Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume III (Pageoph Topical Volumes) 9783764387372, 3764387378

Citation preview

I ••

11

Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume III Edited by Nemesio M. Pérez Sergio Gurrieri Chi-Yu King Yuri Taran

Birkhäuser Basel · Boston · Berlin

Reprint from Pure and Applied Geophysics (PAGEOPH), Volume 165 (2008) No. 1 Editors: Nemesio M. Pérez Environmental Research Division Instituto Tecnológico y de Energias Renovables Polígono Industrial de Granadilla s/n 38611 Granadilla, Tenerife Canary Islands Spain e-mail: [email protected]

Sergio Gurrieri Istituto Nazionale di Geofisica e Vulcanologia Sezione di Palermo V. Ugo La Malfa, 153 90146 Palermo Italy e-mail: [email protected]

Chi-Yu King Earthquake Prediction Research, Inc 381 Hawthorne Ave. Los Altos, CA 94022 USA e-mail: [email protected]

Yuri Taran Volcanology Department Institute of Geophysics UNAM 3000, Av. Universidad Mexico D.F., 04510 Mexico e-mail: taran@geofisica.unam.mx

Library of Congress Control Number: 2006043001 Bibliographic information published by Die Deutsche Bibliothek: Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at

ISBN 978-3-7643-8737-2 Birkhäuser Verlag AG, Basel · Boston · Berlin This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in other ways, and storage in data banks. For any kind of use permission of the copyright owner must be obtained.

© 2008 Birkhäuser Verlag AG Basel · Boston · Berlin P.O. Box 133, CH-4010 Basel, Switzerland Part of Springer Science+Business Media Printed on acid-free paper produced from chlorine-free pulp. TCF ∞ Printed in Germany ISBN 978-3-7643-8737-2 987654321

e-ISBN 978-3-7643-8738-9 www.birkhauser.ch

PURE AND APPLIED GEOPHYSICS Vol. 165, No. 1, 2008

Contents 1

Introduction N. M. Pe´rez, S. Gurrieri, C.-Y. King, Y. Taran

5

Spatial and Temporal Changes of Groundwater Level Induced by Thrust Faulting Y. Chia, J. J. Chiu, Y.-H. Chiang, T.-P. Lee, C.-W. Liu

17

Geochemical Monitoring of Geothermal Waters (2002–2004) along the North Anatolian Fault Zone, Turkey: Spatial and Temporal Variations and Relationship to Seismic Activity S. Su€er, N. Gu€lec¸, H. Mutlu, D. R. Hilton, C. C¸ifter, M. Sayin

45

Coupling Between Seismic Activity and Hydrogeochemistry at the Shillong Plateau, Northeastern India A. Skelton, L. Claesson, G. Chakrapani, C. Mahanta, J. Routh, M. Mo¨rth, P. Khanna

63

Radon Changes Associated with the Earthquake Sequence in June 2000 in the South Iceland Seismic Zone P. Einarsson, P. Theodo´rsson, A´. R. Hjartardo´ttir, G. I. Guðjo´nsson

75

CO2 Degassing over Seismic Areas: The Role of Mechanochemical Production at the Study Case of Central Apennines F. Italiano, G. Martinelli, P. Plescia

95

Changes in the Diffuse CO2 Emission and Relation to Seismic Activity in and around El Hierro, Canary Islands E. Padro´n, G. Melia´n, R. Marrero, D. Nolasco, J. Barrancos, G. Padilla, P. A. Herna´ndez, N. M. Pe´rez

115

SO2 Emission from Active Volcanoes Measured Simultaneously by COSPEC and mini-DOAS J. Barrancos, J. I. Rosello´, D. Calvo, E. Padro´n, G. Melia´n, P. A. Herna´ndez, N. M. Pe´rez, M. M. Milla´n, B. Galle

135

Underground Temperature Measurements as a Tool for Volcanic Activity Monitoring in the Island of Tenerife, Canary Islands A. Eff-Darwich, J. Coello, R. Vin˜as, V. Soler, M. C. Martin-Luis, I. Farrujia, M. L. Quesada, J. de la Nuez

147

Carbon Dioxide Discharged through the Las Can˜adas Aquifer, Tenerife, Canary Islands R. Marrero, D. L. Lo´pez, P. A. Herna´ndez, N. M. Pe´rez

173

List of Publications

Hiroshi Wakita was born in Nishinomiya (Hyogo, Japan) on September 29, 1936, and is actually an Emeritus Professor at The University of Tokyo (Japan). He graduated with B.Sc., M.Sc. and Ph.D. from Gakushuin University (Tokyo) in 1962, 1964 and 1968, respectively. He was a Researcher at the Japan Atomic Energy Research Institution, Tokyo (1964–68), Research Associate at Oregon State University, Corvallis, USA (1968– 71), Research Associate at the University of Tokyo (1971–77), Lecture at The University of Tokyo (1977–78), Associate Professor at The University of Tokyo (1978–86), Professor at The University of Tokyo (1986–97), and Professor at the Gakushuin Women’s College, Tokyo (1998–2007). He was also the Director of the Laboratory for Earthquake Chemistry at The University of Tokyo (1988–97), Associate Editor of Applied Geochemistry (1992–96), President of the Geochemical Society of Japan (1992–93), and Vice-president of the Geochemistry Research Association (1996). He received the Miyake Prize for his contributions on the field of geochemistry from the Geochemistry Research Association in 1989.

Pure appl. geophys. 165 (2008) 1–3 0033–4553/08/010001–3 DOI 10.1007/s00024-007-0295-3

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

Introduction

Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume III is a special publication to honor Professor Hiroshi Wakita for his scientific contributions to science, which have been closely linked with one of the major objectives of the 2008 International Year for the Earth Planet. Reducing natural risks in active tectonic and volcanic environments by searching for and detecting early warning signatures related to earthquakes and volcanic eruptions has been a major research goal for Hiroshi Wakita. The volume III consists of nine original papers written by researchers from Taiwan, Italy, Turkey, Iceland, USA, Sweden, India and Spain dealing with various aspects of the role of terrestrial fluids in earthquake and volcanic processes, which reflect Prof. Wakita’s wide scope of research interests. The volumes I and II consist of 17 and 10 original contributions which were published in Pure and Applied Geophysics on May 2006 and December 2007, respectively. These Pure and Applied Geophysics Hiroshi Wakita volumes should be useful for active researchers in the subject field, and graduate students who wish to become acquainted with them. Professor Wakita founded the Laboratory for Earthquake Chemistry in April 1978 with the aim of establishing a scientific base for earthquake prediction by means of geochemical studies, and served as its director from 1988 until his retirement from the university in 1997. He has made the laboratory a leading world center for the study of earthquakes and volcanic activities by means of geochemical and hydrological methods. Together with his research team and numerous foreign guest researchers whom he attracted, he has made many significant contributions in the above-mentioned scientific fields of interest. This achievement is a testimony for not only his scientific talent, but also his enthusiasm, his open-mindedness, and his drive in obtaining both human and financial support. The nine contributions of this volume III are arranged into two groups. The first group of five papers deals with the movement and signatures of terrestrial fluids related to earthquakes and active tectonic regions. The paper by CHIA et al. describes the observed changes of groundwater level induced by thrust faulting during the Mw 7.6, 1999 Chi-Chi earthquake recorded in 276 monitoring wells in Taiwan. Most of the observed coseismic falls appeared near the seismogenic fault as well as other active faults, while coseismic rises prevailed removed from the fault. The following paper by SU¨ER et al. presents the results of the 2002–2004 geochemical monitoring period of terrestrial fluids in geothermal fields located along an 800-km long E-W transect of the North Anatolian

2

N. M. Pe´rez et al.

Pure appl. geophys.,

Fault Zone (NAFZ) in Turkey, in order to both characterize the chemical nature of the individual fields and identify possible temporal variations associated with localized seismic activity. The paper by SKELTON et al. describes transient hydrogeochemical anomalies observed in a granite-hosted aquifer, which is located at a depth of 110-m, north of the Shillong Plateau, Assam, India. Their onsets preceded moderate earthquakes on December 9, 2004 (MW = 5.3) and February 15, 2005 (MW = 5.0), respectively, 206 and 213 km from the aquifer. The observation of these two hydrogeochemical events with the only two MW C 5 earthquakes in the study area argues in favor of causeand-effect seismic-hydrogeochemical coupling. The paper by EINARSSON et al. describes the observed radon changes in geothermal waters from drill holes related to an earthquake sequence at the transform plate boundary in South Iceland, which included two magnitude 6.5 earthquakes in June 2000. The authors emphasize that these radon anomalies were large and unusual if compared to a 17-year history of radon monitoring in this area. The paper by ITALIANO et al. provides field observations and new experimental data for the potential of the unexpected additional CO2 gas source production by mechanical energy applied to carbonate rocks in active tectonic regions beside mantlederived CO2 or CO2 produced by thermometamorphism. Data collected during the seismic crisis which struck the Central Apennines in 1997–1998 have shown an enhanced CO2 flux not associated with the presence of mantle or thermometamorphic-derived fluids. This earthquake-tectonic-related paper is then followed by four additional contributions dealing with observations related to volcanic processes. The paper by PADRo´N et al. describes the continuous monitoring of diffuse CO2 emission at El Hierro volcanic system, Canary Islands (Spain) and the observed geochemical anomalies before the occurrence of low magnitude seismic events in and around the volcanic island in 2004. The authors applied the material Failure Forecast Method (FFM) on the diffuse CO2 emission data to forecast successfully the first seismic event that took place in El Hierro in 2004. The following volcano-related paper by BARRANCOS et al. provides additional and recent SO2 emission data from eight active volcanoes: Santa Ana (El Salvador), San Cristo´bal and Masaya (Nicaragua), Arenal and Poa´s (Costa Rica), Tungurahua (Ecuador), Sierra Negra (Gala´pagos) and Etna (Italy). This paper also describes a comparison of SO2 emission measurements by COSPEC and mini-DOAS showing that most of the observed relative differences were lower than 10%. The paper by EFF-DARWICH et al. describes the spatial distribution of groundwater temperatures in Tenerife (Canary Islands) thanks to the vast network of *1.500 subhorizontal tunnels which provide most of the water resources for the island. Geological, hydrological and volcanological characteristics seem to be responsible for the actual groundwater temperature spatial distribution which has been characterized during a quiescent period, in order to detect changes in heat flow related to volcanic activity. The last volcano-related paper is also related to the groundwater system at Tenerife, Canary Island (Spain). The paper by MARRERO et al. provides an estimation of the water mass balance and the CO2 budget in Las Can˜adas’ aquifer; the largest aquifer on the island. The relatively high dissolved inorganic carbon

Vol. 165, 2008

Introduction

3

content in the groundwaters explains the ability of this aquifer to dissolve and transfer magmatic CO2, even during quiescence periods. The guest editorial team would like to thank all the contributors, and reviewers involved, who are listed below: R.M. Azzala, Werner Balderer, Alain Bernard, Emily Brodsky, Giorgio Capasso, Carlo Cardellini, Yeeping Chia, Antonio Eff-Darwich, Williams C. Evans, Cinzia Federico, Fausto Grassa, Jens Heinicke, Pedro A. Herna´ndez, David Hilton, George Igarashi, Kohei Kazahaya, Naoji Koizumi, Paolo Madonia, Rayco Marrero, Norio Matsumoto, Agnes Mazot, Eleazar Padro´n, Antonio Paonita, J. W. Rudnicki, Francesco Sortino, Jean-Paul Toutain, Nick Varley, Giuseppe Vilardo and Vivek Walia. Special thanks are due to Kenneth McGee, who served as co-guest editor in The Hiroshi Wakita volume I, for his support of this special issue, to Pedro A. Herna´ndez for his great assistance to the Guest-Editorial team, and to Renata Dmowska, without whose marvellous and tremendous support help the third special volume would not have been possible.

Nemesio M. Pe´rez Environmental Research Division Instituto Tecnolo´ gico y de Energı´as Renovables (ITER) Tenerife, Canary Islands Spain Sergio Gurrieri Istituto Nazionale di Geofisica e Vulcanologia, V. Ugo La Malfa 153 - 90146 Palermo Italy

Chi-Yu King Earthquake Prediction Research, Inc. 381 Hawthorne Ave. Los Altos, CA 94022 USA Yuri Taran Institute of Geophysics Universidad Nacional Auto´noma de Me´xico (UNAM) Mexico D.F 04510 Mexico

Pure appl. geophys. 165 (2008) 5-16

' Birkhauser Verlag, Basel, 2008

0033^553/08/010005-1 2

D , , , . ^ ^ » H Ar^r^li^r l

r-^^,^l.woi..o

’ ^^^^ ^" ^ Applied Geophysics

DOT 10.1007/S00024-007-0293-5

Spatial and Temporal Changes of Groundwater Leve l Induced by Thrust Faulting YEEPIN G CHIA / JESSIE J. CHIU,^ YI-HSUA N CHIANG / TSAI-PIN G L E E / and CHEN-WUM G LIU"^

Abstract Changes of groundwater level, ranging from a fall of 11.10 m to a rise of 7.42 m, induced by thrust faulting during the 1999 M ^ 7.6, Chi-Chi earthquake have been recorded in 276 monitoring wells in Taiwan. Most coseismic falls appeared near the seismogenic fault as well as other active faults, while coseismic rises prevailed away from the fault. Coseismic groundwater level rises and falls correlated fairly well with hypocentral distance in the vicinity of the thrust fault. W e found a major difference of coseismic changes in wells of different depths at most multiple-wel l stations. The recovery process of coseismic groundwater level changes is associated with the confining condition of the aquifer. Cross-formational flow is likely to play an important role in groundwater level changes after the earthquake. In the hanging wall of the thrust fault, an abnormal decline of groundwater level was observed immediately before the earthquake. The underlying mechanism of the unique preseismic change warrants further investigation. Key words: Groundwater, Chi-Chi earthquake. Thrust fault, Coseismic, Postseismic, Preseismic.

1. Introduction Water level in a well-confine d aquifer could be sensitive to crustal strain 1967). Field observations have shown a correlation between the estimated tectonic strains and the coseismic changes of well water level during the 1974 Izu-HantoOki earthquake (WAKITA , 1975). Variations of groundwater level in seismic regions have been used to monitor crustal deformation and to search for an earthquake precursor (BREDEHOEFT ,

(BAKU N and LINDH , 1985; KISSI N et al., 1996).

Coseismic groundwater level changes have been reported in many places around the world (MONTGOMER Y and MANGA , 2003), howeve r most changes are either sparsely distributed or concentrated in a few spots. Postseismic changes may reflect the subsurface flow in response to coseismic changes or permeability changes (ROELOFFS, 1998; WAN G et al., 2004). The observation of the flow process, particularly after small changes, is often impeded by pumping or other hydrologic factors. Preseismic changes are seldom

^ Department of Geosciences, National Taiwan University, Taipei 106, Taiwan. E-mail: [email protected] ^ Atomi c Energy Council, Yonghe , Taipei 234, Taiwan. ^ Department of Bioenvironmen t System s Engineering, National Taiwan University, Taipei 106, Taiwan.

6

Y . Chia et al.

Pure appl. geophys.,

reported (ROELOFFS and QUILTY , 1997; KOIZUM I and TSUKUDA , 1999), and the supporting evidence and underlying mechanism of their relations to fault deformation or earthquakes are not clear (KIN G et al, 2000). Whil e studies pertaining to the distribution and process of earthquake-related hydrologic phenomena are hampered by limited data, preliminary clues have been obtained by examining groundwater level changes recorded by a dense monitoring well network in the vicinity of a thrust fault ruptured during a large earthquake in Taiwan. Here we use monitoring records before, during, and after the earthquake to enhance our understanding of the spatial and temporal distribution as well as the possible mechanisms of groundwater level changes induced by thrust faulting.

2. Earthquake and Monitoring Wells On 21 Septembe r 1999, an earthquake of Mw 7.6 occurred near the town of Chi-Chi in central Taiwan at 1:47 a.m. local time. The hypocentral depth was estimated to be 10 km (SHIN et al., 2000). The best fitting focal mechanism has a nodal plane with a strike of 5 , a dip of 34 and a rake of 65 (CHAN G et al, 2000; KA O and CHEN, 2000). A s shown in Figure 1, widespread surface rupture resulted from thrusting along the Chelungpu fault extended approximately 100 km in the north-south direction (ANGELIE R et al, 2003). The hanging wall is on the east side of the thrust fault. Field investigations and GPS data indicated that the hanging wall move d as much as 10.1 m laterally and 8 m vertically. In contrast, up to 1.5-m lateral displacement and 0.26-m vertical displacement were observed in the footwall (Y u et al., 2001). In the coastal plain of Taiwan, the second generation network of monitoring wells had been installed since 1992 for improving groundwater resource management. A t the time of the Chi-Chi earthquake, 377 monitoring wells were operational in Taiwan. In the vicinity of the seismogenic fault, all wells, except one, are located in the footwall (west side) of the fault. Groundwater level is recorded by the digital data logger at one-hour interval. Som e wells equipped with the analog data logger also provide continuous records. Al l of the monitoring wells were screened in highly permeable sand or gravel layers. An y changes of groundwater level in the aquifer can, therefore, be quickly reflected by changes of water level in the monitoring well.

3. Spatial Distribution of Coseismic Groundwater Changes Coseismic changes of groundwater level in central western Taiwan due to the Chi-Chi earthquake have been discussed by CHIA et al. (2001) and WAN G et al. (2001). The mechanisms of these coseismic changes have been discussed by LEE et al. (2002), WAN G et al. (2003), KOIZUM I et al. (2004) and LA I et al. (2004). Tw o types of coseismic changes of water level were observed: oscillatory changes and persistent changes.

Vol . 165, 2008

Changes of Groundwater Leve l Induced by Faulting

Figure 1 Map of Taiwan showing the spatial distribution of 276 wells located at 126 monitoring stations where coseismic groundwater level rises and falls were observed during the M ^ 7.6 Chi-Chi earthquake on Septembe r 21, 1999.

Oscillatory coseismic changes, recorded on analog records, were the response of water column in the well and pore pressure in the aquifer to passing earthquake waves (COOPER et al., 1965; Liu et ai, 1989). The amplitude of oscillatory changes diminished shortly after the earthquake. Persistent coseismic changes have been proposed to be the

8

Y . Chia et al.

Pure appl. geophys.,

response of formation fluid pressure to shear strain due to the redistribution of stress field resulted from fault movement (MuiR-Woo d and KING , 1993; GE and STOVER , 2000). It is analogous to the response of pore pressure in soils to shear strain under an undrained triaxial test (WANG , 1997). Of the 377 monitoring wells, persistent coseismic groundwater level changes were observed in 276 wells during the Chi-Chi earthquake. Figure 1 illustrates spatial distribution of coseismic groundwater level rises and falls based on hourly records of 276 wells at 126 monitoring stations in the coastal plain of Taiwan. Of those, 203 wells at 80 stations are located within 50 km from the thrust fault. Included are a coseismic rise of 7.42 m and a coseismic fall of 11.09 m (Figs. 2a and 2b), the largest changes ever documented. It is noticed that, in the footwall of the seismogenic fault in central western Taiwan, coseismic falls were primarily observed at the stations near the ruptured segment. In contrast, coseismic rises prevailed at the stations away from the fault. A t some stations in the transition area, both coseismic rises and falls were observed in wells of different depths. Similar distribution patterns were also observed in the footwall of the unruptured segment of the thrust fault in southwestern Taiwan as well as another active fault in southern Taiwan. There is a general trend in the variation of the magnitude of coseismic water level change with hypocentral distance, as shown in Figure 3. The

Figure 2 The largest coseismic rise and coseismic fall of groundwater level observed during the earthquake. Hypocentral distances are denoted on top of each panel, (a) Hourly records (solid line) of the HW 2 well showing a coseismic rise of 7.42 m from 1 a.m. to 2 a.m., while analog records (dotted line) showing a rise of 8.5 m at 1:47 a.m. (b) Hourly records of the JS2 well showing a coseismic fall of 11.09 m.

Vol . 165, 2008

Changes of Groundwater Leve l Induced by Faulting

9

Figure 3 Coseismic groundwater level rise and fall versus hypocentral distance. Here coseismic changes in unconfined aquifers were excluded because they failed to represent the actual changes.

equation of the best-fit regression curve for the scatter plot of coseismic fall versus hypocentral distance up to 70 km is log h = 6.48^.4 4 log d, where h is the magnitude of the coseismic change in meters and d is the hypocentral distance in kilometers. The squared correlation coefficien t (R^ ) of 0.83 indicates a good correlation between coseismic fall and distance from the hypocenter in the vicinity of the thrust fault. The coseismic fall becomes considerably small beyond a hypocentral distance of 70 km . The equation of the best-fit regression curve for the plot of coseismic rise against hypocentral distance is log h = 4.09-2.39 log d, with a squared correlation coefficien t of 0.53. This suggests a moderate correlation between coseismic rise and hypocentral distance. A s shown in Figure 3, large rises were observed primarily at stations located between 30 km and 60 km from the hypocenter. Nevertheless, the magnitude of coseismic rise at these multiple-wel l stations changes drastically with depth. Whil e the rise diminishes gradually beyond 60 km , moderate rises were observed in six wells beyond 130 km . Al l of these six wells are located in northern Taiwan, implyin g that the redistributed stress induced by thrust faulting may concentrate in certain areas far from the seismogenic fault. Seismi c shaking of the saturated porous medium provides another possible way of explaining the occurrence of coseismic rise at a distance from the epicenter (MONTGOMER Y and MANGA , 2003).

