN , T. H. & W ATSON , E. B. 1982. Crystallization of apatite in natural magmas under high pressure, hydrous conditions, with particular reference to orogenic rock series. Contributions to Mineralogy and Petrology, 79, 96–105. G ROVE , T. L. & B RYAN , W. B. 1983. Fractionation of pyroxene– phyric MORB at low pressure: an experimental study. Contributions to Mineralogy and Petrology, 84, 293– 309. G ROVE , T. L., G ERLACH , D. C. & S ANDO , T. W. 1982. Origin of calc-alkaline series lavas at Medicine Lake Volcano by fractionation, assimilation and mixing. Contributions to Mineralogy and Petrology, 80, 160–182. G ROVE , T. L., K INZLER , R. J. & B ARTELS , K. S. 1989. Effects on pressure on alumina substition in igneous augite: an empirical barometer. EOS Transactions of the AGU, 70, 1401–1402. H AMURO , K., A RAMAKI , S. & U TO , K. 1983. The Higashi–Izu-oki Submarine Volcanoes, Part 2, and the Submarine Volcanoes near the Izu Shoto Isalnds. Bulletin of the Earthquake Research Institute, 58, 527–557. H ARAGUCHI , S., I SHII , T., K IMURA , J. I. & O HARA , Y. 2003. Formation of tonalite from basaltic magma at the Komahashi– Daini Seamount, northern Kyushu– Palau Ridge in the Philippine Sea, and growth of Izu– Ogasawara (Bonin)–Mariana arc crust. Contributions to Mineralogy and Petrology, 145, 151– 168. H OCHSTAEDTER , A. F. ET AL . 2000. Across-arc geochemical trends in the Izu– Bonin arc: constraints on source composition and mantle melting. Journal of Geophysical Research, 105, 495– 512. H OCHSTAEDTER , A. G. ET AL . 2001. Across-arc geochemical trends in the Izu –Bonin arc: Contributions

281

from the subducting slab. Geochemistry Geophysics Geosystems, 2, 2000GC000105 [12,7776]. H OLLAND , T. & B LUNDY , J. 1994. Non-ideal interactions in calcic amphiboles and their bearing on amphiboleplagioclase thermometry. Contributions to Mineralogy and Petrology, 116, 433– 447. I KEDA , Y. & Y UASA , M. 1989. Volcanism in nascent back-arc basins behind the Shichoto Ridge and adjacent areas in the Ogasawara arc, northwest Pacific: evidence for mixing between E-type MORB and island arc magmas at the initiation of back-arc rifting. Contributions to Mineralogy and Petrology, 101, 377–393. I SHIZUKA , O., T AYLOR , R. N., M ILTON , J. A. & N ESBITT , R. W. 2003a. Fluid– mantle interaction in an intra-oceanic arc: constraints from high-precision isotopes. Earth and Planetary Science Letters, 211, 221– 236. I SHIZUKA , O., U TO , K. & Y UASA , M. 2003b. Volcanic history of the back-arc of the Izu –Bonin (Ogasawara) arc. In: L ARTER , R. D. & L EAT , P. T. (eds) Intra-Oceanic Subduction Systems: Tectonic and Magmatic Processes. Geological Society Special Publications, London, 187–205. I SHIZUKA , O. ET AL . 2006. Early stages in the evolution of Izu– Bonin arc volcanism: New age, chemical, and isotopic constraints. Earth and Planetary Science Letters, 250, 385– 401. I SHIZUKA , O. ET AL . 2007. Processes controlling along-arc isotopic variation of the southern Izu-Bonin arc. Geochemistry Geophysics Geosystems, 8, Q06008, DOI: 10.1029/2006GC001475. J AROSEWICH , E., N ELEN , N. & N ORBERG , J. 1980. Reference samples for electron microprobe analysis. Geostandard Newsletter, 4, 43– 47. J OHNSON , M. C. & P LANK , T. 1999. Dehydration and melting experiments constrain the fate of subducted sediment. Geochemistry Geophysics Geosystems, 1, 1999GC000014. K AWAMOTO , T. 1992. Dusty and honeycomb plagioclase: indicators of processes in the Uchino stratified magma chamber. Journal of Volcanology and Geothermal Research, 49, 191– 208. K AWATE , S. & A RIMA , M. 1998. Petrogenesis of the Tanzawa plutonic complex, central Japan: exposed felsic middle crust of the Izu-Bonin–Mariana arc. The Island Arc, 7, 342– 358. K ELEMEN , P. B., H ANGHOI , K. & G REENE , A. R. 2004. One view of the geochemistry of subduction-related magmatic arcs, with an emphasis on primitive andesite and lower crust. In: H OLLAND , H. D. & T UREKIAN , K. K. (eds) The Crust, Vol. 3, Treatise on Geochemistry. Elsevier, Oxford, 593– 659. K ILINC , A., C ARMICHAEL , I. S. E., R IVERS , M. L. & S ACK , R. O. 1983. The ferric –ferrous ratio of natural silicate liquids equilibrated in air. Contributions to Mineralogy and Petrology, 83, 136– 140. K ODAIRA , S. ET AL . 2007. Seismological evidence for variable growth of crust along the Izu intraoceanic arc. Journal of Geophysical Research, 112, B05104, DOI: 10.1029/2006JB004593. K OGISO , T., H IRSCHMANN , M. & P ERTERMANN , M. 2004. High-pressure partial melting of mafic