4. Postseismic Changes of Groundwater Level The recovery of groundwater level in response to coseismic changes began immediately after the earthquake in most wells. The recovery process in an aquifer was controlled primarily by groundwater flow, and the recovery rate of coseismic change

10

Y . Chia et al.

Pure appl. geophys.,

varied with tlie confining condition of an aquifer. In a confined aquifer, tlie recovery process of coseismic change could last for several week s to a few months (Fig. 4a). The coseismic change in a confined aquifer, defined as the water level change from 1 a.m. to

Figure 4 Temporal changes of groundwater levels under various confining conditions after the earthquake, (a) Recovery of water level in the JL3 well approximately 5 months after a coseismic rise in a confined aquifer, (b) Recovery of water level in the TC2 well approximately 16 hours after a coseismic fall in a partially confined aquifer, (c) Recovery of water level in the SL l well within one hour after a coseismic rise in an unconfined aquifer, (d) Analog records at HS l showing recovery in approximately 15 minutes after coseismic rise in an unconfined aquifer.

Vol . 165, 2008

Changes of Groundwater Leve l Induced by Faulting

11

2 a.m. Septembe r 21 on hourly records, is close to the actual coseismic change at 1:47 a.m. on analog records (Fig. 2a). On the contrary, the recovery rate of groundwater level in an unconfined or partially confined aquifer is considerably faster. For instance, the TC station is located in a groundwater recharge area. The shallow well, TCI , was installed in an unconfined aquifer while the deep well, TC2, was installed in an aquifer confined partially by the overlying silty layer. The coseismic change in TC2 took 16 hours to recover to a steady level (Fig. 4b). In an unconfined aquifer, such as SLl , a pulse-like water level change is typically observed on hourly records, indicating the recovery process was completed within one hour after the coseismic change (Fig. 4c). On analog records, however, most actual coseismic changes at 1:47 a.m. in the unconfined aquifer were recovered within tens of minutes after the mainshock (Fig. 4d). Consequently, only a fractional or unnoticeable water level change could be observed at 2 a.m. on hourly records. A s the recovery rate of coseismic change depends on the confining condition of an aquifer, the variation pattern of a coseismic water level change and its recovery process could be a potential criterion for characterizing the degree of aquifer confinement . Most monitoring stations are composed of 2 to 5 wells screened at different depths, ranging from 14 m to 300 m. Records of these multiple-wel l stations revealed clues to the variation of coseismic change in the vertical direction as well as to the occurrence of cross-formational flow in the subsurface after the earthquake. They also provide field evidence for interpreting earthquake-related groundwater level anomalies. First, the magnitude of coseismic change fluctuated in wells at different depths. For instance, the coseismic rise at the Y L station, as shown in Figure 5a, varies from 6.55 m in YL l at the depth of 69 m to 1.2 m in YL 4 at 198 m. The DZ station recorded a coseismic fall of 9 cm at DZ2, but a coseismic rise of 28 cm and 14 cm at DZ l and DZ4, respectively (Fig. 5c). Such a difference in the vertical direction provides a possible explanation for difficultie s in matching coseismic groundwater level changes with volumetric strain changes estimated from fault displacement based on simple dislocation models (IGARASH I and WAKITA , 1991; QUILT Y and ROELOFFS, 1997; GRECKSC H et al, 1999). Second, postseismic anomalous changes of groundwater level at some multiple-wel l stations revealed the occurrence of cross-formational flow in the subsurface after the earthquake. Regional groundwater flow is considered a steady-state condition before the earthquake. The variation of coseismic changes at different locations and depths, however, initiated a transient flow in both lateral and vertical directions after the earthquake. For instance, the permeable layers tapped by YL 2 and YL 3 are hydraulically interconnected as evidenced in Figure 5a by the overlap of their groundwater levels before the earthquake. It is noted that the coseismic rise was 6.46 m in YL2 , but only 4.12 m in YL3 . The hydraulic gradient induced by different coseismic changes drove groundwater flow downward, resulting in a decline of water level in YL 2 and a rise in

12

Y . Chia et al.

Pure appl. geophys.,

Figure 5 Temporal change of groundwater levels in different depths at three multiple-wel l stations, (a) The Y L station. The coseismic rise fluctuates in wells of different depths. The water levels at YL 3 and YL 4 continued to rise for a few days after the earthquake, (b) The LY station. Afte r a coseismic fall of 4.62 m and a postseismic fall of 1.80 m, the water level at LY l approached that of LY2 . (c) The DZ station. The cosiesmic rise appeared in DZ2 while coseismic falls were observed in DZ l and DZ4. The coseismic change in DZ3 cannot be identified.

YL 3 until they came close again four days after the earthquake. The continued rise of groundwater level after a coseismic rise, similar to YL3 , was also observed in YL 4 and several other wells.

Vol . 165, 2008

Changes of Groundwater Leve l Induced by Faulting

13

The confining conditions of some aquifers near the surface rupture zone may have been changed by the earthquake. For instance, both LY l and LY2 , located 5 km from the ruptured segment, were artesian wells before the earthquake (Fig. 5b). Instead of a gradual rise after the coseismic fall, the groundwater levels declined further for several months after the earthquake. This phenomenon implies that subsurface fracturing may have been developed through the artesian aquifer during the earthquake. Crossformational flow could be generated when the confinemen t of the aquifer was breached, resulting in a rapid dissipation of pore water pressure after the earthquake (ROJSTACZE R et al, 1995; KIN G et al, 1999). Both wells were no longer artesian after the earthquake.

5. Preseismic Changes of Groundwater Level Several abnormal changes of groundwater level, observed from tens of minutes to a few months before the Chi-Chi earthquake, have been identified. Further examination indicated that these changes were likely to be associated with local pumping, rainfall, or improper data management, except a rapid decline of groundwater level immediately before the earthquake in the SL l well (Fig. 4c). The SL l well, located approximately 1.5 km east of the ruptured fault, is the only monitoring well installed in the hanging wall. It was placed in a shallow unconfined aquifer. The water level in SL l had not been interfered by pumping since the beginning of monitoring in 1997 until the end of 2000. A t the beginning of each recession (falling) stage during this period, the water level usually declined slowly with the rate of decline increasing gradually or remaining steady. The decline had never exceeded 3 cm during the first 3 hours in all recession stages, except for the one right before the Chi-Chi earthquake (Fig. 6). It is observed that, after rainfalls from Septembe r 15 to 19, 1999, the water level in SL l rose gradually on Septembe r 19 and 20, as shown in Figure 7, but declined 4 cm abruptly over a 3-hour period right before a 14-cm pulse-like coseismic rise. By

Figure 6 Water-level decline in the SL l well during the first 3 hours in recession stages from 1997 to 2000. The largest decline, 4 cm, appeared right before the Chi-Chi earthquake.

14

Y . Chia et al.

Pure appl. geophys.,

Figure 7 Ten-day records of groundwater level in the SL l well showing a 4-cm decline from 22:00 Septembe r 20 to 1:00 Septembe r 21 immediately before a 14-cm coseismic rise at 2:00 Septembe r 21 (also shown in Fig. 4c). The decline continued after the earthquake and the rate of decline decreased gradually when the pulse-like coseismic rise and its recovery were ignored.

ignoring coseismic change and its recovery, the rate of water level decline decreased gradually during the beginning of the recession starting from 22:00 Septembe r 20. The 4-cm decline of groundwater level immediately before the earthquake and the unique decreasing rate of decline at the beginning of the recession stage, therefore, becomes a unique hydrologic anomaly in SLl . Whethe r the decline is caused by dilatational deformation in the hanging wall right before the earthquake remains to be investigated.

6. Conclusions The comprehensive data recorded at monitoring well stations during the M^ 7.6 Chi-Chi earthquake provide a preliminary framework of regional distribution of coseismic groundwater level changes. The spatial distribution of observed coseismic change may reflect the redistribution of stress field in the shallow subsurface induced by the displacement of seismogenic fault. Such a complex distribution of coseismic changes is not expected to be consistent with volumetric strains calculated from a simple dislocation model. It is desirable to develop a more sophisticated model, taking into consideration aquifer characteristics and other geologic conditions, to describe coseismic changes induced by fault displacement. Postseismic groundwater level changes are essentially the subsurface hydrologic responses to various coseismic changes. Records from multiple-wel l stations revealed the importance of vertical confinemen t and cross-formational flow in the water level recovery after coseismic changes, providing a basis for interpreting anomalous changes of groundwater level after the earthquake. Moreover, the coseismic change and the postseismic recovery process could be considered a natural hydraulic testing for characterizing the confining condition or estimating hydraulic parameters of aquifers.

Vol . 165, 2008

Changes of Groundwater Leve l Induced by Faulting

15

A n anomalous decline of groundwater level was observed in the hanging wall immediately before the earthquake, although the underlying mechanism of the phenomenon is not clear. A s large earthquakes occur frequently in Taiwan and the monitoring well network continues to expand, more earthquake-related water level records will be available to facilitate further investigations.

Acknowledgements W e gratefully acknowledge access to monitoring records and hydrogeologic infor› mation of the Water Resources Agenc y and the Central Geological Survey of Taiwan. This work is supported by the National Science Council of Taiwan (NSC-94-2116-M002 011) and the Water Resources Agenc y (MOEAWRA0950187) .

REFERENCE S

ANGELIER , J., LEE, J. C , CHU, H . T. , and Hu, J. C. (2003), Reconstruction of fault slip of the September 21st, 1999, Taiwan Earthquake in the asphalted surface of a car park, and co-seismic slip partitioning, J. Struct. Geol. 25, 345-350. BAKU N W . H . and LINDH , A . G. (1985), The Parkfield, California Earthquake prediction experiment, Science 229, 619-624. BREDEHOEFT , J. D. (1967), Response of well-aquifer systems to earth tides, J. Geophys. Res. 72, 3075-3087. CHAN G C.H., Wu , Y.M. , SHIN , T . C , and WAN G C.Y . (2000), Relocation of the 1999 Chi-Chi earthquake in Taiwan, Terrestrial, Atmos . and Ocean. Sci. 11, 581-590. CHIA , Y.P. , WANG , Y.S. , CHIU, J.J., and Liu, C.W . (2001), Changes of groundwater level due to the 1999 Chi-Chi earthquake in the Choshui River alluvial fan in Taiwan, Bull. Seismol . Soc. Am . 91, 1062-1068. COOPER, H . H., Jr., BREDEHOEFT, J. D., PAPADOPULOS , L S. , and BENNETT , R . R . (1965), The response of well-

aquifer systems to seismic waves, J. Geophys. Res. 70, 3915-3926. GE, S . and STOVER , S . C . (2000), Hydrodynamic response to strike- and dip-slip faulting in a half-space, J. Geophys. Res. 105, 25513-25524. GRECKSCH , G., ROTH , F., and KUMPEL , H.J. (1999), Co-seismic well-level changes due to the 1992 Roermond earthquake compared to static deformation of half-space solutions, Geophys. J. Int. 138, 470-478. IGARASHI , G . and WAKITA , H . (1991), Tidal responses and earthquake-related changed in the water level of deep wells, J. Geophys. Res. 96, 4269^278 . KAO , H . and CHEN, W . P . (2000), The Chi-Chi earthquake sequence: Active, out-of-sequence thrust faulting in Taiwan, Science 288, 2346-2349. KING , C.-Y. , AZUMA , S. , IGARASHI , G., OHNO, M . , SAITO , H., and WAKITA , H . (1999), Earthquake related water-

level changes at 16 closely clustered wells in Tono, Central Japan, J. Geophys. Res. 104, 13073-13082. KING , C.-Y. , AZUMA , S., OHNO, M . , ASAI , Y. , HE, P., KTTAGAWA , Y. , IGARASHI , G., and WAKITA , H . (2000), In

search of earthquake precursors in the water-level data of 16 closely clustered wells at Tono, Japan, Geophys. J. Int. 143, 469^77 . KissiN, I.G., BELIKOV , V.M. , and ISHANKULIEV , G.A . (1996), Short-term groundwater level variations in a seismic region as an indicator of the geodynamic regime, Tectonophysic s 265, 313-326. KOIZUMI , N . and TSUKUDA , E . (1999), Preseismic changes in groundwater level and volumetric strain associated with earthquake swarms off the east coast of the hu Peninsula, Japan, Geophys. Res. Lett. 26, 3509-3512. KOIZUMI , N . , LAI , W . - C , KTTAGAWA , Y. , and MATSUMOTO , N . (2004), Comment on "Coseismic hydrological changes associated with dislocation of the September 21, 1999 Chi-Chi earthquake, Taiwan" by Min LEE et al, Geophys. Res. Lett., 31, L13603, doi: 10.1029/2004GL019897.

16

Y . Chia et al.

Pure appl. geophys.,

LAI , W . - C , KOIZUMI , N . , MATSUMOTO , N . , KITAGAWA , Y. , LIN , C.-W. , SHIEH, C.-L. , and LEE, Y . - P . (2004), Ejfects

of seismic ground motion and geological setting on the coseismic groundwater level changes caused by the 1999 Chi-Chi earthquake, Taiwan, Earth Planets Space 56, 873-880. LEE, M. , LIU, T.K. , MA , K.F., and CHANG , Y.M . (2002), Coseismic hydrological changes with dislocation of the September 21, 1999 Chi-Chi earthquake, Taiwan, Geophys. Res. Lett. 29, doi: 10.1029/2002GL015116. LIU, L.B. , ROELOFFS, E., and ZHENG , X . Y . (1989), Seismically-induced water level fluctuations in the Wall well, Beijing, China, J. Geophys. Res. 94, 9453-9462. MONTGOMERY , D . R . and MANGA , M . (2003), Streamflow and water well responses to earthquakes. Science 300, 2047-2049. MuiR-WooD , R. and KING , G . C . P. (1993), Hydrologic signatures of earthquake strain, J. Geophys. Res. 98, 22035-22068. QuiLTY , E.G. and ROELOFFS, E.A . (1997), Water-level changes in response to the 20 December 1994 earthquake near Parkfield, California, Bull. Seismol . Soc. Am . 87, 310-317. ROELOFFS, E.A . (1998), Persistent water level changes in a well near Parkfield, California, due to local and distant earthquake, J. Geophys. Res. 103, 869-889. ROELOFFS, E.A . and Quilty, E. (1997), Water level and strain changes preceding and following the August 4, 1985 Kettleman Hills, California, Earthquake, Pure Appl . Geophys. 149, 21-60. RojSTACZER , S., WOLF , S. , and MICHEL , R . (1995), Permeability enhancement in the shallow crust as a cause of earthquake-induced hydrologic changes. Nature 373, 237-239. SHIN , T.C., Kuo, K.W. , LEE, W.H.K. , TENG , T.L. , and TSAI , Y . B . (2000), A preliminary report on the 1999 Chi-Chi (Taiwan) earthquake, Seismol . Res. Lett. 71, 23-29. WAKITA , H . (1975), Water wells as possible indicators of tectonic strain. Science 189, 553-555. WANG , C . Y. , CHENG, L . H., CHIN, C. V. , and Yu , S. B . (2001), Coseismic hydrologic response of an alluvial fan to the 1999 Chi-Chi earthquake, Taiwan, Geology 29, 831-834. WANG , C.Y. , DREGER , D.S., WANG , C.H., MAYERI , D., and BERRYMAN , J. G. (2003), Field relations among

coseismic ground motion, water level change and liquefaction for the 1999 Chi-Chi (M^ = 7.5) earthquake, Taiwan, Geophys. Res. Lett. 30, doi: 10.1029/2003GL017601. WANG , C . Y. , WANG , C . H., and Kuo, C. H. (2004), Temporal change in groundwater level following the 1999 (M^ = 7.5) Chi-Chi earthquake, Taiwan, Geofluids 4, 210-220. WANG , H . F. (1997), Effects of deviatoric stress on undrained pore pressure response to fault slip, J. Geophys. Res. 102, 17943-17950. Yu , S.B. , Kuo, L.C., Hsu, Y.J. , Su, H.H., Liu, C.C, Hou, C.S., LEE, J.F., LAI , T . C , LIU, C . C , LIU, C.L., TSENG ,

T.F., TsAi , C. S., and SHIN , T.C. (2001), Preseismic deformation and coseismic displacements associated with the 1999 Chi-Chi, Taiwan earthquake. Bull. Seismol . Soc. Am . 91, 995-1012. (Receive d June 1, 2007, revised Novembe r 12, 2007, accepted Novembe r 17, 2007) PubHshed Online First: February 1, 2008

To access this journal online: www.birkhauser.ch/pageoph

Pure appl. geophys. 165 (2008) 17–43 0033–4553/08/010017–27 DOI 10.1007/s00024-007-0294-4

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

Geochemical Monitoring of Geothermal Waters (2002–2004) along the North Anatolian Fault Zone, Turkey: Spatial and Temporal Variations and Relationship to Seismic Activity SELIN SU¨ER,1 NILGU¨N GU¨LEC¸,1 HALIM MUTLU,2 DAVID R. HILTON,3 CANDAN C¸IFTER,4 and MESUT SAYIN4

Abstract—A total of nine geothermal fields located along an 800-km long E-W transect of the North Anatolian Fault Zone (NAFZ), Turkey were monitored for three years (2002–2004 inclusive; 3-sampling periods per year) to investigate any possible relationship between seismic activity and temporal variations in the chemistry and isotope characteristics of waters in the fields. The geothermal fields monitored in the study were, from west to east, Yalova, Efteni, Bolu, Mudurnu, Seben, Kurs¸ unlu-C ¸ ankırı, Hamamo¨zu¨, Go¨zlek and Res¸ adiye. The chemical (major anion-cation contents) and isotopic (18O/16O, D/H, 3H) compositions of hot and cold waters of the geothermal sites were determined in order to both characterize the chemical nature of the individual fields and identify possible temporal variations associated with localized seismic activity. The geothermal waters associated with the NAFZ are dominantly Na-HCO3, whereas the cold waters are of the Ca-HCO3 type. The oxygen- and hydrogen-isotope compositions reveal that the hot waters are meteoric in origin as are their cold water counterparts. However, the lower d18O, dD and 3H contents of the hot waters point to the fact that they are older than the cold waters, and that their host aquifers are recharged from higher altitudes with virtually no input from recent (post-bomb) precipitation. Although no major earthquakes (e.g., with M C 5) were recorded along the NAFZ during the course of the monitoring period, variations in the chemical and isotopic compositions of some waters were observed. Indeed, the timing of the chemical/isotopic changes seems to correlate with the occurrence of seismic activity of moderate magnitude (3 < M < 5) close to the sampling sites. In this respect, Cl, 3H and Ca seem to be the most sensitive tracers of seismically-induced crustal perturbations, and the Yalova and Efteni fields appear to be the key localities where the effects of seismic activity on the geothermal fluids are most pronounced over the monitoring period. The present study has produced a ‘baseline’ database for future studies directed at characterizing the effects of moderate-major earthquakes on the composition of geothermal waters along the NAFZ. Future work involving longer monitoring periods with more frequent sampling intervals should lead to a better understanding of the underlying mechanism(s) producing the observed chemical and isotopic variations. Key words: North Anatolian Fault Zone, geothermal fluid, seismic activity, geochemical monitoring, Turkey. 1 Department of Geological Engineering, Middle East Technical University, 06531 Ankara, Turkey. E-mail: [email protected]; [email protected] 2 Department of Geological Engineering, Eskis¸ ehir Osmangazi University, 26480 Eskis¸ ehir, Turkey E-mail: [email protected] 3 Fluids and Volatiles Lab., Geosciences Research Division, Scripps Inst. of Oceanography, UCSD, La Jolla, CA 92093-0244, USA. E-mail: [email protected] 4 Department of Technical Research and Quality Control, General Directorate of Turkish State Hydraulic Works, TR-06100 Ankara, Turkey. E-mail: [email protected]; [email protected]

18

S. Su¨er et al.

Pure appl. geophys.,

1. Introduction In seismically-active areas, periodic or continuous monitoring of the chemistry of terrestrial fluids is an approach increasingly utilized for understanding both the mechanisms inducing earthquakes and the associated response in the affected region of the crust. Variations in the chemical and isotopic composition of terrestrial fluids are considered to reflect sub-surface physical and chemical processes, such as fluid mixing, micro-fracturing and associated permeability changes (KING, 1986; THOMAS, 1988; ZHANG, 1994; IGARASHI and WAKITA, 1995; SILVER and WAKITA, 1996; WAKITA, 1996;TOUTAIN and BAUBRON, 1999; KING and IGARASHI, 2002; KING et al., 2006). A number of geochemical tracers have been used to investigate such subsurface processes, with both dissolved and free-gas phase samples showing great sensitivity in recording responses associated with seismic activity. For example, monitoring studies of terrestrial fluids in earthquake-prone regions have revealed significant changes in 222Rn (HAUKSSON, 1981; TENG and SUN, 1986; WAKITA et al., 1989; VIRK and SINGH, 1993; IGARASHI et al., 1995; PE´REZ et al., 1998; DAS et al., 2005; WALIA et al., 2006) and H2 emissions (WAKITA et al., 1980; SATO et al., 1986), variations in helium and carbon isotope ratios (3He/4He and 13C/12C) (SANO et al., 1986; 1998; SOREY et al., 1993; ITALIANO and MARTINELLI, 2001; BRA¨UER et al., 2003), fluctuations of diffuse CO2 emission (SALAZAR et al., 2002; PE´REZ et al., 2003; PE´REZ and HERNA´NDEZ, 2007; PADRO´N et al., this volume), as well as changes in gas ratios, such as He/Ar, He/222Rn, N2/Ar, CH4/Ar and CO2/He (SUGISAKI, 1978; KAWABE, 1984; SUGISAKI and SUGIURA, 1986; SUGISAKI et al., 1996; HILTON, 1996; VIRK et al., 2001). In addition to the variations observed by targeting gases, significant variations have also been recorded (either as pre- or post-seismic signals) in the chemical and isotopic composition of water phase samples. Variations in groundwater level/discharge and hydrogeochemistry were observed related to the 1995 Kobe earthquake (M 7.2), and were attributed to mixing of waters with different chloride and tritium contents (TSUNOGAI and WAKITA, 1995; KING et al., 1995; SANO et al., 1998). TOUTAIN et al. (1997) recorded a *36% increase (above background values) in the Cl content of a mineral water prior to a Pyrenean earthquake (M 5.2) in France. PE´REZ et al. (1998) also registered a *10% increase in the Cl content of a mineral water prior to earthquakes (M 4.2) in the NW Iberian Peninsula. NISHIZAWA et al. (1998) reported a two-fold increase in both Cl and SO4 contents in Yugano hot spring waters, a few days after the onset of the 1995 seismic swarm (M < 3 to 4.8) in the Izu Peninsula, central Japan. Likewise, SONG et al. (2006) detected increases in the Cl and SO4 contents (up to 138% and 278% of mean values, respectively) in hot and artesian springs of the Kuantzeling area (west-central Taiwan) prior to a September 1999 earthquake (M 7.3). FAVARA et al. (2001) and ITALIANO et al. (2004) observed temporal variations in the chemical composition and temperature of thermal waters and dissolved gases in the Umbria region. CLAESSON et al. (2004) reported short-term preseismic anomalies of Cu, Mn, Zn, Fe and Cr in the groundwater prior to a magnitude 5.8 earthquake in the Tjo¨rnes

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

19

fracture zone, north Iceland. All these changes were attributed to subsurface mixing processes induced by stress/strain changes associated with localized earthquake activity. In terms of the isotope systematics of the water phase, measurements of dD and d18O concentrations of groundwaters in seismically-active regions are potentially useful for searching earthquake precursors and in elucidating mechanisms operative during earthquakes. O’NEIL and KING (1981) observed an increase in the dD content of groundwaters while the d18O content remained constant during a magnitude 5.7 earthquake at the Oroville Dam in 1975 and a series of events near SAN JUAN BAUTISTA in 1980. This increase suggested that H2O may have either decomposed or reacted to form molecular H2 at depth. The study of BOLOGNESI (2000) reported a large increase in d18O values in geothermal reservoir waters at Vulcano Island, Italy, following seismic activity that occurred near the island. The most prominent increase in d18O was a change from +1.0 ± 0.5% to 3.4 ± 0.5% after a major earthquake (M: 5.5) in April 1978. The response was attributed to earthquake-induced increases in the contribution of hightemperature, d18O-rich magmatic condensates to the geothermal reservoir. Stable isotope variations in groundwaters were also observed before and after the magnitude 5.8 earthquake in the Tjo¨rnes fracture zone, north Iceland, suggesting stress-induced source mixing and leakage of fluid from an external (hotter) basalt-hosted source reservoir, where fluid-rock interaction was more rapid (CLAESSON et al., 2004). In relation to the present work, BALDERER et al. (2002) reported results of sampling campaigns before and after the devastating 1999 earthquakes along the North Anatolian Fault Zone, and pointed out variations in both chemical (Ca, K, Na, NO3, SO4 and Cl) and isotopic (d18O and dD) compositions of the thermal and mineral waters in the Kuzuluk, Bursa and Yalova/Gemlik areas. They attributed the variations in chemical compositions to the mobilization of deep-seated brines in response to the seismic activity and their mixing with waters of surficial origin. On the other hand, variations in d18O and dD values were interpreted as resulting from isotopic exchange with CO2 and H2S gases, respectively, and/or mixing with groundwaters of shallow and/or deep origin with differing recharge conditions. In this study we extend the work of BALDERER et al. (2002) by targeting a total of nine geothermal localities located along the North Anatolian Fault Zone (NAFZ) in an attempt to identify anomalies in the chemical and isotopic composition of fluids which may be related to regional/localized seismic activity. The present approach involves a 3-year monitoring program with 3 sampling periods per year covering 2002 (March-JulyOctober), 2003 (April-July-October) and 2004 (April-June-October). The geothermal fields studied in this project are located along an 800-km long transect of the NAFZ and are, from west to east, Yalova, Efteni, Bolu, Mudurnu, Seben, Kurs¸ unlu-C¸ankırı, Hamamo¨zu¨, Go¨zlek and Res¸ adiye (Fig. 1). Both hot and cold waters were utilized and analyzed for their chemical (major anion-cation contents) and isotopic (18O/16O, D/H, and 3H) compositions.

20

S. Su¨er et al.

Pure appl. geophys.,

Figure 1 Locations of the sampled geothermal fields and the recent seismic activities along the NAFZ (after C ¸ EMEN et al., 2000).

2. NAFZ: Tectonic Setting and Recent Seismicity The North Anatolian Fault Zone (NAFZ), comprising one of the major geologic structures in Turkey, developed during the Neotectonic period in response to intra-continental convergence following the Late Miocene collision of the Arabian promontory with Eurasia (MCKENZIE, 1972; DEWEY and S¸ ENGo¨R, 1979; S¸ENGo¨R et al., 1985; BARKA, 1992). This fault zone is a 1500-km long, few hundred meters to 40 km wide, dextral strike-slip fault system which forms an intracontinental transform boundary between the Eurasian Plate in the north and the Anatolian Plate in the south (KOc¸YIg˘IT et al., 1999a, b). It consists of shorter subparallel fault strands locally displaying an anastomosing fault pattern along much of its length. The NAFZ extends from eastern Anatolia in the east to the Aegean Sea in the west. The fault zone bifurcates into two major strands just east of the Sea of Marmara. The northern strand traverses, and the southern strand bounds, the southern margin of the Sea of Marmara (Fig. 1). The age of dextral motion and the total offset along the fault zone are controversial, covering the range of Middle Miocene - Early Pliocene and 20–85 km, respectively (see BOZKURT, 2001 for a review). The other structural members of the Neotectonic period that

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

21

developed along with the NAFZ are the Bitlis Suture Zone (BSZ), East Anatolian Fault Zone (EAFZ) and the Western Anatolian Graben System (WAGS) (Fig. 1). The NAFZ is known worldwide as one of the great strike-slip fault zones that host many medium to large-scale earthquakes. The most recent destructive effects of the _ NAFZ were experienced in 1999 in Izmit (M 7.4; 17 August, 1999) and Du¨zce (M 7.2; 12 November, 1999) provinces, resulting in more than 25,000 fatalities in the region. The August and November earthquakes occurred, respectively, at depths of 11 km (with two successive shocks in 45 seconds at the centers of Go¨lcu¨k and Arifiye) and 10 km (epicenter Du¨zce), creating a total length of surface rupture of 177 km from Go¨lcu¨k eastwards (KOc¸YIG˘IT et al., 1999a, b; C ¸ EMEN et al., 2000) (Fig. 1). During the monitoring period of the present study (2002–2004 inclusive) the magnitudes of the earthquakes that occurred along the NAFZ were all less than 5.0, with the seismic activity mainly concentrated in the western and central segments of the fault zone (Fig. 1). The most recent earthquakes recorded along the NAFZ during the monitoring period with magnitudes greater than 3.0 include: the Sea of Marmara (M 4.7—23 March, 2002), Armutlu-Yalova (M 3.1—3 July, 2002 and M 3.1—13 July, 2002), Yıg˘ılca-Du¨zce (M 3.1—14 July, 2002, M 3.1—11 October, 2002, M 4.0—25 July, 2003, M 3.6—15 June, 2004, M 4.8—2 July, 2004), Kaynas¸ lı-Du¨zce (M 3.2—1 July, 2003 and M (3.1 and 4.0)—25 July, 2003), Mudurnu (M 3.4—1 November, 2002), C ¸ ınarcık-Yalova (M 3.5—22 July, 2003), and Bolu (M 3.5—7 August, 2003, M 4.7—14 April, 2004, M 4.6—23 June, 2004).

3. Regional Geology and Hydrogeologic Outline The basement rocks in the geothermal areas of this study are represented by Paleozoic metamorphics (schists and marbles). These units are unconformably overlain by Upper Jurassic-Lower Cretaceous limestones and Upper Cretaceous flysch consisting of intercalations of limestone, conglomerate, marl, sandstone, claystone and siltstone. Products of widespread volcanic activity, extending from Late Cretaceous to Miocene and consisting of basaltic-andesitic lava flows, tuffs and agglomerates, are observed either intercalated with, or overlying, the Upper Cretaceous flysch. In turn, these units are overlain by Neogene clastics and lacustrine limestones (S¸ AHINCI, 1970; CANIK, 1972; ¨ ZCAN and U ¨ NAY, 1978; MU¨FTu¨OG˘LU and AKıNCı, 1989; ERZENOG˘LU, 1989; KOC¸AK, 1974; O MTA, 1996). Plio-Quaternary fluviatile deposits form the youngest units of the succession and are accompanied, in the eastern-central segment of the NAFZ, by Plio-Quaternary volcanics of limited areal extent (TATAR et al., 1996). Numerous hot-springs emerge at geothermal fields located along the fault and associated fracture zones. The temperature of the hot-springs ranges between 30C and 74C, with flow rates between 0.1 and 5.6 l/s. Except for the Yalova and Kurs¸ unlu fields, where the primary reservoirs are comprised of Tertiary volcanics, the geothermal reservoir rocks are mainly

S. Su¨er et al.

22

Pure appl. geophys.,

Mesozoic limestones (KOC¸AK, 1974; MTA, 1996). Impervious clayey levels of the flysch facies and the Neogene sequence frequently act as cap rocks to the geothermal systems.

4. Sampling and Analytical Techniques Sampling was performed over a total of nine periods, comprising three discrete sampling intervals per year (March-July-October 2002, April-July-October 2003 and April-June-October 2004). This schedule was maintained for all geothermal fields (Yalova, Efteni, Mudurnu, Seben, Bolu, Kurs¸ unlu, Hamamo¨zu¨, Go¨zlek) with the exception of Res¸ adiye (where sampling ceased in October 2002). Both natural springs and production wells were utilized during sampling (Table 1). Cold water sampling commenced at most locations in 2003 in order to better evaluate possible proximal subsurface processes, such as mixing and boiling, which may contribute to compositional variations. The Res¸ adiye field could be sampled only in the year 2002 since the hotspring source dried-up due to excessive usage for balneological purposes. Furthermore, in April 2004, after sampling in Efteni and Bolu fields on 12 and 13 April, respectively, a second sampling campaign was mounted for both fields following the seismic activity of 14 April at Bolu Town Center (M 4.7). Samples were collected for major anion-cation contents, stable isotope ratios (18O/16O and D/H) and tritium (3H) contents using polyethylene bottles. The anioncation chemistry, tritium, 18O/16O and D/H analyses of water samples were carried out Table 1 Sample numbers, sample types and geographic coordinates of the geothermal fields Locality

Sample no.

Latitude

Efteni

1a-hot 1b-cold 2a-hot 2c-cold 3a-hot 3b-cold 4a-hot 4d-cold 5a-hot 5d-cold 6a-hot 6b-cold 7a-hot 7b-cold 8a-hot 8b-cold 9a-hot 9b-mineral 9c-cold

40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40

Yalova Bolu Mudurnu Seben Hamamo¨zu¨ Go¨zlek Res¸ adiye Kurs¸ unlu

a

450 450 360 370 410 410 270 270 190 190 460 460 330 330 230 230 490 490 500

4200 4200 1600 1400 1000 0500 3300 3500 3100 3100 5900 5900 0800 0800 3100 3100 3400 5700 0400

a

Longitude N N N N N N N N N N N N N N N N N N N

031 031 029 029 031 031 031 031 031 031 035 035 035 035 037 037 033 033 033

010 010 100 130 370 370 140 140 320 320 010 010 400 400 200 200 100 100 100

4100 4100 1200 6800 1600 0500 2700 2600 2400 2400 2400 2400 3600 3600 1500 1500 3200 4300 4300

E E E E E E E E E E E E E E E E E E E

Sample type

Spring Spring Spring Spring Production well Spring Production well Spring Spring Spring Production well Spring Production well Spring Spring Spring Production well Mineral spring Spring

Numbers given beside production wells represent production depths, A: Artesian, P: Pumping.

(P)-83 m (A)-80 m

(A) (A)-800 m

(A)-165 m

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

23

at the laboratories of the Turkish State Hydraulic Works (DSI). Anion-cation analyses were performed by conventional methods (titration for Ca, Mg, HCO3, CO3 and Cl, flame photometry for Na and K, and spectrophotometry for SO4). Tritium contents were determined by a liquid scintillation counting system (Packard Tri Carb 2260 XL) after electrolytic enrichment (ALTAY and C ¸ IFTER, 1996). d18O and dD ratios were measured on a Micromass 602C mass spectrometer, using the standard water equilibration (EPSTEIN and MAYEDA, 1953) and zinc reduction (COLEMAN et al., 1982) methods, respectively.

5. Results The results of the chemical and isotopic analyses, as well as the in-situ temperature, pH and EC measurements, are given for all sampling periods in Table 2. This dataset includes measurements on both the hot and cold water samples. As seen in Table 2, temperatures range between 37.4–72.6C for the hot waters, and 6.7–20.5C for the cold waters. The hot waters are slightly acidic to slightly alkaline in character, with pH values ranging between 5.9 and 8.0. In contrast, the cold waters are relatively more alkaline, with pH values from 6.5 to 8.8. The TDS values (Total Dissolved Solid content) are much higher for the hot waters (ranging up to 11181 mg/l) compared to those for the cold waters (ranging up to 996 mg/l). Among all samples, Kurs¸ unlu 9a hot water (collected from a production well) has the highest TDS value which may be due, in part at least, to steam phase separation which has resulted in calcite scaling in the well. Regarding the stable isotope compositions (expressed with respect to V-SMOW), the d18O values of the hot waters range between -13.4% and -8.4%, while the cold waters have values ranging from -12.7% to -8.1%. The dD values, on the other hand, range between -98.3% and -64.2% for the hot, and between -86.1% and -54.7% for the cold waters. Tritium (3H) contents of the hot waters lie between 0–12.55 TU, with the majority being less than 5 TU. For the cold waters, the range is 3.00–15.70 TU, although most of the samples have concentrations above 8 TU. 5.1. Hydrogeochemical Facies The hydrogeochemical facies of the waters, defined by the dominant cation-anion pairs, are represented as pie diagrams in Figure 2. As observed, the hot waters are mostly Na-HCO3 type waters (Efteni, Seben, Go¨zlek, Res¸ adiye and Kurs¸ unlu samples), except for the Ca-HCO3 type at Bolu and Mudurnu, and the Na-SO4 type at Yalova (Fig. 2). The only geothermal field which displays a mixed-character water is Hamamo¨zu¨ since none of its individual ions exceeds 50% of the total composition. The cold waters at this locality, on the other hand, display Ca and/or Mg-HCO3 character.

S. Su¨er et al.

24

Pure appl. geophys.,

Table 2 Results of chemical analyses Locality

T

Efteni-1a 23/3/02 42.3 8/7/02 41.4 27/10/02 43.4 07/4/03 43.0 09/7/03 43.1 15/10/03 43.3 12/4/04 43.6 14/4/04 43.4 28/6/04 44.7 11/10/04 44.4 average 43.2 r 1.0 VC% 2.3 Efteni-1b 07/4/03 11.2 09/7/03 20.0 15/10/03 12.6 12/4/04 18.5 28/6/04 19.8 11/10/04 20.0 average 17.0 r 4.0 VC% 23.7 Yalova-2a 24/3/02 60.1 09/7/02 60.6 28/10/02 56.9 08/4/03 64.3 10/7/03 65.0 16/10/03 63.4 13/4/04 65.7 28/6/04 53.8 12/10/04 60.6 average 61.2 r 3.9 VC% 6.4 Yalova-2c 08/4/03 13.5 10/7/03 20.0 16/10/03 18.2 13/4/04 14.4 28/6/04 17.1 12/10/04 20.2 average 17.2 r 2.8 VC% 16.3 Bolu-3a 24/3/02 41.6 09/7/02 41.6

pH

EC

HCO3

Cl

SO4

Na

K

Ca

Mg

TDS

d18O

6.5 7.4 7.7 6.2 6.3 6.4 6.2 6.2 6.3 6.3 6.5 0.5 8.0

3090 3020 3050 3070 3010 3080 3053 38 1

1835 1793 1702 1704 1765 1580 1787 1756 1608 1560 1709 96 6

197 209 206 201 196 198 202 206 195 192 200 5 3

1 1 4 5 5 5 4 4 4 2 3 2 44

450 520 298 308 347 232 339 329 302 263 339 86 25

18 18 13 13 12 11 12 12 12 13 13 3 19

164 104 141 219 63 86 170 168 53 223 139 61 44

135 148 163 109 194 206 140 143 193 105 154 35 23

2800 2793 2526 2557 2582 2318 2654 2617 2365 2358 2557 171 7

-11.1 -11.2 -11.3 -11.7 -10.8 -11.6 -11.5 -11.0 -11.8 -10.5 -11.3 0.4 3.7

-83.1 0.00 ± -83.5 0.05 ± -80.0 2.20 ± -81.8 0.00 ± -81.9 4.85 ± -79.5 4.55 ± -86.0 0.00 ± -87.5 4.00 ± -84.7 5.80 ± -81.5 0.90 ± -82.9 2.24 2.6 2.35 3.1 105.00

1.65 1.55 1.70 1.60 1.95 1.90 2.05 1.90 1.90 1.85

8.8 8.8 8.6 8.5 8.3 8.5 8.6 0.2 2.4

195 208 225 253 209 229 22 10

134 145 120 133 13 10

9 4 8 7 3 39

7 6 8 7 1 20

2 0 2 0 1 0 2 0 1 0 33 33

14 11 7 11 4 33

23 23 23 23 0 1

190 191 168 183 13 7

-11.1 -12.1 -11.7 -11.5 -11.6 -10.8 -11.5 0.4 3.9

-80.9 -78.6 -80.8 -83.4 -78.1 -78.8 -80.1 2.0 2.5

3.00 ± 4.90 ± 5.65 ± 6.75 ± 4.50 ± 4.45 ± 4.88 1.26 25.87

1.60 2.00 1.90 2.20 1.90 1.95

7.9 7.8 7.7 7.4 7.6 7.7 7.3 7.8 7.5 7.6 0.2 2.6

1922 1902 1913 1916 1923 1912 1917 6 0

48 28 31 43 43 81 43 49 39 45 15 34

101 69 96 97 97 95 104 94 94 94 10 11

875 950 816 750 764 820 770 810 824 820 62 8

280 6 330 6 241 5 243 5 209 4 290 4 269 4 299 4 199 5 262 5 43 1 16 18

151 2 154 0 162 16 157 7 199 2 154 7 158 0 142 0 162 38 160 8 16 12 10 152

1463 1537 1365 1301 1316 1451 1349 1398 1361 1393 77 6

-11.3 -11.4 -11.7 -11.1 -11.7 -11.2 -11.6 -11.4 -10.5 -11.3 0.4 3.2

-75.0 8.90 ± -75.4 12.55 ± -72.2 3.30 ± -75.0 3.70 ± -77.3 0.05 ± -73.8 0.00 ± -80.3 1.10 ± -77.1 0.00 ± -76.2 1.90 ± -75.8 3.50 2.3 4.41 3.0 126.11

1.90 2.00 1.75 1.70 1.80 0.60 2.05 1.90 1.90

7.4 7.6 7.8 7.5 7.1 7.6 7.5 0.3 3.4

600 557 574 572 606 584 19 3

332 298 320 300 322 262 306 25 8

14 16 15 14 14 14 14 1 5

38 28 30 34 29 86 41 23 55

15 2 11 1 37 2 13 2 12 1 16 2 17 2 10 0 57 26

513 461 488 467 488 497 486 19 4

-8.8 -8.8 -9.0 -9.6 -9.8 -8.4 -9.1 0.5 5.8

-60.4 -54.7 -57.3 -61.0 -61.8 -59.1 -59.0 2.7 4.5

1.80 2.10 0.90 2.35 1.85 1.85

7.0 6.2

-

817 512

7 625 5 675

105 102 75 100 106 108 99 12 13

8 5 10 4 4 7 6 2 38

dD

52 18 344 112 1975 -11.9 -88.8 55 17 304 53 1622 -12.2 -87.7

3

H

7.10 ± 8.95 ± 8.10 ± 12.35 ± 5.90 ± 6.10 ± 8.08 2.39 29.60

9.75 ± 1.90 4.65 ± 1.80

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

25

Table 2 (Contd.) Locality

T

28/10/02 43.0 08/4/03 42.5 10/7/03 43.1 16/10/03 41.8 13/4/04 37.4 14/4/04 40.8 27/6/04 43.4 12/10/04 44.5 average 42.0 r 1.9 VC% 4.6 Bolu-3b 08/4/03 6.7 10/7/03 13.7 16/10/03 11.6 13/4/04 19.7 27/6/04 18.1 12/10/04 16.9 average 14.6 r 5.4 VC% 36.7 Mudurnu-4a 25/3/02 38.3 10/7/02 38.3 29/10/02 44.3 09/4/03 39.1 11/7/03 39.2 17/10/03 39.3 14/4/04 39.4 27/6/04 39.9 13/10/04 40.1 average 39.8 r 1.8 VC% 4.5 Mudurnu-4d 25/3/02 11.3 10/7/02 15.4 29/10/02 16.9 09/4/03 12.2 11/7/03 15.7 17/10/03 17.4 14/4/04 13.2 27/6/04 15.5 13/10/04 18.7 average 15.1 r 2.5 VC% 16.3