282

S. M. STRAUB

lithologies in the mantle. Journal of Petrology, 45, 2407–2422, DOI: 101093/petrology/egh057. K URITANI , T., Y OKOYAMA , T., K OBAYASHI , K. & N AKAMURA , E. 2003. Shift and rotation of composition trends by magma mixing: 1983 eruption at Mikayejima Volcano, Japan. Journal of Petrology, 44, 1895–1916. L ANGMUIR , C. H. 1989. Geochemical consequences of in-situ crystallisation. Nature, 340, 199 –205. L ANGMUIR , C. H., K LEIN , E. M. & P LANK , T. 1992. Petrological systematics of mid-ocean ridge basalts: constraints on melt generation beneath ocean ridges. In: M ORGAN , J. P., B LACKMAN , D. K. & S INTON , J. M. (eds) Mantle Flow and Melt Generation at MidOcean Ridges. Geophysical Monographs, American Geophysical Union, Washington, DC, 183–280. L ANGMUIR , C. H. ET AL . 2008. Petrogenesis of Torishima and adjacent volcanoes of the Izu –Bonin arc: one end member of the global spectrum of arc basalts compositions. Contributions to Mineralogy and Petrology, unpublished data. L EAKE , B. E. 1978. Nomenclature of amphiboles. Canadian Mineralogy, 16, 501–520. L EAT , P. T., S MELLIE , J. L., M ILLAR , I. L. & L ARTER , R. D. 2003. Magmatism in the South Sandwich Arc. In: L ARTER , R. D. & L EAT , P. T. (eds) Intra-Oceanic Subduction Systems: Tectonic and Magmatic Processes. Geological Society Special Publications, London, 285–313. L EE , J. M., S TERN , R. J. & B LOOMER , S. H. 1995. Forty million years of magmatic evolution in the Mariana arc: the tephra record. Journal of Geophysical Research, 100, 17671–17687. L EEMAN , W. P. 1983. The influence of crustal structure on compositions of subduction-related magmas. Journal of Volcanology and Geothermal Research, 18, 561 –588. L INDSLEY , D. H. & A NDERSEN , D. J. 1983. A twopyroxene thermometer. Journal of Geophysical Research, 88(suppl.), A887– A906. L UHR , J. F. & C ARMICHAEL , I. S. E. 1985. Jorullo Volcano, Michoaca´n, Mexico (1759–1774): the earliest stages of fractionation in calc-alkaline magmas. Contributions to Mineralogy and Petrology, 90, 142– 161. M ACHIDA , S. & I SHII , T. 2003. Backarc volcanism along the en echelon seamounts: The Enpo seamount chain in the northern Izu– Ogasawara arc. Geochemistry Geophysics Geosystems, 4, 9006, DOI: 10.1029/ 2003GC000554. M C G UIRE , A. V., D YAR , M. D. & F RANCIS , C. A. 1992. Mineral standards for electron microprobe analysis for oxygen. American Mineralogy, 77, 1087–1091. M ETRICH , N. & C LOCCHIATTI , R. 1996. Sulfur abundance and its speciation in oxidized alkaline melts. Geochimica et Cosmochimica Acta, 60, 4151– 4160. M OORE , G., V ENNEMAN , T. & C ARMICHAEL , I. S. E. 1998. An empirical model for the solubility of H2O in magmas up to 3 kilobars. American Mineralogy, 83, 36– 42. N ILSSON , K. & P EACH , C. 1993. Sulfur speciation, oxidation state, and sulfur concentration in backarc magmas. Geochimica et Cosmochimica Acta, 57, 3807–3813.