HCO3

Cl

SO4 Na

K

Ca

Mg

TDS

d18O

dD

1980 1961 1960 1957 1957 1957 0 0

787 955 786 663 793 788 808 648 756 120 16

11 10 16 9 14 14 11 14 11 4 32

373 326 325 482 499 451 298 412 447 127 29

47 22 33 38 40 41 39 46 41 10 23

18 18 14 16 16 16 15 17 16 1 8

328 360 335 356 369 361 273 351 338 30 9

27 41 20 15 27 30 48 16 39 29 74

1589 1731 1528 1579 1758 1702 1493 1504 1648 148 9

-8.6 -11.4 -12.1 -11.5 -12.4 -11.8 -12.5 -10.6 -11.5 1.2 10.2

-64.2 -82.7 -84.0 -79.4 -87.3 -89.1 -82.3 -82.1 -82.8 7.3 8.8

3.35 ± 3.30 ± 1.00 ± 1.85 ± 2.05 ± 3.65 ± 0.00 ± 0.90 ± 3.05 2.76 90.47

1.75 1.65 1.85 0.70 2.00 1.70 1.70 1.80

7.5 7.7 7.7 7.9 8.4 8.4 8.0 0.4 5.1

78 86 84 89 90 88 3 4

25 41 51 34 49 46 41 10 24

3 5 4 9 4 7 5 2 46

15 8 10 9 5 4 8 4 49

4 3 1 4 4 5 3 1 39

2 2 3 2 2 3 2 0 18

11 1 12 2 12 5 9 4 11 3 10 4 11 3 1 1 11 46

60 72 85 70 77 78 74 9 12

-11.6 -12.4 -12.2 -12.5 -12.0 -10.8 -11.9 0.6 5.4

-79.1 -79.2 -75.1 -84.5 -82.8 -75.7 -79.4 3.7 4.7

15.20 ± 11.95 ± 8.40 ± 11.20 ± 7.30 ± 8.70 ± 10.46 2.92 27.90

2.00 2.15 0.90 2.30 1.90 2.05

6.3 6.3 6.2 6.1 6.3 6.3 6.2 6.2 6.1 6.2 0.1 1.0

1162 1151 1162 1152 1148 1150 1150 2 0

767 787 689 780 744 695 753 726 634 731 50 7

1 3 7 11 13 8 11 8 12 8 4 49

30 10 34 32 29 31 25 27 21 27 7 28

30 34 23 46 16 5 20 18 23 24 12 48

9 8 6 7 5 7 6 6 7 7 1 18

24 43 46 43 70 56 25 66 26 44 17 39

1030 1066 949 1080 1010 950 1039 976 896 1000 61 6

-11.8 -12.2 -12.2 -11.7 -12.1 -12.0 -12.1 -12.7 -11.1 -12.0 0.4 3.7

-83.3 6.00 ± -88.0 3.90 ± -81.6 0.00 ± -87.3 0.65 ± -82.9 0.00 ± -84.3 0.25 ± -86.9 2.30 ± -86.0 0.00 ± -82.1 0.30 ± -84.7 1.49 2.4 2.15 2.8 144.68

1.80 1.85 1.60 1.65 1.75 0.60 2.00 1.80 1.90

7.3 7.1 6.9 7.0 7.1 7.1 7.2 7.0 7.0 7.1 0.1 1.6

791 743 808 743 797 839 793 48 6

415 432 476 464 495 461 518 432 462 35 8

8 5 9 9 12 11 7 13 9 3 28

72 50 35 36 37 24 28 75 45 19 43

24 28 18 17 13 15 14 19 18 5 29

4 108 32 5 130 16 4 82 46 4 97 37 2 69 63 3 114 25 4 76 58 4 99 47 4 97 40 1 21 16 21 21 40

662 666 669 662 692 653 704 689 674 18 3

-10.5 -11.4 -11.6 -12.3 -11.1 -11.1 -11.1 -10.9 -10.5 -11.2 0.6 5.0

-80.7 -82.7 -82.0 -82.6 -76.7 -74.5 -83.4 -79.1 -81.9 -80.4 3.1 3.8

2.05 1.95 1.70 1.85 2.05 0.80 2.20 1.90 1.95

pH

EC

6.1 6.1 5.9 6.3 6.2 6.0 6.2 6.1 6.2 0.3 4.7

168 182 144 161 133 147 199 126 174 159 24 15

3

H

12.46 ± 9.30 ± 6.35 ± 4.95 ± 8.15 ± 5.20 ± 5.25 ± 5.80 ± 4.55 ± 6.89 2.61 37.95

S. Su¨er et al.

26

Pure appl. geophys.,

Table 2 (Contd.) Locality

T

Seben-5a 25/3/02 70.1 10/7/02 69.9 29/10/02 72.3 09/4/03 70.7 11/7/03 71.8 17/10/03 72.3 14/4/04 71.6 27/6/04 72.6 13/10/04 71.5 average 71.4 r 1.0 VC% 1.4 Seben-5d 25/3/02 13.1 10/7/02 29/10/02 15.1 09/4/03 14.5 11/7/03 14.5 17/10/03 15.7 14/4/04 14.4 27/6/04 16.5 13/10/04 16.2 average 15.0 r 1.1 VC% 7.4 Hamamo¨zu¨-6a 26/3/02 41.3 11/7/02 41.1 30/10/02 42.6 10/4/03 42.1 12/7/03 42.7 18/10/03 42.5 15/4/04 42.5 25/6/04 43.6 14/10/04 44.4 average 42.5 r 1.0 VC% 2.4 Hamamo¨zu¨-6b 30/10/02 15.8 10/4/03 10.6 12/7/03 19.3 18/10/03 17.4 15/4/04 12.7 25/6/04 19.5 14/10/04 19.0 average 16.3 r 3.5 VC% 21.4

HCO3 Cl SO4

Na

K

Ca

Mg

TDS

d18O

2210 2180 2200 2190 2180 2190 2187 6 0

1201 1171 1109 1221 1202 1194 1294 1098 1186 63 5

80 138 60 98 65 138 67 57 62 55 66 101 64 48 25 94 61 91 16 36 26 39

530 427 441 408 434 427 486 363 439 50 11

40 31 32 32 28 30 29 31 32 4 12

58 40 51 57 46 51 38 38 47 8 17

0 19 6 21 21 20 3 19 14 9 66

2047 1847 1842 1863 1848 1889 1962 1667 1871 109 6

-12.4 -12.3 -11.7 -11.8 -12.5 -12.7 -12.3 -12.3 -10.9 -12.1 0.6 4.7

7.3 7.3 7.2 7.5 7.4 7.4 7.4 7.5 7.4 0.1 1.2

1076 1070 1073 1128 1107 1079 1105 25 2

332 329 353 331 625 331 356 272 366 108 29

12 11 14 16 13 20 12 16 14 3 21

400 291 249 259 37 342 240 307 266 107 40

100 62 13 53 66 64 57 63 60 24 40

6 6 6 5 6 6 5 6 6 0 9

106 112 114 103 98 132 77 84 103 17 17

40 40 59 44 42 44 59 56 48 8 18

996 -8.7 -69.3 850 -8.3 -68.7 806 -8.5 -66.8 810 -8.9 -67.2 886 -9.2 -67.8 938 -10.9 -75.4 804 -8.6 -66.6 803 -8.1 -65.4 862 -8.9 -68.4 73 0.9 3.1 8 9.6 4.5

7.8 7.7 7.2 7.0 7.3 7.3 7.2 7.4 7.3 7.4 0.3 3.4

516 505 516 514 513 514 514 1 0

234 221 235 244 244 244 234 254 231 238 10 4

36 39 33 35 36 33 36 35 37 36 2 5

15 28 25 29 24 25 23 26 18 24 4 19

49 8 62 8 44 6 50 6 38 5 41 6 42 5 39 1 45 5 46 6 7 2 16 35

35 37 39 39 48 30 39 36 41 38 5 13

16 13 17 16 17 24 18 26 16 18 4 23

394 408 399 418 413 404 397 417 393 405 10 2

-12.4 -12.3 -11.9 -11.7 -12.5 -11.9 -12.0 -12.3 -11.3 -12.0 0.4 3.1

-87.2 -88.2 -92.3 -85.3 -85.3 -83.7 -90.5 -85.1 -89.1 -87.4 2.8 3.3

2.10 ± 1.75 ± 1.15 ± 0.55 ± 1.40 ± 0.05 ± 0.75 ± 1.50 ± 0.60 ± 1.09 0.66 60.02

1.65 1.55 1.65 1.50 1.75 0.60 2.10 1.80 1.80

7.6 7.3 7.4 7.8 7.4 7.5 7.5 7.5 0.2 2.1

636 505 609 602 611 626 613 12 2

351 340 330 311 345 321 320 331 15 4

9 21 12 9 11 8 11 11 4 38

31 38 46 41 34 35 32 37 5 15

17 2 13 2 11 1 28 2 14 2 12 4 15 2 16 2 6 1 39 42

84 98 45 66 93 66 97 79 20 26

24 17 47 20 18 27 11 23 12 50

517 528 491 478 515 474 488 499 21 4

-10.7 -11.8 -10.2 -9.1 -9.5 -9.7 -9.6 -10.1 0.9 9.0

-78.3 -73.8 -70.8 -66.3 -82.1 -75.8 -74.4 -74.5 5.1 6.8

9.65 ± 10.55 ± 12.50 ± 10.00 ± 10.45 ± 12.25 ± 10.45 ± 10.84 1.10 10.15

1.80 1.90 1.90 0.95 2.25 2.10 2.15

pH

EC

6.7 6.9 6.5 6.4 6.6 6.8 6.5 6.4 6.5 6.6 0.2 2.6

dD

3

H

-92.3 0.00 ± -91.4 0.85 ± -90.0 0.00 ± -88.4 0.80 ± -86.5 0.00 ± -85.8 0.40 ± -89.5 0.00 ± -91.1 0.00 ± -86.0 1.65 ± -89.0 0.41 2.5 0.58 2.8 141.94 8.40 ± 10.20 ± 10.20 ± 11.20 ± 9.70 ± 8.30 ± 9.20 ± 10.30 ± 9.69 1.00 10.33

1.60 1.50 1.60 1.55 1.75 0.60 2.00 1.80 1.95

1.80 1.90 1.90 1.80 1.00 2.25 2.00 2.00

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

27

Table 2 (Contd.) Locality

T

Go¨zlek-7a 27/3/02 38.4 11/7/02 38.3 31/10/02 39.3 10/4/03 39.6 12/7/03 38.9 18/10/03 38.8 15/4/04 38.8 25/6/04 39.3 14/10/04 40.1 average 39.1 r 0.6 VC% 1.5 Go¨zlek-7b 31/10/02 16.1 10/4/03 16.3 12/7/03 17.2 18/10/03 19.2 15/4/04 15.8 25/6/04 20.5 14/10/04 19.6 average 17.8 r 1.9 VC% 10.7 Res¸ adiye-8a 27/3/02 41.3 12/7/02 41.1 average 41.2 r 0.1 VC% 0.3 Res¸ adiye-8b 27/3/02 12.5 12/7/02 17.4 31/10/02 15.9 average 15.3 r 2.5 VC% 16.4 Kurs¸ unlu-9a 28/3/02 55.9 13/7/02 55.5 01/11/02 57.4 11/4/03 57.4 12/7/03 57.9 19/10/03 58.1 16/4/04 56.0 26/6/04 58.2 15/10/04 59.9 average 57.4

pH

EC

HCO3

Cl

SO4

Na

K

Ca Mg

TDS

d18O

dD

3

H

7.9 8.0 7.6 7.4 8.0 7.8 7.8 7.7 7.7 7.8 0.2 2.3

497 488 497 496 494 495 495 1 0

251 248 238 226 258 245 250 267 256 249 12 5

14 13 14 17 15 14 16 14 15 15 1 8

5 40 37 35 35 34 30 33 28 31 10 33

78 89 67 78 68 73 77 92 71 77 9 11

6 6 4 5 4 4 4 3 4 5 1 21

17 18 22 18 32 17 19 19 19 20 5 23

5 6 11 6 9 12 9 5 10 8 3 34

376 420 393 384 420 400 405 432 404 404 18 4

-13.3 -13.0 -12.5 -13.4 -12.3 -13.4 -12.8 -13.0 -11.8 -12.8 0.6 4.3

-95.5 -95.3 -96.7 -92.8 -91.3 -90.1 -98.3 -94.2 -93.9 -94.2 2.6 2.7

7.9 7.5 7.5 7.8 7.8 7.7 7.7 7.7 0.2 2.0

1063 1157 1255 1293 1242 1304 1280 33 3

363 440 429 411 438 336 403 43 11

43 54 65 64 54 60 57 8 14

228 200 200 248 212 290 230 35 15

47 50 38 53 45 56 48 6 13

4 4 3 4 4 4 4 0 10

80 110 68 131 81 133 101 28 28

70 64 95 58 85 65 73 14 20

834 921 898 969 919 945 914 46 5

-10.7 -10.1 -9.9 -9.9 -10.5 -9.6 -9.4 -10.0 0.5 4.7

-80.1 11.20 ± 1.80 -74.2 9.00 ± 1.80 -75.2 9.96 ± 1.85 -74.9 9.50 ± 0.95 -82.2 7.95 ± 2.15 -76.7 11.40 ± 2.05 -73.6 10.45 ± 2.00 -76.7 9.92 3.3 1.23 4.3 12.35

786 40 821 195 804 118 25 110 3 93

760 840 800 57 7

2 55 9 125 9 46 7 75 4 43 60 57

20 42 25 29 12 40

6.4 6.4 6.4 0.0 0.3

- 1805 - 1829 - 1817 17 1

7.8 6.5 7.6 7.3 0.7 9.4

481 481 -

180 238 235 218 32 15

7.2 7.1 6.9 6.9 7.1 7.1 6.6 7.0 7.0 7.0

11300 11100 11030 11270 11070 11140 11160

6100 4906 5612 6002 6809 6184 5984 5942

49 280 85 54 328 92 52 304 89 4 34 5 7 11 6

2.10 ± 2.45 ± 0.55 ± 1.40 ± 2.12 ± 0.25 ± 0.00 ± 1.80 ± 0.55 ± 1.25 0.92 73.88

1.65 1.65 1.60 1.55 1.65 0.60 1.90 1.80 1.80

3805 -12.9 -93.3 4160 -12.6 -92.5 3982 -12.7 -92.9 251 0.2 0.6 6 1.9 0.6

3.05 ± 1.70 2.10 ± 1.65 2.58 0.67 26.09

19 30 23 24 5 21

320 494 381 398 88 22

-12.7 -11.6 -11.9 -12.1 0.6 4.7

-86.1 -84.5 -82.2 -84.3 1.9 2.3

9.20 ± 1.80 9.25 ± 1.70 7.55 ± 1.80 8.67 0.97 11.16

- - 756 52 2650 44 73 28 773 46 2090 219 113 11 797 58 2445 220 39 22 762 103 2478 225 52 58 813 108 2490 832 108 21 758 48 2630 241 15 31 767 52 2450 47 68 0 775 67 2462 261 67 24

9703 8158 9192 9680 11181 9907 9368 9598

-8.5 -8.5 -8.4 -8.7 -8.5 -8.6 -8.8 -8.6 -7.5 -8.5

-88.5 -88.5 -87.2 -87.5 -89.3 -90.6 -94.9 -84.2 -83.8 -88.3

0.35 ± 0.00 ± 0.00 ± 1.35 ± 0.00 ± 0.40 ± 0.00 ± 0.00 ± 0.26

3 3 2 3 1 22

40 48 41 43 4 10

1.70 1.55 1.65 1.60 1.60 0.60 1.95 1.80

S. Su¨er et al.

28

Pure appl. geophys.,

Table 2 (Contd.) Locality

T

r 1.4 VC% 2.4 Kurs¸ unlu-9b 28/3/02 12.6 13/7/02 14.7 01/11/02 15.3 11/4/03 10.6 average 13.3 r 2.1 VC% 16.1 Kurs¸ unlu-9c 28/3/02 9 13/7/02 14.9 01/11/02 13.3 11/4/03 11.9 12/7/03 12.5 19/10/03 13.4 16/4/04 11.0 26/6/04 14.1 15/10/04 14.5 average 13.2 r 1.3 VC% 10.1

pH

EC

0.2 2.6

101 1

HCO3 Cl SO4

Na

K

581 10

22 27 3 40

184 266 7 102

6.5 6.4 6.5 2660 1476 6.5 2440 1429 6.4 2550 1452 0.0 156 34 0.2 6 2

85 33 82 37 84 35 2 3 2 8

458 431 445 19 4

8.2 7.8 8.1 7.7 7.9 8.0 8.0 8.0 8.1 8.0 0.2 1.9

22 2 7 6 4 8 5 9 8 6 78

416 406 427 437 474 455 455 19 4

241 240 247 256 250 271 275 245 253 13 5

7 5 10 11 12 14 6 9 9 3 33

8 13 7 6 31 6 5 9 11 9 80

Ca

Mg

TDS

d18O

36 53

18 74

905 9

0.4 4.7

dD

66 72 64 62 57 75 66 65 66 6 9

8 4 10 15 3 11 15 12 10 5 48

353 337 346 355 356 387 372 349 357 16 4

-10.4 -10.0 -10.0 -9.8 -9.3 -10.3 -9.0 -9.8 -9.5 -9.8 0.4 4.4

H

3.3 0.47 3.8 179.48

- -11.3 -86.3 - -9.7 -83.5 43 71 32 2199 -10.2 -77.7 45 130 1 2154 -10.7 -77.9 44 101 16 2177 -10.5 -81.3 1 42 22 32 0.7 4.2 2 42 136 1 6.5 5.2 1 1 1 0 1 1 0 0 1 0 60

3

-76.9 -81.5 -77.8 -74.9 -75.0 -78.6 -79.5 -73.5 -78.8 -77.4 2.6 3.3

7.25 ± 9.85 ± 9.65 ± 11.85 ± 9.65 1.88 19.52

1.80 1.70 1.80 1.90

15.70 ± 13.75 ± 13.25 ± 14.75 ± 13.85 ± 14.25 ± 10.80 ± 14.45 ± 13.85 1.44 10.36

2.00 1.90 1.85 2.00 2.05 1.10 2.10 2.05

Temperature in C, Electrical Conductivity (EC) in mmho/cm, major ion concentrations and total dissolved solid contents in mg/l. d18O and dD expressed in % with respect to V-SMOW. The analytical errors associated with d18O and dD are 0.1 % and 1 % , respectively. Tritium concentrations given as Tritium Units (TU) together with the associated analytical errors. r is the standard deviation, VC is the variation coefficient (= {standard deviation / mean} x 100)

The bicarbonate character of most waters seems to be compatible with the dissolution of reservoir rocks that are dominated by Mesozoic limestones, whereas the dominancy of Na cation in the hot waters can be attributed to ion exchange (of the waters) with the overlying sediments, including the impermeable clayey levels. As an exception to the dominantly bicarbonate nature of the NAFZ waters, the Yalova thermal waters display sulphate character. In general, sulphate-rich waters can originate from either i) dissolution of evaporites containing gypsum (as previously suggested by EISENLOHR (1995) for the sulphate waters of Armutlu located in close proximity to Yalova), or ii) oxidation of sulphide-bearing minerals (contained in Mesozoic-Paleozoic rocks). Alternatively, the sulphate character of the Yalova thermal waters might be genetically connected to young organic accumulations in Izmit Bay (located to the north of Yalova) and may reflect the release of sulphur (entrapped in these organic-rich sediments) induced by proximal seismic activity.

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

29

Figure 2 a) Pie diagrams of the hot water samples, b) Pie diagrams of the cold water samples (average data of all sampling periods are taken, TDS contents are not reflected in the pie diameters of the diagrams).

5.2. Stable Isotope Compositions The oxygen- and hydrogen-isotope compositions of the waters are presented in the d18O vs. dD diagram in Figure 3, where the data plotted represent the mean values of the measurements throughout the monitoring period. As can be seen from the figure, both hot and cold waters lie close to the Global Meteoric Water Line (CRAIG, 1961). This observation reveals that the hot waters, as well as the cold waters, are overwhelmingly meteoric in origin. However, the hot water sample (no. 9a) from the Kurs¸ unlu field has a

30

S. Su¨er et al.

Pure appl. geophys.,

Figure 3 d18O vs. dD diagram of the sampled waters (data represent the average of nine sampling periods, GMWL: Global Meteoric Water Line (CRAIG, 1961), MMWL: Mediterranean Meteoric Water Line (GAT and CARMI, 1970)).

relatively high d18O value which we interpret as reflecting the effects of (i) water-rock exchange, and/or (ii) calcite scaling in the production well from which the sample was collected. The spatial distributions of the oxygen- and hydrogen- isotope compositions of the water samples are depicted in Figures 4 and 5 as d18O vs. locality and dD vs. locality diagrams, respectively. As observed, the hot waters in almost all fields show more negative (lower) values of d18O and dD than the cold waters, and this situation persists for all sampling periods. This is a rather interesting feature as hot waters might be predicted to have higher d18O values than cold waters owing to intense water-rock interaction at high temperatures. The only explanation for the low values recorded for the hot waters in this study seems to be their recharge characteristics which must occur at higher altitudes (compared to the altitudes where cold water aquifers are recharged). However, owing to the low number of cold waters sampled in the study, no attempt has been made here to estimate the recharge altitudes of the various waters: this would require sampling of several cold waters discharging from varying elevations.

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

31

Figure 4 d18O vs. locality diagrams for all sampling periods (filled symbols show hot, empty symbols show cold waters, gray-filled symbol shows the mineral water sample of Kurs¸ unlu field).

32

S. Su¨er et al.

Pure appl. geophys.,

Figure 5 dD vs. locality diagrams for all sampling periods (symbols are same as Fig. 4).

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

33

5.3. Tritium Contents The spatial distribution of the tritium measurements are shown in Figure 6 as plots of tritium concentrations (expressed as Tritium units, TU) vs. sampling localities. Inspection of Figure 6 reveals that, for each field, the tritium concentrations of the hot waters are lower (majority < 5 TU) than those of the cold waters (majority > 8 TU). Given that (i) the level of tritium in the atmosphere was 5–25 TU in the 1950s, while it increased to about 2900 TU in 1963 following the start of nuclear testing in 1960 (IAEA, 1992), and ii) tritium has a half life of 12.26 years, the rather low tritium concentration levels of the hot waters in this study suggest that their aquifers were recharged by precipitation exceeding approximatly 50 years. The cold water aquifers, on the other hand, appear to have been recharged with a component of relatively younger precipitation.

6. Discussion Temporal variations were recorded in both the chemical and isotopic compositions of the water samples as well as in their temperature, pH and EC values during the course of the monitoring program. 6.1. Temperature and pH Variations Examination of Table 2 reveals a temporal variation in temperature, with the amplitude of fluctuations (expressed by variation coefficient, VC) ranging between 7.4 (Seben) and 36.7 (Bolu) for the cold waters, and between 0.3 (Res¸ adiye) and 6.4 (Yalova) for the hot waters. The observation that fluctuations are higher in the cold waters points to the fact that the waters from shallow aquifers either reflect admixture with recentlyrecharged waters or that they have been considerably more affected by seasonal changes in the ambient temperature. In contrast, the pH variations appear to be site-dependent rather than affected by seasonal effects: In some localities the amount of fluctuations in the hot waters is higher than those in their cold water companions, whereas, at other localities, the situation is reversed (Table 2). The highest pH variation in the hot waters was recorded in the Efteni field (8% of the mean), and the lowest variation was recorded in the Res¸ adiye field (0.3% of the mean) where the variation in the cold waters is the highest (9.4% of the mean) among all the fields. 6.2. Chemical Variations As regards chemical compositions, an important feature to note is that the extent of observed variation is essentially dependent on the levels of ionic concentrations (the lower the concentration, the higher the variation coefficient). In this respect, the variations are higher in cold waters due to their lower TDS contents. However,

34

S. Su¨er et al.

Pure appl. geophys.,

Figure 6 Tritium concentrations vs. locality diagrams for all sampling periods (symbols are same as Fig. 4).

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

35

considerable fluctuations are also observed in the hot waters of Efteni and Yalova which are located along the western-central segment of the NAFZ. The most striking features, in this respect, can be summarized as follows: 6.2.1. Efteni geothermal field. In the July 2002 sampling period, there is about 1.3-fold decrease in the Ca content (compared to the mean of all the periods) for the Efteni hot water. This decrease is accompanied by an increase in the Cl content and pH value, the latter reaching its maximum in the October 2002 period (Fig. 7). Given that any pH increase in waters is accompanied by a decrease in calcite dissolution, and that other studies in progress on Efteni waters reveal synchronous increases in d13C values and CO2/3He ratios of the geothermal fluids (DE LEEUW et al., in prep.), these variations likely reflect disturbance in the hydrothermal system via degassing. This leads to (i) an increase in the CO2/3He ratio as well as the d13C values of the residual waters (as He and 12C partition into the gas phase in preference to CO2 and 13C, respectively) and (ii) an increase in pH-inducing calcite precipitation and hence a decrease in the Ca content. The occurrence of travertine deposition around the Efteni hot-spring is consistent with these observations. The recorded variations appear to correlate with the 14 July, 2002 Yıg˘ılca-Du¨zce (M 3.1) and the 15 July, 2002 Yıg˘ılca-Du¨zce (M 2.8) earthquakes (Fig. 7). In this respect, degassing in the hydrothermal system might have been induced by a decrease in regional stress/strain levels related to the afore-mentioned earthquakes. In April 2003, the pH value at Efteni seems to have returned to the mean value observed prior to the July 2002 events: This was accompanied by an increase in Ca content. The July 2003 and June 2004 periods at Efteni also deserve attention as there seems to be about 0.5-fold decrease in Ca concentration paralleling the two-fold increase in the Mg content of the hot water (compared to the mean values). Given that Mg is the dominant cation of the cold water in the Efteni field (Fig. 2 and Table 2), the variations observed in these 2 periods suggest mixing between hot waters and shallow, cold waters. It is important to note that (i) the period of mixing in July 2003 appears to follow the July 1, 2003 Kaynas¸ lı-Du¨zce (M 3.2), and/or precede the July 25, 2003 Yıg˘ılca-Du¨zce (M 3.1 and 4.0) earthquakes, and (ii) the mixing in June 2004 correlates with the June 15, 2004 Yıg˘ılca Du¨zce (M 3.6), June 23, 2004 Bolu (M 4.6) and/or the July 2, 2004 Yıg˘ılca-Du¨zce (M 4.8) earthquakes (Fig. 7). Interestingly, however, no significant change was recorded in the ionic concentrations (and/or in temperature-pH) of the Efteni water prior to or following the nearby Bolu earthquake on April 14, 2004 (M 4.7). 6.2.2. Yalova geothermal resort. For the Yalova hot water, compared to the mean of all the sampling periods, there is about a 0.7-fold decrease in Cl and accompanying 1.2-fold increase in SO4 contents in the July 2002 period (Fig. 7). Since Cl is considered to be a conservative constituent of the hot waters, the decrease in the Cl content points to a hot-cold water mixing process, and seems to be correlated with the seismic activity which occurred on the 3rd and/or 13th of July, 2002 in Armutlu-Yalova (M 3.1). Given that the

36

S. Su¨er et al.

Pure appl. geophys.,

Figure 7 Temporal variations of some selected parameters for the Efteni and Yalova hot waters (solid vertical lines: timing of the nearby earthquakes, Apr-I(04) and Apr-II(04) stand for sampling campaigns before and after the 14/4/2004 Bolu earthquake).

sampling date in Yalova was the 9th of July, it is difficult to decide, however, whether the above mentioned Cl decrease is the precursor of the 13th of July, 2002 seismic activity or if it is the post-earthquake response to the 3rd of July, 2002 activity. Regarding the increase in SO4, however, mixing cannot be a viable mechanism as the dominant anion of the cold waters in the Yalova field is HCO3. This increase may be due to the possible release of sulphur (entrapped in the organic sediments) induced by the seismic activity of this period.

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

37

Another important variation in the Yalova hot water concerns the drastic increase in the Mg content in the October 2004 period. Although the sampling date in this period (12/10/ 2004) correlates with the October 10th, 2004 C¸ınarcık-Yalova earthquake (M 3.0), there is no other significant variation recorded in other parameters. This leads to the suspicion that the variation may be an analytical artifact unrelated to any seismic activity. 6.3. Tritium Variations As in the case of chemical variations, the variation coefficient is higher in those waters which have lower levels of tritium concentration. In this respect, hot waters display considerably higher tritium concentration variations than cold waters. The variations in tritium concentrations, however, require particular caution due to high analytical errors (about ± 2 TU) resulting from the high background level in the counter. Nevertheless, significant variations beyond the limits of analytical error are observed in the Bolu, Mudurnu and Yalova fields, in the March and July 2002 sampling periods (Fig. 8). Although the anomalies in Bolu (March 2002) and Mudurnu (March and July 2002) appear to correlate with the March 23rd, 2002 Sea of Marmara (M 4.7), July 14th, 2002 Yıg˘ılca-Bolu (M 3.1), and the July 15th, 2002 Yıg˘ılca-Du¨zce (M 2.8) earthquakes, there is no concomitant variation, in these periods, in the chemical composition of the hot waters in the relevant fields. At Yalova, in the July 2002 period, a significant tritium increase (outside the limits of analytical errors) was recorded (Fig. 8) accompanying the decrease in Cl content (Fig. 7) and supporting the idea of hot-cold water mixing (section 6.2) that can be correlated with the 3rd and/or the13th of July 2002 Armutlu-Yalova earthquake (M 3.1). Apart from July 2002, a high tritium content in the Yalova hot water sample was also detected in March 2002. Although not coupled with any significant variation in anion-cation contents, this

Figure 8 Temporal variations of tritium contents (TU) for the Yalova, Bolu and Mudurnu hot waters (Apr-I(04) and Apr-II(04) stand for sampling campaigns before and after the 14/4/2004 Bolu earthquake).

38

S. Su¨er et al.

Pure appl. geophys.,

tritium anomaly correlates with the seismic activity that occurred in the Sea of Marmara (M 4.7) on the same day as sampling (23 March, 2002). It is particularly important to note that in March 2002, an anomalous value of +5.79% was recorded in the Yalova geothermal field in d13C values of dissolved CO2—this value contrasted with significantly lower (< 0%) values, for all other sampling periods (DE LEEUW et al., in prep.). In the April 2004 sampling period, a significant increase was detected in the TU contents of Efteni (outside the limits of analytical error, 4-fold increase) and Bolu (within the limits of analytical error) in the second sampling campaign performed following the April 14, 2004 earthquake in Bolu (M 4.7). At Mudurnu, on the other hand, an apparent increase in the tritium content was detected on the same day as the April 14, 2004 Bolu earthquake. Although the afore-mentioned fields are located along different strands of the NAFZ (Mudurnu located on the southern strand, and Efteni and Bolu fields located on the northern strand), their response to the seismic activity was the same. Therefore, the increase observed in the tritium contents of these fields can reflect an immediate response probably related to a hot-cold water mixing process triggered by the earthquake in Bolu. 6.4. d18O – dD Variations Temporal variations in d18O and dD values are less than 5% of the mean for hot waters, and up to 10% of the mean for cold waters. This probably reflects the effects of seasonal variations (in the amount of precipitation and/or evaporation) which are more pronounced in cold waters emanating from shallow aquifers. The most prominent variations in hot waters are observed in the Bolu field where a nearly 1.5-fold increase

Figure 9 Temporal variations of d18O and dD values for the Efteni and Bolu waters (Apr-I(04) and Apr-II(04) stand for sampling campaigns before and after the 14/4/2004 Bolu earthquake).

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

39

was recorded in the October 2002 period for both d18O and dD values (Fig. 9). One possible means of producing variations in the d18O and dD values of waters is isotopic exchange with CO2 and H2S gases, respectively, as has been previously proposed by BALDERER et al. (2002) for the increase in d18O and dD values of Kuzuluk mineral water and the Bursa thermal spring. Given that the isotopic variations in Bolu in October 2002 correlate with the October 21st, 2002 Bolu earthquake (M 2.6), seismicity-induced gas emissions could be a possible consequence. In fact, in the Bolu field during the October 2002 period, there seems to be a slight decrease in pH and increase in HCO3 values (as expected from CO2 dissolution in water) supporting the link between CO2 emission and observed d18O variation. Regarding the dD variation, on the other hand, although it was not monitored, any possible H2S gas release is anticipated to increase the SO4 content of the waters. During the October 2002 period, however, about a two-fold decrease was recorded in SO4 content of the Bolu hot water compared to the preceding period. Another alternative means to increase d18O and dD values is evaporation. Such a process could have resulted from possible adiabatic boiling associated with a pressure decrease due to either (i) a disturbance in the regional stress/strain distribution (possibly seismicity-induced), or (ii) excessive pumping in the well from which the sample was collected. This latter suggestion is difficult to evaluate, however, as pumping records are unavailable. The Efteni field also deserves attention in regard to d18O – dD variations: There is a slight (about 3–4% above the mean value) decrease in dD values of both hot and cold waters in the April 2004 period. The sampling dates in this field are 12 and 14 April, 2004, where the second sampling was performed immediately after the seismic activity which occurred on April 14, 2004. Considered in this framework, dD values appear to have started decreasing before the earthquake and increased to their original values after the earthquake. However, no accompanying variations were recorded in d18O values, and it is difficult to assess the mechanism responsible for dD variations.

7. Conclusions

1. The geothermal waters along the NAFZ are mostly Na-HCO3 in character with the exceptions of Na-SO4 type waters (at Yalova) and Ca-HCO3 type waters (at Bolu and Mudurnu). In contrast, the cold waters are mostly of Ca-HCO3 type. While the dominant HCO3 character in the hot and the cold waters seems to be compatible with the dissolution of reservoir rocks that are dominated by Mesozoic limestones, ion exchange with the overlying sediments is probably responsible for the dominancy of Na cation in the hot waters. 2. The oxygen-and hydrogen- isotope compositions suggest a meteoric origin for both hot and cold waters. The high d18O value observed in Kurs¸ unlu hot water, as an exception, is probably related to more extensive water-rock interaction and/or the

40

S. Su¨er et al.

Pure appl. geophys.,

scaling problem observed in the production well from which samples were collected, either higher recharge altitudes for the aquifers of the former compared to the latter, or different climatic conditions during infiltration. 3. Tritium contents of the cold waters are higher than those of the hot waters, revealing that the cold water aquifers are recharged by more recent precipitation. The low TU contents of the hot waters, on the other hand, suggest deep circulating hot waters with long residence times (> 50 years). 4. The monitoring program covering a total of nine sampling periods over three years has revealed temporal variations in both chemical and isotopic compositions of the geothermal waters. Some of these variations appear to correlate with seismic activity occurring close in time and space to the sampling sites. These variations probably reflect the effects of disturbances in the hydrologic/hydrogeochemical system that might have been induced by changes in the regional stress/strain distribution. In this respect, Cl, tritium and Ca appear to be the most sensitive geochemical parameters, and Yalova and Efteni are the key localities for further monitoring studies. 5. Continuing monitoring of water compositions, coupled with gas chemistry (e.g., CO2/ He, and d13C) should lead to a better understanding of their relationship to seismic activities.

Acknowledgements _ ¨ BITAK This study was supported by TU (YDABAG-100Y097) and NSF (Grant no. ¨ zdemir and Hu¨seyin Sendir for EAR-0229508) projects. We would like to thank Yavuz O ¨ zcan Eyu¨pog˘lu for her assistance their support during the field work, and Sabahat O during tritium analyses. We are grateful for helpful comments and constructive reviews by N. Pe´rez and W. Balderer, which improved our manuscript.

REFERENCES _¸ leri Genel Mu¨du¨rlu¨g˘u¨, Yayın No. ALTAY, T. and C¸IFTER, C. (1996), Trityum laboratuvarı el kitabı. Devlet Su Is IZ-903 (in Turkish). _ BALDERER, W., LEUENBERGER, F., SUNER, F., YALC¸ıN, T., and STICHLER, W. (2002), Effects of C ¸ ınarcık-Izmit August 17, 1999 earthquake on the composition of thermal and mineral waters as revealed by chemical and isotope investigation, Geofisica Internacional 41 (4), 385–391. BARKA, A.A. (1992), The North Anatolian Fault Zone, Annales Tecton. 6, 164–195. BOLOGNESI, L. (2000), Earthquake-induced variations in the composition of the water in the geothermal reservoir at Vulcano Island, Italy, J. Volcanol. Geoth. Res. 99, 139–150. BOZKURT, E. (2001), Neotectonics of Turkey - A Synthesis, Geodinamica Acta 14, 3–30. BRA¨UER, K., KA¨MPF, H., STRAUCH, G., and WEISE, S.M. (2003), Isotopic evidence (3He/4He, d13C(CO2)) of fluidtriggered intraplate seismicity, J. Geophy. Res. 108 (B2), 2070, doi:10.1029/2002JB002077. CANIK, B. (1972), Du¨zce-Derdin hamamı civarının jeoloji ve hidrojeoloji etu¨du¨, MTA Report 6359, Ankara (in Turkish).

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

41

¨ ZAKSOY, V., and ERKMEN, C. (2000), Turkish earthquakes reveal C ¸ EMEN, I., Go¨KTEN, E., VAROL, B., KıLıc¸, R., O dynamics of fracturing along a major strike-slip fault zone, EOS, Transact. Am. Geophys. Union, 81/28. CLAESSON, L., SKELTON, A., GRAHAM, C., DIETL, C., MO¨RTH, M., Torssander, P., and Kockum, I. (2004), Hydrogeochemical changes before and after a major earthquake, Geology 32, 641–644. COLEMAN, M.L., SHEPPARD, T.J., DURHAM, J.J., ROUSE, J.E., and MOORE, G.R. (1982), Reduction of water with zinc for hydrogen isotope analysis, Anal. Chem. 54, 993–995. CRAIG, H. (1961), Isotopic variations in meteoric waters, Science 133, 1702–1703. DAS, N.K., BHANDARI, R.K., GHOSE, D., SEN, P., and SINHA B. (2005), Anomalous fluctuation of radon, gamma dose and helium emanating from thermal spring prior to earthquake, Curr. Sci. 89, 1399–1404. DE LEEUW, G.A.M., HILTON, D.R., GU¨LEc¸, N., and MUTLU, H., He-C relationships in geothermal waters from the North Anatolian Fault Zone (NAFZ), Turkey (in prep). DEWEY, J.F. and S¸ ENGo¨R, A.M.C. (1979), Aegean Sea and surrounding regions: Complex multiplate and continuum tectonics in a convergent zone, Geol. Soc. Am. Bull. Part I, 90, 84–92. EISENLOHR, Th. (1995), Die Thermalwa¨sser der Armutlu-Halbinsel (NW-Tu¨rkei) und deren Beziehung zu Geologie und aktiver Tektonik, Diss. ETH-Zu¨rich, Nr. 11340 (1995). EPSTEIN, S. and MAYEDA, T. (1953), Variation of 18O content of waters from natural sources, Geochim. et Cosmochim. Acta 4 (5), 213–224. ERZENOG˘LU, Z. (1989), Amasya Gu¨mu¨s¸ hacıko¨y Hamamo¨zu¨ (Arkutbey) kaplıcası H-1 no’lu sıcaksu sondajı kuyu bitirme raporu, MTA Report 8714, Ankara (in Turkish). FAVARA, R., ITALIANO, F., and MARTINELLI, G. (2001), Earthquake-induced chemical changes in the thermal waters of the Umbria region during the 1997–1998 seismic swarm, Terra Nova 13, 227–233. GAT, J.R. and CARMI, I. (1970), Evolution in the isotopic composition of atmospheric waters in the Mediterranean Sea area, J. Geophys. Res. 75, 3039–3048. HAUKSSON, E. (1981), Radon content of ground water as an earthquake precursor: Evaluation of worldwide data and physics basis, J. Geophys. Res. 86, 9397–9410. HILTON, D.R. (1996), The helium and carbon isotope systematics of a continental geothermal system: Results from monitoring studies at Long Valley Caldera (California, U.S.A.), Chem. Geol. 127, 269–295. IAEA (1992), Statistical treatment of data on environmental isotopes in precipitation. Tech. Rep. Series 331, Vienna. IGARASHI, G., SAEKI, T., TAKAHATA, N., SUMIKAWA, K., Tasaka, S., SASAKI, Y., and SANO, Y. (1995), Groundwater radon anomaly before the Kobe earthquake in Japan, Science 269, 60–61. IGARASHI, G. and WAKITA, H. (1995), Geochemical and hydrological observations for earthquake prediction in Japan, J. Phys. Earth 43, 585–598. ITALIANO, F. and MARTINELLI, G. (2001), Anomalies of mantle-derived helium during the 1997–1998 seismic swarm of Umbria-Marche, Italy, Geophysi. Res. Lett. 28, 839–842. ITALIANO, F., MARTINELLI, G., and RIZZO, A. (2004), Geochemical evidence of seismogenic-induced anomalies in the dissolved gases of thermal waters: A case study of Umbria (Central Apennines, Italy) both during and after the 1997–1998 seismic swarm, Geochem. Geophys. Geosyst. 5, 11, doi:10.1029/2004GC000720. KAWABE, I. (1984), Anomalous changes of CH4/Ar ratio insubsurface gas bubbles as seismo-geochemical precursors at Matsuyama, Japan, Pure Appl. Geophys. 122, 194–214. KING, C.-Y. (1986), Gas geochemistry applied to earthquake prediction: An overview, J. Geophys. Res. 91, 12269–12281. KING, C.-Y., KOIZUMI, N., and KITAGAWA, Y. (1995), Hydro-geochemical anomalies and the 1995 Kobe earthquake, Science 269, 38–39. KING, C.-Y., and IGARASHI, G. (2002), Earthquake-related hydrologic and geochemical changes, International Handbook of Earthquake and Engineering Seismology 81A, 637–645. KING, C.-Y., ZHANG, W., and ZHANG, Z. (2006), Earthquake-induced groundwater and gas changes. Puer Appl. Geophys. Topical Volume Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume I, 633– 646. KOc¸AK, A. (1974), C ¸ avundur (Kurs¸ unlu-C ¸ ankırı) kaplıca bo¨lgesinin hidrojeoloji etu¨du¨ raporu, MTA Report 6373, Ankara (in Turkish). ¨ ZACAR, A., and TEKSO¨Z, B. (1999a), Neotectonic framework of Turkey: A KOc¸YIG˘IT, A., BOZKURT, E., CIHAN, M., O special emphasis on the 17 August 1999 Go¨lcu¨k- Arifiye earthquake (NE Marmara, Turkey), Proc. Internat. Con. Earthquake Hazard and Risk in the Mediterranean Region, Near East University, Cyprus.