N IMIS , P. & U LMER , P. 1998. Clinopyroxene barometry of magmatic rocks Part 1: An expanded structural geobarometer for anhydrous and hydrous, basic and ultrabasic systems. Contributions to Mineralogy and Petrology, 133, 122–135. N IU , Y. & B ATIZA , R. 1997. Trace element evidence from seamounts for recycled oceanic crust in the Eastern Pacific mantle. Earth and Planetary Science Letters, 148, 471– 483. N OTSU , K., I SSHIKI , N. & H IRANO , M. 1983. Comprehensive strontium isotope study of Quaternary volcanic rocks from the Izu– Ogasawara arc. Geochemistry Journal, 17, 289–302. P ANJASAWATWONG , Y., D ANYUSHEVSKY , L. V., C RAWFORD , A. J. & H ARRIS , K. L. 1995. An experimental study of the effects of melt composition on plagioclase –melt equilibria at 5 and 10 kbar: implications for the origin of magmatic high-An plagioclase. Contributions to Mineralogy and Petrology, 118, 420– 432. P EARCE , J. A., K EMPTON , P. D., N OWELL , G. M. & N OBLE , S. R. 1999. Hf– Nd element and isotope perspective on the nature and provenance of mantle and subduction zone components in Western Pacific arc-basin systems. Journal of Petrology, 40, 1579– 1611. P ROUTEAU , G., S CAILLET , B., P ICHAVANT , M. & M AURY , R. 1999. Fluid-present melting of ocean crust in subduction zones. Geology, 27, 1111– 1114. R APP , R. P. & W ATSON , E. B. 1995. Dehydration melting of metabasalt at 8 –32 kbar: Implications for continental growth and crust-mantle recycling. Journal of Petrology, 36, 891– 931. R APP , R. P., S HIMIZU , N., N ORMAN , M. D. & A PPLEGATE , G. S. 1999. Reaction between slab-derived melts and peridotite in the mantle wedge: experimental constraints at 3.8 GPa. Chemical Geology, 160, 335–356. S ACK , R. O., C ARMICHAEL , I. S. E., R IVERS , M. & G HIORSO , M. S. 1980. Ferric– ferrous equilibria in natural silicate liquids at 1 bar. Contributions to Mineralogy and Petrology, 75, 369–376. S CHIANO , P. ET AL . 1995. Hydrous silica-rich melts in the sub-arc mantle and their relationship with erupted arc lavas. Nature, 377, 595–600. S CHMIDT , A. 2001. Temporal and spatial evolution of the Izu Island Arc, Japan, in terms of Sr– Nd– Pb isotope geochemistry. Doctoral Thesis; http://e-diss.uni-kiel. de/diss_465/ Thesis, Christians-Albrecht-Universita¨t, Kiel. S HAW , D. M. 1970. Trace element fractionation during anatexis. Geochimica et Cosmochimica Acta, 34, 237–243. S ISSON , T. & G ROVE , T. 1993. Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology, 113, 143–166. S MITH , I. E. M., W ORTHINGTON , T., S TEWART , R. B., P RICE , R. C. & G AMBLE , J. A. 2003. Felsic volcanism in the Kermadec arc, SW Pacific: crustal recycling in an oceanic setting. In: L ARTER , R. D. & L EAT , P. T. (eds) Intra-Oceanic Subduction Systems: Tectonic