42

S. Su¨er et al.

Pure appl. geophys.,

¨ ZACAR, A., and TEKSO¨Z, B. (1999b), 12 Kasım 1999 Dag˘dibi (Du¨zceKOc¸YIG˘IT, A., BOZKURT, E., CIHAN, M., O Bolu) depremi, Geological Reconnaissance Report, Department of Geological Engineering, Middle East Technical University, Ankara (in Turkish). MCKENZIE, D.P. (1972), Active tectonics of the Mediterranean region, Geophys. J. R. Astron. Soc 30, 109–185. MTA (1996), Geothermal inventory of Turkey, General Directorate of Turkish Mineral Research and Exploration (MTA), Ankara, 480 pp (in Turkish). MU¨FTU¨OG˘LU, E. and AKıNCı, K. (1989), Bolu-Seben Keseno¨zu¨ kaplıcası ve c¸evresi jeotermal enerji aramaları sismik etu¨d raporu, MTA Report 8856, Ankara (in Turkish). NISHIZAWA, S., IGARASHI, G., SANO, Y., SHOTO, E., TSAKA, S., and SASAKI, Y. (1998), Radon, Cl and SO4 anomalies in hot-spring water associated with the 1995 earthquake swarm off the east coast of the Izu Peninsula, Central Japan, Appl Geochem. 13, 89–94. O’NEIL, J.R. and KING, C.-Y. (1981), Variations in stable-isotope ratios of ground waters in seismically active regions of California, Geophys. Res. Lett. 8, 429–432. ¨ ZCAN, M. and U ¨ NAY, T. (1978), Bolu-Seben-Keseno¨zu¨ kaplıcası ve c¸evresi jeotermal enerji aramaları rezistivite O etu¨du¨ raporu, MTA Report 8269, Ankara (in Turkish). PADRO´N, E., MELIA´N, G., MARRERO, R., NOLASCO, D., BARRANCOS, J., HERNA´NDEZ, P.A., and PE´REZ, N.M. (2007), Changes on diffuse CO2 emission and relation to seismic activity in and around El Hierro, Canary Islands, Spain. Pure Appl. Geophys. Topical Volume, Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume II, in press. PE´REZ, N.M., SALAZAR, J.M.L., IGARASHI, G., WAKITA, H., and NAKAI, S. (1998), Precursory geochemical signatures of recent earthquakes in Galicia, Spain, Internat Workshop on Frontiers in Monitoring Science and Technology for Earthquake Environments, Tono Geoscience Center, JNC, Japan, November, 16–19, 1998. PE´REZ, N.M., SALAZAR, J.M.L., HERNA´NDEZ, P.A., MELIA´N, G., HERRERA, M., and STRAUCH, W. (2003), Precursory diffuse CO2 emission signature of the 2002 magnitude 4.2 Managua earthquake, Nicaragua, 8th Workshop on Volcanic Gases-CCVG 2003 Abstracts, 58. PE´REZ, N.M. and HERNA´NDEZ, P.A. (2007), Earthquake forecasting research in active volcanic areas by means of diffuse CO2 emission studies. In Geochemical Precursors for Earthquakes, MacMillan Advan. Res. Ser. 94–103. S¸ AHINCI, A. (1970), Mudurnu kaplıcası hidrojeoloji etu¨d raporu, MTA Report 4367, Ankara (in Turkish). SALAZAR, J.M.L., PE´REZ, N.M., HERNA´NDEZ, P.A., SORIANO, T., BARAHONA, F., OLMOS, R., CARTAGENA, R., LO´PEZ, D., LIMA, N., MELIA´N, G., GALINDO, I., PADRO´N, E., SUMINO, H., and NOTSU, K. (2002), Precursory diffuse carbon dioxide degassing signature related to a 5.1 magnitude earthquake In El Salvador, Central America, Earth Plan. Sci. Lett. 205, 81–89. SANO, Y., NAKAMURA, Y., WAKITA, H., NOTSU, K., and KOBAYASHI, Y. (1986), 3He/4He ratio anomalies associated with the 1984 western Nagano earthquake: Possibly induced by a diapiric magma, J. Geophys. Res. 91 / B12, 12291–12295. SANO, Y., TAKAHATA, N., IGARASHI, G., KOIZUMI, N., and STURCHIO, N.C. (1998), Helium degassing related to the Kobe earthquake, Chem. Geol. 150, 171–179. SATO, M., SUTTON, A.J., MCGEE, K.A., and RUSSEL-ROBINSON, S. (1986), Monitoring of hydrogen along the San Andreas and Calaveras faults in Central California in 1980–1984, J. Geophys. Res. 91 (B12), 12315–12326. S¸ ENGO¨R, A.M.C., GO¨RU¨R, N., and S¸ AROG˘LU, F. (1985), Strike-slip faulting and related basin formation in zones of tectonic escape: Turkey as a case study, The Society of Economic Paleontologists and Mineralogists, Special Publication No. 37, 227–264. SILVER, P. and WAKITA, H. (1996), A search for earthquake precursors, Science 273, 77–78. SONG, S.R., KU, W.Y., CHEN, Y.L., LIU, C.M., CHEN, H.F., CHA, P.S., CHEN, Y.G., YANG, T.F., CHEN, C.H., LIU, T.K., and LEE, M. (2006), Hydrogeochemical anomalies in the springs of the Chiayi area in west-central Taiwan as possible precursors to earthquakes, Pure Appl. Geophys. 163, 675–691. SOREY, M.L., KENNEDY, B.M., EVANS, W.C., FARRAR, C.D., and SUEMNICHT, G.A. (1993), Helium isotope and gas discharge variations associated with crustal unrest in Long Valley Caldera, California, 1989–1992, J. Geophys. Res. 98 (B9), 15871–15889. SUGISAKI, R. (1978), Changing He/Ar and N2/Ar ratio of fault air may be earthquake precursors, Nature 275, 209–211. SUGISAKI, R., and SUGIURA, T. (1986), Gas anomalies at three mineral springs and a fumarole before an inland earthquake, Central Japan, J. Geophys. Res. 91 (B12), 12296–12304.

Vol. 165, 2008

Monitoring of Geothermal Waters along NAFZ, Turkey

43

SUGISAKI, R., ITO, K., NAGAMINE, K., and KAWABE, I. (1996), Gas geochemical changes at mineral springs associated with the 1995 southern Hyogo earthquake (M = 7,2), Japan, Earth Plan. Sci. Lett. 139, 239–249. TATAR, O., PIPER, J.D.A., PARK, R.G., and GU¨RSOY, H. (1996), Paleomagnetic study of block rotations in the Niksar overlap region of the North Anatolian Fault Zone, Central Turkey, Tectonophysics 244, 251–266. TENG, T. L., and SUN, L.F. (1986), Research on groundwater radon as a fluid phase precursor to earthquakes, J. Geophys. Res. 91 (B12), 12305–12313. THOMAS, D. (1988), Geochemical precursors to seismic activity, Pure Appl. Geophys. 126, 241–266. TOUTAIN, J.P., and BAUBRON, J.C. (1999), Gas geochemistry and seismotectonics: A Review, Tectonophysics 304, 1–27. TOUTAIN, J.P., MUNOZ, M., POITRASSON, F., and LIENARD, A.C. (1997), Spring water chloride ion anomaly prior to a M = 5.2 Pyrenean earthquake, Earth Plan. Sci. Lett. 149, 113–119. TSUNOGAI, U. and WAKITA, H. (1995), Precursory chemical changes in groundwater: Kobe earthquake, Japan, Science 269, 61–63. VIRK, H.S., and SINGH, B. (1993), Radon anomalies in soil-gas and groundwater as earthquake precursor phenomenon, Tectonophysics 227, 215–224. VIRK, H.S., WALIA, V., and KUMAR, N. (2001), Helium/Radon precursory anomalies of Chamoli earthquake, Garhwal Himalaya, India, J. Geodyn. 31, 201–210. WAKITA, H., NAKAMURA, Y., KITA, J., FUJII, N., and NOTSU, K. (1980), Hydrogen release: New indication of fault activity, Science 210, 188–190. WAKITA, H., IGARASHI, G., NAKAMURA, Y., and NOTSU, K. (1989), Coseismic radon changes in groundwater, Geophys. Res. Lett. 16, 417–420. WAKITA, H. (1996), Geochemical challenge to earthquake prediction, Proc. Natl. Acad. Sci., USA 93, 3781– 3786. WALIA, V., VIRK, H.S., and BAJWA, B.S. (2006), Radon precursory signals for some earthquakes of magnitude > 5 occurred in NW Himalaya: an overview, Pure and Appl. Geophys. Topical Volume, Terrestrial Fluids, Earthquakes and Volcanoes: The Hiroshi Wakita Volume I, 711–722. ZHANG, W. (1994), Research on hydrogeochemical precursors of earthquakes, J. Earth Prediction Res. 3, 170– 182. (Received October 2, 2007, revised October 11, 2007, accepted October 17, 2007)

To access this journal online: www.birkhauser.ch/pageoph

Pure appl. geophys. 165 (2008) 45–61 0033–4553/08/010045–17 DOI 10.1007/s00024-007-0288-2

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

Coupling Between Seismic Activity and Hydrogeochemistry at the Shillong Plateau, Northeastern India ALASDAIR SKELTON,1 LILLEMOR CLAESSON,1 GOVINDA CHAKRAPANI,2 CHANDAN MAHANTA,3 JOYANTO ROUTH,1 MAGNUS MO¨RTH,1 and PARAM KHANNA4

Abstract—Transient hydrogeochemical anomalies were detected in a granite-hosted aquifer, which is located at a depth of 110 m, north of the Shillong Plateau, Assam, India, where groundwater chemistry is mainly buffered by feldspar alteration to kaolinite. Their onsets preceded moderate earthquakes on December 9, 2004 (MW = 5.3) and February 15, 2005 (MW = 5.0), respectively, 206 and 213 km from the aquifer. The ratios [Na+K]/Si, Na/K and [Na+K]/Ca, conductivity, alkalinity and chloride concentration began increasing 3–5 weeks before the MW = 5.3 earthquake. By comparison with field, experimental and theoretical studies, we interpret a transient switchover between source aquifers, which induced an influx of groundwater from a second aquifer, where groundwater chemistry was dominantly buffered by the alteration of feldspar to smectite. This could have occurred in response to fracturing of a hydrological barrier. The ratio Ba/Sr began decreasing 3–6 days before the MW = 5.0 earthquake. We interpret a transient switchover to anorthite dissolution caused by exposure of fresh plagioclase to groundwater interaction. This could have been induced by microfracturing, locally within the main aquifer. By comparison with experimental studies of feldspar dissolution, we interpret that hydrogeochemical recovery was facilitated by groundwater interaction and clay mineralization, which could have been coupled with fracture sealing. The coincidence in timing of these two hydrogeochemical events with the only two MW C 5 earthquakes in the study area argues in favor of cause-and-effect seismichydrogeochemical coupling. However, reasons for ambiguity include the lack of similar hydrogeochemical anomalies coupled with smaller seismic events near the monitoring station, the >200 km length scale of inferred seismic-hydrogeochemical coupling, and the potential for far-field effects related to the Great Sumatra– Andaman Islands Earthquake of December 26, 2004. Key words: Hydrogeochemistry, seismic-hydrogeochemical coupling, water-rock interaction, Shillong Plateau, India.

1. Introduction Changes in groundwater chemistry have been reported before and after several earthquakes (ULOMOV and MAVASHEV, 1971; IGARASHI et al., 1995; TSUNOGAI and WAKITA, 1995; CLAESSON et al., 2004; SILVER and WAKITA, 1996). These include pre-seismic 1 2 3 4

Department of Geology and Geochemistry, Stockholm University, 106 91 Stockholm, Sweden. Department of Earth Sciences, Indian Institute of Technology, Roorkee 247667, India. Department of Civil Engineering, Indian Institute of Technology, Guwahati 781039, India. Wadia Institute of Himalayan Geology, Dehra Dun, India.

46

A. Skelton et al.

Pure appl. geophys.,

changes in the concentrations of Rn, CO2, CH4, N2, H2, He, Cl, F, SO4, Fe, Cr, Mn, Zn and Cu and post-sesimic changes in the concentrations of B, Ca, K, Li, Mo, Na, Rb, S, Si, Sr, Cl, and SO4 (CLAESSON et al., 2004). Such hydrogeochemical changes are often detected far (>100 km) from the earthquake focus (SILVER and WAKITA, 1996). These have been related to a range of processes, but the most widely-accepted are (1) the exposure of fresh mineral surfaces to groundwater interaction and (2) the mixing or switching of source aquifers, induced by pre-seismic damage (SCHOLZ et al., 1973; THOMAS, 1988). However, the mechanism whereby this can occur so far from an earthquake’s focus remains poorly understood (WYSS, 1997; GELLER et al., 1997). Here, we combine (1) hydrogeochemical monitoring, (2) petrographic observation of the granite, which hosts the aquifer, (3) reaction stoichiometry and (4) theoretical considerations, imposed by the critical earthquake model (BUFE and VARNES, 1993; SORNETTE and SAMMIS, 1995; JAUMe´ and SYKES, 1999; JOHANSEN et al., 2000; Zo¨LLER and HAINZL, 2002), to test the hypothesis that hydrogeochemical changes detected in a granite-hosted aquifer, near the Shillong Plateau, northeastern India, were coupled with the only two MW C 5 earthquakes in the region during two years of sampling.

2. Study Area The Brahamaputra valley and the northern Bengal Basin, which flank the 1.6–2 km elevated Shillong Plateau in northeastern India are both densely populated and strongly affected by earthquakes (Fig. 1). The region, which is subjected to frequent M C 5 earthquakes, suffered major loss of life and infrastructural damage caused by the great 1897 Assam earthquake. The Shillong Plateau is interpreted as a ‘pop-up’ structure, which is bounded to the south by a northward-dipping reverse fault (the Dawki fault), and to the north by an inferred buried southward-dipping reverse fault (the Oldham fault) (BILHAM and ENGLAND, 2001; MITRA et al., 2005).

3. Hydrogeochemical Monitoring From October 1, 2004 to December 1, 2005, 171 groundwater samples were collected, on average every 2–3 days at a commercial bottling plant (Silver Drop). The groundwater source is a granite-hosted aquifer at a depth of 110 m located at latitude 26 120 24.6600 and longitude 91 410 27.2800 . Groundwater was pumped from this aquifer at a constant rate of 5000 liters/hour during operational hours (daytime only). The mean pH and temperature of the groundwater, reported at the plant, was 7.7 and 25C, respectively. This location is on the northern flank of the Shillong Plateau and close to the subterranean culmination of the inferred Oldham fault. We extended this time series to include the period December 2003 to September 2004, by analyzing 10 groundwater samples, which had been bottled at the plant for commercial purposes. Groundwater samples were shipped to Stockholm

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

47

Figure 1 Location map (modified from SRIVASTAVA and SINHA, 2004) showing major tectonic features, regional geology, the location of the hydrogeochemical monitoring station (open triangle), north of the Shillong plateau, and the epicenters of the MW = 5.3 earthquake on December 9, 2004 (206 km from Silver Drop), the MW = 5.0 earthquake on February 15, 2005 (213 km from Silver Drop), the mb = 3.9 on January 12, 2004 (29 km from Silver Drop) and the mb = 4.2 September 27, 2004 (52 km from Silver Drop).

University and analyzed for pH, conductivity, alkalinity (by titration), cations: Al, As, B, Ba, Be, Ca, Cd, Ce, Co, Cr, Cu, Eu, Fe, K , La, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Rb, Sc, Si, Sr, Ti, V, Y, Yb, Zn and Zr (using an Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES, Varian Vista Pro Ax) and anions: Cl and SO4 (by ion chromatography, Dionex DX-300). Ion balance errors, computed using PHREEQCI were less than 5% for >70% of the data and less than 10% for >97% of the data. Two years of data were collected to rule out hydrogeochemical variation related to seasonal rainfall variation in the region. This is relatively systematic, ranging from a monthly average of

48

A. Skelton et al.

Pure appl. geophys.,

4 mm in December to 239 mm in July. Unfortunately, we were unable to obtain local meteorological data.

4. Seismic Activity During the study, two MW C 5.0 earthquakes, listed in the HARVARD CMT catalogue (http://www.globalcmt.org/), occurred within a 2 9 2 rectangle, centered on Silver Drop (Fig. 1). These earthquakes occurred south of the Shillong Plateau. On December 9, 2004, a MW = 5.3 earthquake occurred at latitude 24 390 3600 and longitude 92 430 1200 , and, on February 15, 2005, a MW = 5.0 earthquake occurred at latitude 24 310 1200 and longitude 92 360 3600 . The distances between the epicenters of these earthquakes and the sampled aquifer were 206 km and 213 km, respectively. In this study, we aim to test the hypothesis that hydrogeochemical changes detected Silver Drop (see below) were coupled with these two earthquakes. Before proceeding, we must therefore consider both larger events occurring outside our ‘‘study area’’ and smaller events occurring near to Silver Drop. The epicenters of the next nearest MW C 5 and MW C 6 earthquakes listed in the HARVARD CMT catalogue, which occurred during the study were 366 km (MW = 5.2 on March 25, 2005) and 904 km (MW = 6.0 on March 27, 2004) from Silver Drop, respectively. These events were probably too distant to have generated crustal strain at Silver Drop, similar to that generated by the MW C 5 earthquakes on December 9, 2004 and February 15, 2005. No MW C 7 earthquakes occurred within 2000 km of Silver Drop during the study. However, similar crustal strains could have been generated both by the MW = 9.0 (HARVARD CMT) Sumatra–Andaman Islands earthquake on December 26, 2004 (2585 km from Silver Drop) (cf. HILL et al., 1993), and by two of the 26 smaller events (mb < 5) listed for the study area (Table 1) in the NEIC/PDE catalogue (http://earthquake.usgs.gov/regional/neic/). These events occurred on January 12, 2004 (mb = 3.9), 29 km from Silver Drop, and on September 27, 2004 (mb = 4.2), 52 km from Silver Drop.

5. Results Only Al, Ba, Ca, K, Mg, Na, Si, Sr, SO4 and Cl occurred in concentrations that significantly exceeded the limits of analytical detection. There was no temporal variation in the concentrations of Si (34.24 ± 0.05 ppm), Mg (5.84 ± 0.02 ppm), Al (6.5 ± 1.2 ppb) and SO4 (2.28 ± 0.04 ppm). Si data are plotted in Figure 2a. pH remained approximately constant at 7.71 ± 0.05. In November 2004 (3–5 weeks before the MW = 5.3 earthquake), the concentrations of Na and K began increasing from their respective baseline values of 15.7 ± 0.3 ppm and 1.32 ± 0.01 ppm. Respective maxima of 28–37 ppm and 1.47–1.48 ppm were reached in

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

49

Table 1 Earthquakes listed in the NEIC/PDE (http://earthquake.usgs.gov/regional/neic/) and HARVARD CMT (http:// www.globalcmt.org/) catalogues for a 2 9 2 rectangle centered on Silver Drop. Body-wave magnitudes (mb) are given for entries from the NEIC/PDE catalogue (normal text) and moment magnitudes (MW) are given for entries from the HARVARD CMT catalogue (italic text). The final entry is for the Great Sumatra–Andaman Islands Earthquake. Entries for the December 9, 2004 and February 15, 2005 earthquakes for both NEIC/PDE and HARVARD CMT catalogues are listed (bold text) Date

Longitude

Latitude

Depth

mb/MW

Distance

Catalogue

2/12/03 6/12/03 12/1/04 12/5/04 13/7/04 4/8/04 9/8/04 27/9/04 2/11/04 5/11/04 12/11/04 24/11/04 7/12/04 9/12/04 9/12/04 21/1/05 21/1/05 15/2/05 15/2/05 27/2/05 11/3/05 3/5/05 29/5/05 24/6/05 17/7/05 12/9/05 11/11/05 12/12/05 27/12/05 31/12/05 26/12/04

90.39 90.25 91.9 93.44 92.78 90.26 91.8 92.15 92.58 93.71 92.08 90.94 92.43 92.54 92.72 92.66 92.72 92.52 92.61 91.55 90.49 91.06 92.42 93.14 93.39 90.53 93.04 92.31 93.99 90.38 94.26

25.79 25.6 26.36 25.23 26.21 25.92 27.58 26.13 26.44 24.09 27.27 27.33 24.41 24.76 24.66 27.42 27.44 24.55 24.52 25.37 27.34 25.76 26.93 26.48 26.41 25.88 25.46 25.96 24.78 25.73 3.09

23 26 38 57 71 61 16 37 48 56 45 10 72 34 39.4 52 55 35 27.2 24 52 33 56 54 26 35 51 35 86 35 28.6

3.8 4.4 3.9 4.5 4.2 4.2 4.1 4.2 4.2 4.7 4 4 4.3 5.5 5.3 3.5 3.7 5.1 5 4.2 4.4 4.3 4.3 4.3 4.8 4.2 3.7 4.3 4.7 4.2 9.0

152 174 29 223 121 162 153 52 102 325 126 150 216 186 206 173 178 206 213 94 183 86 114 164 190 134 171 74 301 155 2585

NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE HARVARD CMT NEIC/PDE NEIC/PDE NEIC/PDE HARVARD CMT NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE NEIC/PDE HARVARD CMT

January 2005 and the concentrations of Na and K returned to their respective baseline values thereafter (Figs. 2b, c). Similar behavior was seen for alkalinity, conductivity and Cl concentration (Figs. 2g–i). Between February 9 (six days before the MW = 5.0 earthquake) and February 17, 2005 (two days after the earthquake), the concentrations of Ca and Sr increased and the concentration of Ba decreased from respective baseline values of 18.6 ± 0.5 ppm, 90.7 ± 0.4 ppb and 20.4 ± 0.2 ppb, to respective maxima/minima of 28.8 ppm, 138 ppb and 7.4 ppb (Figs. 2d–f). Their concentrations returned to baseline values thereafter.

50

A. Skelton et al.

Pure appl. geophys.,

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

51

b

Figure 2 Time series for (a) Na, (b) K, (c) Ca, (d) Sr, (e) Ba, (f) Si, (g) total alkalinity, (h) conductivity and (i) Cl. The timing of (1) the MW = 5.3 earthquake on December 9, 2004 (206 km from Silver Drop) and the MW = 5.0 earthquake on February 15, 2005 (213 km from Silver Drop) are shown by solid vertical lines; (2) the mb = 3.9 on January 12, 2004 (29 km from Silver Drop) and the mb = 4.2 September 27, 2004 (52 km from Silver Drop) are shown by dotted vertical lines; and (3) the MW = 9.0 Sumatra–Andaman Islands earthquake on December 26, 2004 (2585 km from Silver Drop) is shown by a dashed vertical line.

6. Water-rock Interaction GARRELS and MACKENZIE (1967) showed that granite alteration by interaction with groundwater and therefore groundwater chemistry is depth-dependant. They showed that the chemistry of shallow circulating groundwater is buffered by feldspar alteration to kaolinite, whereas deeper groundwater is buffered by feldspar alteration to smectite. To test whether such behavior can explain hydrogeochemical changes observed in this study, we conducted a petrographic study of altered granite which was collected near the bottling plant. This granite contained 46.6 ± 4.5 modal % quartz, 40.6 ± 4.4 modal % microcline with minor lamellae of albite (Ab>99), 11.8 ± 2.9 modal % plagioclase (zoned from An16 to An3) with patches of K-feldspar and 1.0 ± 0.9 modal % biotite. Petrographic examination and electron microprobe analysis of the granite showed that (i) plagioclase is extensively altered to smectite along narrow fracture surfaces, which exhibit preferred crystallographic orientation, and (ii) plagioclase and microcline are locally altered to aggregates of clay minerals (kaolinite + smectite ± gibbsite) and magnetite along wider non-crystallographic fractures and at grain edges (Fig. 3). Quartz is unaltered and biotite shows only minor alteration to clay minerals along cleavage planes. We interpret intracrystalline alteration of feldspar to smectite by groundwater interaction at deeper levels, overprinted by intercrystalline alteration of feldspar to kaolinite by groundwater interaction at shallower levels. This is similar to the depthdependent alteration of feldspar reported by GARRELS and MACKENZIE (1967).

7. Reaction Stoichiometry The alteration of feldspar to kaolinite and smectite caused by interaction with groundwater can occur by the (simplified) reactions: 2ðNa;KÞAlSi3 O8 þ 2CO2 þ 11H2 O ! Al2 Si2 O5 ðOHÞ4 þ 2ðNa;KÞþ þ 2HCO 3 þ 4H4 SiO4 ðalbite;microclineÞ

ðkaoliniteÞ

½1

52

A. Skelton et al.

Pure appl. geophys.,

Figure 3 Thin section image viewed in polarized light (a) showing preferential alteration of plagioclase adjacent to a fracture in granite collected near the bottling plant. Image size: ca. 5 9 3 mm. Back-scattered electron (BSE) image (b) showing alteration of plagioclase to smectite along narrow, sometimes crystallographically-aligned fractures and kaolinite along wider fractures and at grain edges.

2ðNa;KÞAlSi3 O8 þ 2CO2 þ 6H2 O ! Al2 Si4 O10 ðOHÞ2 þ 2ðNa;KÞþ þ 2HCO 3 þ 2H4 SiO4 ðalbite;microclineÞ

ðsmectiteÞ

½2 CaAl2 Si2 O8 þ 2CO2 þ 2H4 SiO4 ! Al2 Si4 O10 ðOHÞ2 þ Ca2þ þ 2HCO 3 þ 2H2 O ðanorthiteÞ

ðsmectiteÞ

½3

The stoichiometry of reactions [1–3] has been simplified by ignoring (1) structurallybound H2O in kaolinite and smectite, and (2) Na, Ca and Mg in smectite. Because reactions [1] and [2] release [Na + K] and H4SiO4 into solution in the molar ratios 1:2 and 1:1, respectively (see GARRELS and MACKENZIE, 1967) and reaction [3] releases Ca, but no

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

53

H4SiO4 into solution and based on the assumption that feldspar alteration was a dominant control of groundwater chemistry, we studied their time-dependent activities by plotting temporal variations of [Na+K]/Si, Na/K, Ba/Sr and [Na+K]/Ca ratios (Figs. 4a–d). Electron microprobe analysis indicates that smectite in the granite contained small amounts of Na2O (1.2–1.4 wt. %) and CaO (0.8–1.0 wt. %), but negligible MgO. This would reduce the molar ratio of [Na + K] and H4SiO4 released into solution by reaction [2] to * 0.9:1. Feldspar may also have been altered by reaction with hydrochloric acid, but this reaction was volumetrically minor (alkalinity & 20 9 molar Cl), yielding similar stoichiometric ratios for [Na+K]/Si, Na/K, Ba/Sr and [Na+K]/Ca. Finally, the reason for examining the Ba/Sr ratio is that this ratio has been shown to be a particularly sensitive indicator of the relative inputs from K- and Ca-bearing minerals (LAND et al., 2000). Because Ba substitutes for K in microcline and Sr substitutes for Ca in the anorthitic component of plagioclase, the Ba/Sr ratio was used to compare the dissolution rates of microcline by reaction [1] and anorthitic plagioclase by reaction [3].

8. Hydrogeochemical Events

In November 2004 (3–5 weeks before the MW = 5.3 earthquake), the molar [Na+K]/Si ratio increased from its baseline value of 0.58 ± 0.01, reaching a maximum of 1.0–1.4 in January 2005 (Fig. 4a). This is consistent with a switchover between source aquifers, with groundwater in the main aquifer dominantly buffered by reaction [1] and groundwater in the second aquifer dominantly buffered by reaction [2]. This could occur if (for example) the second aquifer was at a deeper level or if the residence time for groundwater in this aquifer was longer. This switchover can be visualized on a plot of molar Na+/Ca2+ against molar HCO–3/H4SiO4 after GARRELS (1967) (Fig. 5). The interpretation of an open system switchover between groundwater aquifers is further supported by the following observations: (1) The concentration maximum for [Na + K] was not mirrored by a concentration minimum for Si (Figs. 2a-c). This would have been expected for a closed system switchover between reactions [1] and [2] (cf., GARRELS and MACKENZIE, 1967). (2) Clay mineral aggregates occupy fractures in plagioclase and microcline (Fig. 3). These could record switchover(s) between chemically-distinct groundwater aquifers during clay mineral growth, but they might alternatively record simultaneous growth of two or more clay minerals. Possible mechanisms which could cause the interpreted switchover between groundwater aquifers include (1) fracturing of a hydrological barrier between two aquifers, and (2) a change in the relative pressures between two aquifers (THOMAS, 1988). The [Na+K]/Si maximum coincided with a Na/K maximum of 30–40 (compared with a baseline of 20.1 ± 0.3, Fig. 4b), which is consistent with experimental and theoretical

54

A. Skelton et al.

Pure appl. geophys.,

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

55

b

Figure 4 Groundwater molar [Na+K]/Si, Na/K, Ba/Sr and [Na+K]/Ca ratios. The timing of (1) the MW = 5.3 earthquake on December 9, 2004 (206 km from Silver Drop) and the MW = 5.0 earthquake on February 15, 2005 (213 km from Silver Drop) are shown by solid vertical lines; (2) the mb = 3.9 on January 12, 2004 (29 km from Silver Drop) and the mb = 4.2 September 27, 2004 (52 km from Silver Drop) are shown by dotted vertical lines; and (3) the MW = 9.0 Sumatra–Andaman Islands earthquake on December 26, 2004 (2585 km from Silver Drop) is shown by a dashed vertical line. The wide dot/dashed horizontal lines at (Na+K)/Si = 0.5 and 1 dashed lines are the predicted groundwater molar [Na+K]/Si ratio buffered by reactions [1] and [2], respectively. The narrow dotted horizontal lines denote the 2r limits for all data excluding November 2004 – February 2005. The solid curve is the best-fit to the critical earthquake model for equations [4]. Inset shows the molar Ba/Sr ratio in February, 2005 and the MW = 5.0 earthquake.

studies, showing that sodic feldspar dissolves more rapidly than potassic feldspar (BUSENBERG and CLEMENCY, 1976; LASAGA, 1984; STILLINGS and BRANTLEY, 1995). The Cl maximum (Fig. 2i) with no corresponding maximum in SO4 concentration as reported by (e.g.) TSUNOGAI and WAKITA (1995) might relate to the rapidity of SO4 adsorption on clay minerals. In February 2005, the [Na + K]/Si ratio began decreasing. We suggest that this hydrogeochemical ‘‘recovery’’ records effective isolation of the main aquifer from the second aquifer. Possible mechanisms which could cause its effective isolation include (1)

Figure 5 Plot of molar Na+/Ca2+ against molar HCO-3/H4SiO4 after GARRELS (1967). The inferred ‘‘switchover’’ between source aquifers, with groundwater in the main aquifer dominantly buffered by reaction [1] and groundwater in the second aquifer dominantly buffered by reaction [2] is highlighted.

56

A. Skelton et al.

Pure appl. geophys.,

clogging of fractures by newly-formed clay minerals or (2) readjustment of the relative pressures between the aquifers. Renewed buffering of groundwater chemistry by reaction [1] in the main aquifer during or after its isolation could explain the progressive recovery of the [Na + K]/Si ratio towards1:2. Figure 4a shows that recovery is initially rapid, becoming progressively slower until some steady state is approached. This is of interest, because experimental investigations of the artificial weathering of feldspar record sequential nonparabolic, parabolic and linear kinetic stages (e.g., BUSENBERG and CLEMENCY, 1976). Parabolic kinetics was initially attributed to diffusion across a residual layer formed by non-stoichiometric dissolution. However, this was later shown to be an experimental artifact relating to the initial dissolution of fine and ultrafine particles, produced during sample preparation (HOLDEN and BERNER, 1979). However, because production of fine and ultrafine particles may also occur during fracturing, we suggest that this experimental ‘‘artifact’’ might nevertheless provide a reasonable representation of the hydrogeochemical response to fracturing or microfracturing in a natural aquifer. Furthermore, we note that parabolic kinetics could also reflect rate-limiting diffusion across a layer of newly-formed clay minerals. Based on this analysis, we reach the tentative conclusion that hydrogeochemical recovery in the main aquifer at Silver Drop (Fig. 2) is facilitated by progressive fracture healing related to water-rock interaction and consequent clay mineralization, following initial fracturing of a hydrological barrier between two aquifers. Between February 9 (six days before the MW = 5.0 earthquake) and February 17, 2005 (two days after this earthquake), the Ba/Sr ratio decreased from its baseline of 0.144 ± 0.001, to a minimum value of 0.045. We interpret the transient influence of reaction [3] on the chemistry of the sampled groundwater. This increases the release rates of Ca and Sr (Figs. 2d and e). Figure 2f shows a corresponding decrease in the release rate of Ba. We interpret this as closed system behavior: Reaction [3] consumes CO2 from the system, lowering its activity and slowing the forwards progress of reaction [1] and the release rate of Ba (substituting for K). The lack of a corresponding minimum for K (Fig. 2c) reflects its lower sensitivity to water-rock interaction, compared with Ba (LAND et al., 2000). We suggest that reaction [3] was induced by the exposure of unaltered plagioclase to groundwater. This could be caused by local microfracturing in the main aquifer, and would induce a transient stage during which (anorthitic) plagioclase would be preferentially dissolved (BUSENBERG and CLEMENCY, 1976; STILLINGS and BRANTLEY, 1995). Figure 3 shows preferential dissolution of plagioclase in the granite, which hosts the studied aquifer. The Ba/Sr ratio recovered to its baseline by April 2005. We interpret that this recovery occurred because unaltered anorthite in direct contact with groundwater was exhausted (perhaps due to coating by newly-formed clay minerals) and the release of Ca and Sr by reaction [3] ceased. The availability of CO2 and consequently the release rate of Ba increased. We observe a similar pattern, with rapid initial recovery, becoming progressively slower, until some steady state is approached. We conclude that this event involved closed system water-rock interaction within the main aquifer. Both events are seen on Figure 4d, which shows temporal variation of [Na+K]/Ca. In November 2004, the [Na+K]/Ca ratio increased from its baseline value of 1.54 ± 0.03, to

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

57

a maximum of 2.5–3.8 in January 2005. We interpret that this maximum corresponds to switching between aquifers, resulting in an influx of groundwater from a second aquifer, where groundwater chemistry is buffered by reaction [2]. The [Na+K]/Ca ratio reached a minimum of 0.59 on February 12, 2005. We interpret that this minimum records the transient influence of reaction [3] on the chemistry of groundwater in the main aquifer.

9. Seismic-hydrogeochemical Coupling Coincidence of timing is not proof that a hydrogeochemical shift is caused by an earthquake. Nevertheless, we consider the observation that the respective onsets of two distinct hydrogeochemical shifts preceded the only two MW C 5 earthquakes to occur during two years of sampling, convincing evidence of a cause-and-effect association. As such, it is tempting to relate these hydrogeochemical shifts to pre-seismic damage (causing fracturing in the aquifer region). However, elastic half-space models (OKATA, 1992) indicate that the permanent static strain created by a moderate earthquake (which is a probable upper limit for preseismic strain (JOHNSTON et al., 1987)) will be diminishingly small at a distance of 200 km from its focus. On the other hand, precursory hydrogeochemical and hydrological anomalies are often reported at similar distances from earthquake foci, usually confined within the same fault system (SILVER and WAKITA, 1996; ROELOFFS, 1988). This paradox is addressed by the ‘‘critical earthquake model’’ (BUFE and VARNES, 1993; SORNETTE and SAMMIS, 1995; JAUME´ and SYKES, 1999; JOHANSEN et al., 2000; Zo¨LLER and HAINZL, 2002). In this model, a large earthquake is viewed as the culmination of a sequence of seismic cycles, which occur at increasingly large scales within a volume of crust with lateral dimensions, several times greater than the length of the seismic rupture (see also: JOHANSEN et al., 2000; KNOPOFF et al., 1996; ALLe`GRE et al., 1982). For a moderate earthquake, the average radius of this region may still be less than 200 km (WYSS, 1979). However, we note the existence of a pathway of structural weakness, linking the Dawki and (inferred) Oldham faults (BILHAM and ENGLAND, 2001; MITRA et al., 2005) (Fig. 6), which might extend the affected region (cf., ROELOFFS, 1988). The critical earthquake model predicts that cumulative damage before an earthquake follows a power law of the form (BUFE and VARNES, 1993): D  A þ Bðtc  tÞz ;

½4

where A and B are empirically-determined constants, 0 < z < 1, t is time and the earthquake occurs at time, tc. D is cumulative damage before the earthquake. Both cumulative earthquake frequency (SORNETTE and SAMMIS, 1995) and groundwater ion concentrations (JOHANSEN et al., 2000) have been used as proxies for D. Here, we use preseismic [Na + K]/Si as a proxy for D and determine its goodness-of-fit for equation [4] (Fig. 4a). The best-fit value of tc was December 9, 2004 (± 1 day) (R2 = 0.78). We have not incorporated the log-periodic corrections introduced by SORNETTE and SAMMIS (1995) for two reasons. Firstly, our dataset is insufficient to support the additional variables

58

A. Skelton et al.

Pure appl. geophys.,

Figure 6 Schematic cross section of the Shillong ‘pop-up’ structure (modified from MITRA et al., 2005). The pathway of structural weakness between the epicenter of the MW C 5 earthquakes which occurred on December 12, 2004 and February 15, 2005, south of the Shillong Plateau and the hydrogeochemical monitoring station, links between the Dawki and Oldham faults along the base of the crust.

required (cf. MAIN et al., 1999). Secondly, the validity of these corrections has been questioned (JAUME´ and SYKES, 1999). The applicability of the critical earthquake model to smaller earthquakes (M < 6.5) has also been questioned (Zo¨LLER and HAINZL, 2002). On the other hand, potentially useful results have been obtained for earthquakes as small as mb = 3.5 (BREHM and BRAILE, 1998). The goodness-of-fit of the critical earthquake model to our data and the accuracy with which tc is estimated supports the interpretation of a causal association between the MW = 5.3 earthquake on December 9, 2004 and the first hydrogeochemical anomaly. This would imply that this anomaly occurred in response to a switchover between source aquifers triggered by pre-seismic fracturing. However, we cannot unequivocally exclude far-field effects related to the Great Sumatra–Andaman Islands Earthquake on December 26, 2004 (cf. HILL et al., 1993). On the other hand, precursory changes 2000 km from the epicenter seem unlikely. The lack of any measurable hydrogeochemical signal coupled with the two earthquakes which occurred nearest to Silver Drop on January 12, 2004 (mb = 3.9) and September 27, 2004 (mb = 4.2) weakens the argument for coupling between seismicity and hydrogeochemistry. The resolution of the Ba/Sr data is insufficient to perform a similar analysis for the second hydrogeochemical anomaly. However, the shorter duration of this anomaly and the closed system hydrogeochemical behavior are consistent with weaker pre-seismic damage associated with the second and smaller earthquake. This would imply that this anomaly occurred in response to limited pre-seismic microfracturing, locally within the main aquifer. On the other hand, inferred coupling between the second hydrogeochemical anomaly and the smaller earthquake on February 15, 2005, raises further doubts as to why other events did not yield measurable hydrogeochemical signals.

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

59

10. Conclusion We conclude that (1) prior to November 2004, groundwater chemistry in the main aquifer was dominantly buffered by the alteration of feldspar to kaolinite, (2) the first hydrogeochemical anomaly, likely recorded a transient switchover between source aquifers, which induced an influx of groundwater from a second aquifer, where groundwater chemistry was dominantly buffered by the alteration of feldspar to smectite, (3) this could have been coupled with the MW = 5.3 earthquake on December 9, 2004, (4) the second hydrogeochemical anomaly likely recorded a transient switchover to anorthite dissolution induced by exposure of fresh plagioclase to groundwater interaction, (5) this could have been caused by microfracturing in the main aquifer coupled with the MW = 5.0 earthquake on February 15, 2005, and (6) subsequent hydrogeochemical recovery was facilitated by groundwater interaction and clay mineralization, which could have been coupled with post-seismic fracture sealing. The main argument in support of seismic-hydrogeochemical coupling is the coincidence in timing of two hydrogeochemical events with two MW C 5 earthquakes. Reasons for ambiguity include the lack of similar hydrogeochemical anomalies temporally coupled with other seismic events, the >200-km length scale of inferred seismichydrogeochemical coupling, and the potential for far-field effects related to the Great Sumatra–Andaman Islands Earthquake of December 26, 2004. The hydrogeochemical anomalies reported in this study meet some of the validation criteria of the IASPEI (International Association of Seismology and Physics of the Earth’s Interior) subcommission on earthquake prediction (WYSS, 1991, 1997) in that a relation to pre-seismic stress and that some dependence on distance from the earthquake foci is inferred (Table 1). However, hydrogeochemical data was collected from only one site, and even although the hydrogeochemical anomalies are recorded using several instrumental methods the reported anomalies ([Na+K]/Si, Ba/Sr, conductivity, alkalinity) are not truly independent of one another. We thus suggest that to resolve these and similar ambiguities, and to confirm a cause-and-effect association between seismicity and hydrogeochemistry, requires extended time series hydrogeochemical monitoring, conducted simultaneously at several sites, preferably in several seismically-active regions. With respect to site selection, we note that the first hydrogeochemical anomaly detected at Silver Drop caused a measurable change in conductivity and is thus amenable to low cost online monitoring.

Acknowledgements Colin Graham and Heiko Woith are thanked for constructive scientific input to this manuscript. Klara Hajnal and Heike Siegmund are thanked for analytical work. Pedro A. Herna´ndez Pe´rez and an anonymous reviewer are thanked for constructive comments which improved an earlier version of this manuscript. This research was supported by VR SIDA and Carl Tryggers Stiftelse.