UNIFORM PROCESSES OF DIFFERENTIATION and Magmatic Processes. Geological Society Special Publications, Bath. S OBOLEV , A. V., H OFMANN , A. W., S OBOLEV , V. S. & N IKOGOSIAN , I. K. 2005. An olivine-free mantle source of Hawaiian shield basalts. Nature, 434, 590–597. S OBOLEV , A. V. ET AL . 2007. The amount of recycled crust in sources of mantle-derived melts. Science, 316, 412– 417, 10.1126/science.1138113. S PENCER , K. J. & L INDSLEY , D. H. 1981. A solution model for co-existing iron–titanium oxides. American Mineralogy, 66, 1189–1201. S PULBER , S. D. & R UTHERFORD , M. J. 1983. The origin of rhyolite and plagiogranite in oceanic crust: an experimental study. Journal of Petrology, 24, 1– 25. S TERN , R. J. & B LOOMER , S. H. 1992. Subduction zone infancy: examples from the Eocene Izu– Bonin– Mariana and Jurassic California arcs. Geological Society of American Bulletin, 104, 1621–1636. S TERN , R. J., F OUCH , M. J. & K LEMPERER , S. L. 2003. An overview of the Izu– Bonin–Mariana subduction factory. In: E ILER , J. (ed.) Inside the Subduction Factory. American Geophysical Union, Washington, DC. S TORMER , J. C. 1983. The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multicomponent iron-titanium oxides. American Mineralogy, 68, 568 –594. S TRAUB , S. M. 1995. Contrasting compositions of Mariana Trough fallout tephra and Mariana island arc volcanics: a fractional crystallization link. Bulletin of Volcanology, 57, 403– 421. S TRAUB , S. M. 2003. The evolution of the Izu Bonin– Mariana volcanic arcs (NW Pacific) in terms of major elements. Geochemisry Geophysics Geosystems, 4, 1018, DOI: 10.1029/2002GC000357. S TRAUB , S. M. & L AYNE , G. D. 2003. The systematics of chlorine, fluorine and water in Izu arc front volcanic rocks: Implication for volatile recycling in subduction zones. Geochimica et Cosmochimica Acta, 67, 4179–4203. S TRAUB , S. M. & M ARTIN -D EL P OZZO , A. L. 2001. The significance of phenocryst diversity in tephra from recent eruptions at Popocatepetl volcano (Central Mexico). Contributions to Mineralogy and Petrology, 140, 487– 510. S TRAUB , S. M., L A G ATTA , A. B., M ARTIN -D EL P OZZO , A. L. & L ANGMUIR , C. H. 2008. Evidence from high Ni olivines for a hybridized peridotite/pyroxenite source for orogenic andesites from the central Mexican Volcanic Belt. Geochemistry Geophysics Geosystems, 9, Q03007, DOI: 10.1029/ 2007GC001583. S TRAUB , S. M., L ANGMUIR , C. H., M ARTIN -D EL P OZZO , A. L., L A G ATTA , A. B. & G OLDSTEIN , S. L. 2006. Mantle origin of andesites in the central

283

Mexican Volcanic Belt. Geochimica et Cosmochimica Acta, 70, A620. S TRAUB , S. M., L AYNE , G. D., S CHMIDT , A. & L ANGMUIR , C. H. 2004. Volcanic glasses at the Izu arc volcanic front: new perspectives on fluid and sediment melt recycling in subduction zones. Geochemistry Geophysics Geosystems, 5, Q01007, DOI: 10.1029/2002GC000408. S UHEYIRO , K. ET AL . 1996. Continental crust, crustal underplating, and low-Q upper mantle beneath an oceanic island arc. Science, 272, 390–392. T AKAHASHI , E. ET AL . 2007. Crustal structure and evolution of the Mariana intra-oceanic island arc. Geology, 35, 203– 206; DOI: 10.1130/G23212A. T AMURA , Y. & T ATSUMI , Y. 2002. Remelting of an andesitic crust as a possible origin for rhyolitic magma in oceanic arcs: an example from the Izu – Bonin Arc. Journal of Petrology, 43, 1029–1047. T AMURA , Y. ET AL . 2005. Are arc basalts dry, wet or both? Evidence from the Sumisu caldera volcano, Izu Bonin arc, Japan. Journal of Petrology, 46, 1769– 1803. T ATSUMI , Y. & K OGISO , T. 2003. The subduction factory: its role in the evolution of the Earth’s crust and mantle. In: L ARTER , R. D. & L EAT , P. T. (eds) Intra-oceanic Subduction Systems. The Geological Society of London, Bath, 55–80. T AYLOR , B. 1992. Rifting and the volcanic-tectonic evolution of the Izu-Bonin-Mariana Arc. In: T AYLOR , B., F UJIOKA , K. ET AL . (eds) Proceedings of ODP Scientific Research. Ocean Drilling Program, College Station TX, 627 –651. T AYLOR , R. N. & N ESBITT , R. W. 1998. Isotopic characteristics of subduction fluids in an intra-oceanic setting, Izu– Bonin-Arc, Japan. Earth and Planetary Science Letters, 164, 79–98. W OODHEAD , J. 1990. Reply to the comments of Conrey (1990). Journal of Petrology, 31, 963–966. X U , Y. & W ISE , S. W. 1992. Middle Eocene to Miocene calcareous nannofossils of Leg 125 from the western Pacific ocean. In: F RYER , P., P EARCE , J. A., S TOKKING , L. B. ET AL . (eds) Proceedings of ODP Scientific Research 125. Ocean Drilling Program, College Station TX, 43– 70. Y OKOYAMA , T., K OBAYASHI , K., K URITANI , T. & N AKAMURA , E. 2003. Mantle metasomatism and rapid ascent of slab components beneath island arcs: evidence for 238U– 230Th– 226Ra disequilibria of Miyakejima volcano, Izu arc, Japan. Journal of Geophysical Research, 108, 2329, DOI: 10.1029/ 2002JB002103. Y UASA , M. & N OHARA , M. 1992. Petrographic and geochemical along-arc variations of volcanic rocks on the volcanic front of the Izu–Ogasawara (Bonin) arc. Bulletin of the Geological Survey of Japan, 43, 421– 456.

Index

antecrysts 134, 135 Aolian Islands 34 Cathedral Peak Granodiorite 203– 204 conclusions 230– 231 discussion 223– 225 compositional variation 226–230 internal contacts 225– 226 field relations and petrography aplitic dykes 209 Cathedral Peak petrography 209– 211 interaction zones 211–214 ladder dykes 211 magmatic fabrics 215– 216 magmatic faults 215 microgranitoid enclaves 209 potassium feldspar concentrations 214 Schlieren 209 geological setting 204–205 method analytical methods 205 geochronology 206– 209 isotope data 212 major element oxides by XRF 207 mineral chemistry of plagioclase 205–206 sampling strategy 205 trace elements by ICP– MS 210– 211 trace elements by XRF 208 U– Pb zircon data 213 results feldspar chemistry 221–223 geochemistry 216– 218 Sr and Nd isotopes 218–221 U– Pb geochronology 223, 224 crustal thickness 18– 19 correlation with lava dome proportion 21, 23–25 crystal populations 133 combined textural analysis approach 138 –139 crystal size distributions 140 dihedral angles 140– 143 Kameni enclaves 140 Santorini volcano 139– 140 spatial distribution patterns 140 composition 133– 134 antecrysts 134, 135 final texture 134– 136 microlites 134, 135 phenocrysts 134, 135 xenocrysts 134, 135 conclusion 145 future developments linking textures with microgeochemical analysis 145 textural analysis 143– 145 quantifying textural parameters 136 crystal shapes 136

crystal size distribution 136–137 dihedral angles 138 spatial distribution patterns (SDPs) 137–138 crystal size distribution, three dimensional 48 differentiation, uniform processes 261–262 conclusions 279–280 discussion implication for crustal growth 275– 276 melt differentiation processes 276–279 geological setting 262 results amphibole 268 –271 apatite 271 clino- and orthopyroxenes 267– 268 crystallization pressure 274–275 crystallization temperature 273 Fe–Ti-oxides 271– 273 glass compositions 264 –266 mineral chemistry 266 –275 oxygen fugacity 273– 274 plagioclase 266– 267 samples and analytical methods 262–263 preparation 263– 264 diffusion in a spherical grain 251 equilibrium fractionation factor 251 Fourier’s law 22 gas phase exsolution 50 Gibbs free energy 62 Global Volcanism Program (GVP) 16 reliability of eruption style data 20–22 Hawaiian tholeiite basalt 51–52 irregular arcs 16 magma migration 26 Izu Bonin volcanic arc 261– 262 conclusions 279–280 discussion implication for crustal growth 275– 276 melt differentiation processes 276–279 geological setting 262 results glass compositions 264 –266 mineral chemistry 266 –275 samples and analytical methods 262–263 preparation 263– 264 Kı¯ lauea volcano 83 background and previous work 83–87 tectonics and plumbing system 85 conclusions 111

286 Kı¯ lauea volcano (Continued) interpretation 103 eruption types and eruption rate 110–111 Halema‘uma‘u eruption 103–106 magmatic history 106– 109 seismicity along magma supply path 109–110 significance of intrusion types 110 observations eruption efficiency 94 eruption periods 91 precursory sequences to eruption and intrusion 92– 94 seismic and deformation history 95–103 seismic and tilt sequences 94–95 seismic and tilt data 87–88 eruption and intrusions 90, 91 limitations on quantitative analysis of magma supply 91 map 88 Uwekahuna tilt data analysis 88–91 study assumptions 87 tilt and volume changes over eruptive cycles 112–114 lava dome proportion 16 correlation with crustal thickness 21, 23–25 correlation with plate convergence rate 22, 23– 25 correlation with surface heat flux 20, 22–23 linear elastic fracture mechanics (LEFM) 73– 74 magma chamber, short-lived stratified 149, 166 analytical methods 151–152 discussion 162 inferred magmatic processes 165– 166 petrological features of andesitic end-member magmas 162– 163 petrological features of basaltic andesite magmas 163– 164 petrological features of basaltic end-member magmas 163 three magmas for tephra layers 162 geological features of tephra layers 151 geological outline of Zao volcano 149–151 mineral chemistry 155 glass inclusion compositions 157– 158 Z-To5 layer 155–157 Z-To6 and 7 layers 157 petrography of tephra layers 152 Z-To5 layer 152–154 Z-To6 and 7 layers 154–155 whole rock compositions 158–162 mantle to crust magma transfer in volcanic arcs 15– 16 conclusions 27 findings 20 238U/230Th disequilibria 24 correlation with suface heat flux 22– 23 geochronological evidence 25– 26 magma migration in irregular arcs 26 reliability of GVP eruption style data 20– 22

INDEX methodology arcs with Holocene effusive eruptions 17 average viscosity determination 16 crustal thickness 18–19 plate convergence rate 19 surface heat flux 18 volcanic arc characterization 16– 18 results 19–20 microlites 134, 135 phenocrysts 134, 135 plate convergence rate 19 correlation with lava dome proportion 22, 23– 25 Poiseuille equation 72 Popocatepetl volcano 117 conclusions 130 discussion dome evolution 128– 129 magmatic plumbing 128 magnetic anomalies 127–128 observations seismicity 120–122 volcanic activity 119– 120 study methods eruptive activity 117– 118 magnetic signals 122– 127 magnetism 118 seismicity 119 spring water 122 spring water and ash 119 random sphere distribution line (RSDL) 138 regular arcs 16 Santorini volcano 139–140 Kameni enclaves 140 crystal size distributions 140 dihedral angles 140–143 spatial distribution patterns 140 Sciare del Fuoco 35–36, 37 silicic magma production rates 169–170 amphibole, role of 174 conclusions 179 effect of recharge 175 model I – mixing before eruption 170– 171 model II – melt assimilation during differentiation 171– 174 physical implications 175–179 sill growth model 72– 74 Snake River Plain – Yellowstone (SRPY) 235–236 basalt intrusion in middle– upper crust 244 discussion and summary 254– 256 fundamental characteristics of magmatic province crustal and mantle rhyolites 242– 244 rhyolite compositional variation 239 rhyolite petrology and temperature 238–239 rhyolite source 239 –241 volcano-tectonic overview 236–237 volume and scale of rhyolite systems 237–238

INDEX oxygen isotopic constraints on rhyolite genesis diffusivity and fractionation factor coefficients 252 oxygen isotope problem 250–251 theoretical considerations 251 –252 physical models 244 magma reservoir model 244–246 thermal models for rhyolite generation 246–250 spatial distribution patterns (SDPs) of crystals 137– 138 Stromboli volcano 33– 34 analytical methods 38 bulk composition and glass chemistry 44–48 conclusions 59–62 crystal size distribution and intrabubble distances 48– 50 golden pumice 51 scorias and lavas 50–51 geological setting and volcanological outline 34–35 eruptions, historic and current 36– 38 geological map 36 structural framework 35– 36 volcano 35 melt and magma viscosity 51–53 petrography and mineral compositions 38–44 petrology and phase relationships 53–54 plumbing system and magma chamber shape 57–59, 60 P –T path summary 58 thermobarometric constraints 54–57 surface heat flux 18 correlation with lava dome proportion 20, 22–23 Tamman–Vogel– Fulcher (TVF) 51– 53 Tindari– Letojanni fault system 34 Tuolumne Batholith (TB) 203– 204 conclusions 230– 231 discussion 223– 225 compositional variation 226–230 internal contacts 225– 226 field relations and petrography aplitic dykes 209 Cathedral Peak petrography 209– 211 interaction zones 211–214 ladder dykes 211 magmatic fabrics 215– 216 magmatic faults 215 microgranitoid enclaves 209 potassium feldspar concentrations 214 Schlieren 209 geological setting 204–205 method analytical methods 205 geochronology 206– 209 isotope data 212 major element oxides by XRF 207 mineral chemistry of plagioclase 205–206 sampling strategy 205 trace elements by ICP– MS 210– 211 trace elements by XRF 208 U– Pb zircon data 213 results feldspar chemistry 221–223 geochemistry 216– 218

Sr and Nd isotopes 218– 221 U–Pb geochronology 223, 224 Tuolumne Intrusive Suite (TIS) 183– 184 discussion low-temperature mineral equilibrium 193 origin of chemical variation 193– 195 significance of mapped plutonic units 198 –199 source of isotopic variability 195–198 spatial and temporal variations in composition 193 time–space patterns of geochemical variation 198 geological setting 184– 186 petrology major- and trace-element geochemistry 188– 190 methods 186 mineral chemistry 186 –188 radiogenic isotope geochemistry 191–193 thermobarometry 188 summary 199 viscous dissipation in sill growth 71– 72 conclusions 79– 80 implications for numerical modelling 78 implications for sill thickness prediction 78–79 near-tip region 74–75 propagation regime 75– 76 sill growth model 72– 74 sill growth regime 77–78 volcanic arcs, first-order observations 15–16 conclusions 27 findings 20 238U/230Th disequilibria 24 correlation with suface heat flux 22–23 geochronological evidence 25–26 magma migration in irregular arcs 26 reliability of GVP eruption style data 20– 22 methodology arcs with Holocene effusive eruptions 17 average viscosity determination 16 crustal thickness 18– 19 plate convergence rate 19 surface heat flux 18 volcanic arc characterization 16–18 results 19–20 Williams expansion 73–74 xenocrysts 134, 135 Zao volcano 149, 166 analytical methods 151– 152 discussion 162 inferred magmatic processes 165– 166 petrological features of andesitic end-member magmas 162–163 petrological features of basaltic andesite magmas 163–164 petrological features of basaltic end-member magmas 163 three magmas for tephra layers 162

287

288 Zao volcano (Continued) geological features of tephra layers 151 geological outline 149–151 mineral chemistry 155 glass inclusion compositions 157– 158

INDEX Z-To5 layer 155– 157 Z-To6 and 7 layers 157 petrography of tephra layers 152 Z-To5 layer 152– 154 Z-To6 and 7 layers 154– 155 whole rock compositions 158– 162