60

A. Skelton et al.

Pure appl. geophys.,

REFERENCES ALLE`GRE, C.J., LE MOUEL, J. L., and PROVOST, A. (1982), Scaling rules in rock fracture and possible implications for earthquake prediction, Nature 297, 47–49. BILHAM, R. and ENGLAND, P. (2001), Plateau ‘pop-up’ in the great 1897 Assam earthquake, Nature 410, 806– 809. BUFE, C.G. and VARNES, D. J. (1993), Predictive modeling of the seismic cycle in the Greater San Francisco Bay Region, J. Geophys. Res. 98, 9871–9983. BUSENBERG, E. and CLEMENCY, C.V. (1976), The dissolution kinetics of feldspars at 25C and 1 atm CO2 partial pressure, Geochim. Cosmochim. Acta 40, 41–49. CLAESSON, L., SKELTON, A., GRAHAM, G., DIETL, C., Mo¨RTH, M., TORSSANDER, P., and KOCKUM, I. (2004), Hydrogeochemical changes before and after a major earthquake, Geology 32, 641–644. GARRELS, R.M. Genesis of some ground waters from igneous rocks. In Researches in Geochemistry, Vol. II (ed. P.H. Abelson) (Wiley & Sons, New York 1967), pp. 405–420. GARRELS, R.M. and MACKENZIE, F.T. (1967), Origin of the chemical composition of some springs and lakes, Adv. Chem. Ser. 67, 222–242. GELLER, R.J., JACKSON, D.D., KAGAN, Y.Y., and MULARGIA, F. (1997), Earthquakes cannot be predicted, Science 275, 1616. HILL, D.P. et al. (1993), Seismicity remotely triggered by the magnitude 7.3 Landers, California earthquake, Science 260, 1617–1623. HOLDEN, G.R. and BERNER, R.A. (1979), Mechanisms of feldspar weathering – I. Experimental studies, Geochim. Cosmochim. Acta 43, 1161–1171. IGARASHI, G., SAEKI, S., TAKAHATA, N., SUMIKAWAY, K., TASAKA, S., SASAKI, Y., TAKAHASHI, M., and SANO, Y. (1995), Ground-water radon anomaly before the kobe earthquake in japan, Science 269, 60–61. JAUME´, S.C. and SYKES, L.R. (1999), Evolving towards a critical point: A review of accelerating seismic moment/ energy release prior to large and great earthquakes, J. Appl. Geophys. 155, 279–305. JOHANSEN, A., SALEUR, H., and SORNETTE, D. (2000), New evidence of earthquake precursory phenomena in the 17 January 1995 Kobe earthquake, Japan, Eur. Phys. J. B 15, 551–555. JOHNSTON, M.J.S., LINDE, A.T., GLADWIN, M.T., and BORCHERDT, R.D. (1987), Fault failure with moderate earthquakes, Tectonophys. 144, 189–206. KNOPOFF, L., LEVSHINA, T., KEILIS-BOROK, V. I., and MATTONI, C. (1996), Increased long-range intermediatemagnitude earthquake activity prior to strong earthquakes in California, J. Geophys. Res. 101, 5779–5796. ¨ HLANDER, B. (2000), Ba/Sr, Ca/Sr and 87Sr/86Sr ratios in soil water LAND, M., INGRI, J., ANDERSSON, P.S., and O and groundwater: Implications for relative contributions to stream water discharge, Appl. Geochem. 15, 311– 325. LASAGA, A.C. (1984), Chemical kinetics of water-rock interaction, J. Geophys. Res. 89, 4009–4025. MAIN, I.G., LEONARD, T., PAPASOULIOTIS, O., HATTON, C.G., and MEREDITH, P.G., (1999), One slope or two? Detecting statistically significant breaks of slope in geophysical data, with application to fracture scaling relationships, Geophys. Res. Lett. 26, 2801–2804. MITRA, S., PRIESTLEY, K., BHATTACHARYA, A. K., and Gaur, V.R. (2005), Crustal structure and earthquake focal depths beneath northeastern India and southern Tibet, Geophys. J. Int. 160, 227–248. OKADA, Y. (1992), Internal deformation due to shear and tensile faults in a half-space, Bull. Seismol. Soc. Am. 82, 1018–1040. ROELEFFS, E.A. (1988), Hydrological precursors to earthquakes: A review, Pure Appl. Geophys. 126, 177–209. SCHOLZ, C.H., SYKES, L.R., and AGGARWAL, Y.P. (1973), Earthquake prediction: A Physical Basis, Science 181, 803–810. SILVER, P.G. and WAKITA, H. (1996), A search for earthquake precursors, Science, 273, 77–78. SORNETTE, D. and SAMMIS, C.G. (1995), Complex critical exponents from renormalization group theory of earthquakes: Implications for earthquake predictions, J. Phys. Int. France 5, 607–619. SRIVASTAVA, R.K. and SINHA, A.K. (2004), Early Cretaceous Sung Valley ultramafic-alkaline-carbonatite complex, Shillong Plateau, Northeastern India: Petrological and genetic significance, Mineral. Petrol. 80, 241–263. STILLINGS, L.L. and BRANTLEY, S.L. (1995), Feldspar dissolution at 25C and pH 3: Reaction stoichiometry and the effect of cations, Geochim. Cosmochim. Acta 59, 1483–1496.

Vol. 165, 2008

Coupling Between Seismic Activity and Hydrogeochemistry

61

THOMAS, D. (1988), Geochemical precursors to seismic activity, Pure Appl. Geophys. 126, 241–266. TSUNOGAI, U. and WAKITA, H. (1995), Precursory chemical changes in groundwater: Kobe earthquake, Japan, Science 269, 61–63. ULOMOV, V.I. and MAVASHEV, B.Z. (1971), The Tashkent Earthquake of 26 April, Tashkent, Akad. Nauk Uzbek. SSR, FAN, 188–192. WYSS, M. (1979), Estimating the maximum expectable magnitude of earthquakes from fault dimensions, Geology 7, 336–340. WYSS, M, Evaluation of Proposed Earthquake Precursors (ed. M. Wyss), (AGU, Washington DC, 1991), 94 pp. WYSS, M. (1997), Second round evaluation of proposed earthquake precursors, Pure Appl. Geophys. 149, 3–16. Zo¨LLER, G., and HAINZL, S. (2002), A systematic spatiotemporal test of the critical point hypothesis for large earthquakes, Geophys. Res. Lett. 29, 1558. (Received February 13, 2007, revised November 10, 2007, accepted December 10, 2007) Published Online First: February 1, 2008

To access this journal online: www.birkhauser.ch/pageoph

Pure appl. geophys. 165 (2008) 63–74 0033–4553/08/010063–12 DOI 10.1007/s00024-007-0292-6

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

Radon Changes Associated with the Earthquake Sequence in June 2000 in the South Iceland Seismic Zone ´ STA RUT HJARTARDO´TTIR,1 and PA´LL EINARSSON,1 PA´LL THEODO´RSSON,2 A GUðO´N I GUðJO´NSSON2

Abstract—An earthquake sequence at the transform plate boundary in South Iceland, that included two magnitude 6.5 earthquakes in June 2000, was anticipated on the basis of a centuries-long seismicity pattern in the area. A program of radon monitoring in geothermal water from drill holes, initiated in 1999, rendered distinct and consistent variations in radon in association with these events. All seven sampling stations in a 50 9 30 km zone covering the epicentral area showed a consistent pattern. Four types of change could be identified: 1) Preseismic decrease of radon. Anomalously low values were measured 101–167 days before the earthquakes. 2) Preseismic increase. Spikes appear in the time series at six stations 40–144 days prior to the earthquakes. These anomalies were large and unusual if compared to a 17-years history of radon monitoring in this area. 3) Coseismic step, most likely related to the coseismic change in groundwater pressure observed over the entire area. 4) Postseismic return of the radon values to the preseismic level about three months later, also concurrent with groundwater pressure changes. Key words: South Iceland Seismic Zone, radon, earthquake precursor, co-scismic changes.

1. Introduction Various studies have shown that an increase in the concentration of radon in groundwater is an earthquake precursor, see e.g., reviews by HAUKSSON (1981) and KING (1985), and more recent studies by WAKITA (1996), ROELOFFS (1999), TRIQUE et al. (1999), and ZMAZEK et al. (2002). Even though numerous examples of premonitory radon anomalies have been identified and described in the literature, statistical analysis of the relationship between radon and earthquakes has been difficult because of the lack of long-time series from a network of recording stations in active seismic or volcanic areas. A program of radon monitoring was initiated for this purpose in the plate boundary areas of Iceland in 1977 (HAUKSSON and GODDARD, 1981; HAUKSSON, 1981).

1

Institute of Earth Sciences, University of Iceland, Sturlugata 7, 101 Reykjavı´k, Iceland. E-mail: [email protected] 2 Science Institute, University of Iceland, Dunhaga 3, 107 Reykjavı´k, Iceland.

64

P. Einarsson et al.

Pure appl. geophys.,

A network of 11 sampling stations was established, nine of which were located within or near the South Iceland Seismic Zone; a transform zone where the most damaging historical earthquakes in Iceland have occurred. Judged from the time pattern of previous events a sequence of large earthquakes was expected in this zone with high probability (STEFA´NSSON et al., 1993; EINARSSON et al., 1981). This labor-intensive radon program gave promising results (Jo´NSSON and EINARSSON, 1996) but was discontinued in 1993 because of declining funding and deteriorating instruments. Radon monitoring was resumed in 1999 with new and improved instruments and time-saving sample preparation (THEODo´RSSON, 1996; GUDJo´NSSON and THEODo´RSSON, 2000) during which the sample preparation time was reduced from 3 hours to less than 10 minutes. The expected earthquakes occurred in June 2000 (EINARSSON et al., 2000; STEFa´NSSON et al., 2000) within the network of monitoring stations and after one year of radon monitoring. The earthquake sequence occurred along a 90-km stretch of the plate boundary and contained two magnitude 6.5 (Mx) earthquakes (on June 17 and 21) and several ´ RNADo´TTIR et al., 2004). This paper magnitude 5+ events (see e.g., PEDERSEN et al., 2003; A documents the radon time series and reports on the radon changes detected in association with these events.

2. The South Iceland Seismic Zone The South Iceland Seismic Zone is a transform-type plate boundary; a branch of the mid-Atlantic plate boundary that crosses Iceland (Fig. 1). Plate divergence in the southern part of Iceland is accommodated by two sub-parallel rift zones: the Western and the Eastern Volcanic Zones. The gap between them is bridged in the south, near 64N, by a zone of high seismic activity, the South Iceland Seismic Zone, which takes up the transform motion between the Reykjanes Ridge and the Eastern Volcanic Zone (EINARSSON, 1991). The two rift zones and the transform demarcate a block or a microplate, the Hreppar microplate. It has been argued that rifting is dying out in the Western Rift Zone, and is being taken over by the Eastern Rift Zone, or that the partition of rifting between the rift zones may be uneven and changes with time (SIGMUNDSSON et al., 1995). The South Iceland Seismic Zone has been defined by destruction areas of historical earthquakes, Holocene surface ruptures and instrumentally determined epicenters. It is oriented E-W and is 10–15 km wide. Destruction areas of individual earthquakes and surface faulting (Fig. 2) show, however, that each event is associated with faulting on N-S striking planes, perpendicular to the main zone. The overall left-lateral transform motion along the zone, i.e., between the Hreppar microplate to the north and the Eurasia plate to the south, thus appears to be accommodated by right-lateral faulting on many parallel, transverse faults and counter-clockwise rotation of the blocks between them, ‘‘bookshelf faulting’’ (EINARSSON et al., 1981).

Vol. 165, 2008

Radon Changes and the South Iceland Earthquakes of June 2000

65

Figure 1 The active plate boundary of Iceland passes near the center of the Iceland hotspot. The radon program is centered on the transform zone of South Iceland, SISZ. The Western and Eastern Volcanic Rift Zones are marked with W and E, respectively. Arrows show the direction of relative plate movements across the plate boundaries. The rate is 18.5 to 19.5 mm/year.

Earthquakes in South Iceland tend to occur in major sequences in which most of the zone is affected. These sequences last from a few days to about three years. Each sequence typically begins with a magnitude 7 event in the eastern part of the zone, followed by smaller events farther west. Sequences of this type occurred in 1896, 1784, 1732–34, 1630–33, 1389–91, 1339 and 1294. Apart from the historic gap between 1391 and 1630, the sequences thus occur at intervals that range between 45 and 112 years (EINARSSON et al., 1981), and it has been argued that a complete strain release of the whole zone is accomplished in about 140 years (STEFA´NSSON and HALLDo´RSSON, 1988). The lengthy time since the last sequence led to a long-term forecast published in 1985 (EINARSSON, 1985), later refined by STEFA´NSSON et al. (1993), of a major earthquake sequence within the next decades. The original forecast gave a 80% probability for the occurrence of a major earthquake sequence within the next 25 years (i.e., within the 1985–2010 time window). The magnitude of the first event was estimated in the range 6.3–7.5 and the most likely location was given in the eastern part of the seismic zone. In the refined version the magnitude range was reduced to 6.3–7.0 and two seismic gaps were identified, at 20.3W and 20.7W. The forecast was fulfilled in June 2000 when

66

P. Einarsson et al.

Pure appl. geophys.,

Figure 2 The location of the radon sampling stations within the South Iceland Seismic Zone shown with triangles. Thin lines show Holocene surface fractures, formed in earthquakes before 2000. The two long, thick lines show the source faults of the two earthquakes of June 17 and 21, 2000 as delineated by aftershocks; the former one on the easternmost fault, the latter on the western fault. The sense of faulting was right-lateral strike-slip on both faults. Elevation contours are at 50 m intervals.

two magnitude 6.5 events occurred in the zone; one at 20.37W and the other at 20.71W.

3. The 2000 Earthquake Sequence The earthquakes of 2000 were the largest in the zone since 1896 and 1912. They occurred within the SIL-network of seismographs operated by the Icelandic Meteorological Office (see e.g., website http://www.vedur.is/, STEFA´NSSON et al. 1993). The sequence began on June 17 at 15:40 with a magnitude 6.5 event in the eastern part of the zone (Fig. 2). This immediately triggered a flurry of activity along at least a 90-km-long stretch of the plate boundary to the west, apparently triggered by the passing S waves from the first event. Among them was an event with an anomalously low seismic radiation but a ´ RNADo´TTIR et al., 2004). An earthquake of moment equivalent to a magnitude 5.9 (A magnitude 5.7 (mb) followed two minutes later on a small, parallel fault, about 3–4 km to the west of the first shock. An event of magnitude 4.9 (mb) then occurred on the Reykjanes Peninsula, 90 km to the west, about 5 minutes after the first shock. Several other significant shocks also occurred this day along this segment of the plate boundary (PAGLI et al., 2003). A second mainshock similar in magnitude to the first event occurred about 20 km west of the first one on June 21 at 00:51. It was clearly preceded by a clustering of ´ RNADo´TTIR microearthquakes along the eventual source fault (STEFA´NSSON et al., 2000). A et al. (2003, 2004) present evidence that triggering played a large role in the occurrence of

Vol. 165, 2008

Radon Changes and the South Iceland Earthquakes of June 2000

67

events within the sequence, both dynamic triggering by the passing waves and triggering by regional changes in the Coulomb stress due to faulting. The aftershock distribution, moment tensor inversions, distribution of surface faulting, and modeling of surface deformation measured by GPS and InSAR confirm that the mainshocks of the sequence occurred on N-S striking faults, transverse to the zone itself (CLIFTON and EINARSSON, 2005; ´ RNADo´TTIR et al., 2001; PEDERSEN et al., 2003). The sense of faulting was right-lateral A strike-slip conforming to the model of ‘‘bookshelf faulting’’ for the South Iceland Seismic Zone. The two mainshocks occurred on pre-existing faults and were accompanied by surface ruptures consisting primarily of en-e´chelon tension gashes and push-up structures (CLIFTON and EINARSSON, 2005). The main zones of rupture were about 15 km long and fractured the crust down to 10 km depth. The surface faults coincided with the epicentral distributions of aftershocks. Fault displacements were of the order of 0.1–1 m. Faulting along conjugate, left-lateral strike-slip faults also occurred, but was less pronounced than that of the main rupture zones. The maximum fault displacement at depth, determined by modeling of geodetic data, was 2–2.5 m. The source faults of the two largest earthquakes are shown in Figure 2. Large hydrological changes were observed in a wide area surrounding the seismically active zone. Pressure changes in boreholes followed a regular pattern conforming with crustal stress changes (BJo¨RNSSON et al., 2001; Jo´NSSON et al., 2003). Pressure decreased in areas to the NE and SW of the epicenters but increased in the quadrants to the NW and SE. These changes were large, but were reversed and equilibrated in less than three months. A post-earthquake crustal deformation signal was detected by InSAR that correlates with the water pressure changes (Jo´NSSON et al., 2003).

4. Previous Radon Studies in South Iceland The relationship between radon and earthquakes has been studied in this area since 1977, when the first equipment for this purpose was installed. The instruments were operated until 1993. The radon monitoring network consisted of up to 9 stations. Samples (0.6 l) of geothermal water were collected from drill holes every few weeks and sent to the laboratory for radon analysis. The resulting time series varied in length from 3 to 16 years. Many earthquake-related radon anomalies were identified (Jo´NSSON and EINARSSON, 1996). They are represented by both positive and negative excursions from the mean values, and occur mostly prior to the seismic events, i.e., within a few weeks. For a statistical analysis of the anomalies and comparison with the seismicity time series, significant earthquakes were selected according to the criteria of HAUKSSON and GODDARD (1981): M  2:4 log D  0:43

and

M  2;

ð1Þ

where M is the magnitude and D is the distance to a radon monitoring station. Thus 98 independent seismic events were selected. They were in the magnitude range 2–5.8. The main conclusions were as follows:

68

P. Einarsson et al.

Pure appl. geophys.,

1. 2. 3. 4.

Radon anomalies were observed before 30 of the significant events. 35% of all observed anomalies were related to seismicity. 80% of the anomalies observed before earthquakes were positive. If a positive anomaly is detected at one station, the probability of a significant earthquake occurring afterwards is 38%. 5. Some sampling sites were found to be more sensitive than others. The sensitivity appears to depend on local geological conditions. 6. A few radon anomalies appeared to be related to eruptive activity of the neighboring Hekla volcano.

5. Revival of the Radon Monitoring A new radon program was initiated in 1999 using a new time-saving technique and an instrument developed at our institute. It is based on a novel liquid scintillation technique where counting only Bi-218/Po-218 pulse pairs gives high sensitivity with a simple construction (THEODo´RSSON, 1996; GUDJo´NSSON and THEODo´RSSON, 2000). Scintillations other than those associated with radon-222 are thus excluded. Samples of 200 ml are taken from the geothermal drill holes at the sampling sites (Table 1) and analyzed in the laboratory. About 60% of the radon from the water samples is transferred to a scintillator in a 15 ml liquid scintillation counting vial by circulating and bubbling air for four minutes between the two liquids. The scintillation liquid is mineral oil. The scintillations are subsequently counted in a laboratory-made automatic sample changer. 226 counts per hour correspond to 1 Bq/l of radon in the water. The system represents a significant progress in the radon measuring technique where high sensitivity is needed. The technique also saves considerable time compared to previous procedures. Sample preparation time was reduced from 3 hours to less than 10 minutes. Sampling from geothermal wells in the South Iceland Seismic Zone began a year before the destructive earthquakes of June 2000 occurred. Water samples were taken

Table 1 Location, depth and temperature of geothermal holes sampled for radon Station

Latitude

Longitude

Depth, m

Teperature, C

Bakki ¨ xnalækur O Selfoss, hole 13 Hlemmiskeið Flu´ðir, hole 5 Kalda´rholt Laugaland, hole 3

6356.6 6359.1 6356.8 6400.6 6407.7 6400.2 6355.0

2116.6 2111.3 2057.5 2033.2 2019.5 2028.8 2025.0

886 953 500 85 321 38 1100

120 100 & 85 66 94 62 100

Vol. 165, 2008

Radon Changes and the South Iceland Earthquakes of June 2000

69

about twice a week from geothermal drill holes at seven sites (Fig. 2, Table 1) and sent to the Science Institute, University of Iceland, for analysis. These holes, ranging in depth between 38 m and 1100 m, were mostly the same as used in the earlier radon program. The June 2000 earthquakes originated within our sampling network. The nearest station, Laugaland, is only 2 km away from the source fault of the first earthquake (Fig. 2).

6. The Radon Changes The radon time series for the seven monitoring stations are shown in Figure 3. The raw data are plotted as individual points connected by thin lines. The numbers denote counts of disintegrations per hour in the sample. The thick, broken line shows a sevenpoint running average of the data. The vertical bars give the time of significant earthquakes in the area. The height of the bar reflects the ‘‘excess magnitude’’ of the event, ME, defined on the basis of the criteria of equation (1): ME ¼ M  ð2:4 log D  0:43Þ:

ð2Þ

Largest probabilities of radon anomalies are expected prior to events with positive excess magnitude. The scale in the plots is arbitrary and the relative height of the bars is only presumed to reflect the likelihood that the respective earthquake is associated with a radon excursion at that particular station. The radon variations form a distinct pattern that can be related to the earthquakes. A typical behavior is seen at the station Hlemmiskeid. The radon values prior to the earthquakes of June 2000 are relatively stable, varying by less than ± 50% around the mean. The largest deviations are positive, spike-like excursions that occur 59 and 115 days before the earthquakes. These are the highest values measured at this station during the two years of operation. Several of the lowest values are measured in a limited time interval 125–167 days before June 17. A pronounced coseismic step is observed at this station. The mean value drops by about 50% at the time of the earthquakes. About three months later the mean value returns to the pre-earthquake level. The main features of the Hlemmiskeið time series can be identified in the other time ¨ xnalækur series has a stable level in the series as well, except at Kalda´rholt. The O beginning, low values 124–138 days prior to the earthquakes, and a peak value 114 days before them. At Selfoss the same behavior is observed, but subdued. Low values occur 124–142 days before, and high values 114 and 144 days before the events. At Bakki anomalous values are observed, but the background level of radon is very low. Therefore the negative deviations are more difficult to identify. The values are apparently low in the time interval 117–134 days before the earthquakes, and 3 out of the 4 highest values are recorded 57, 50, and 43 days before June 17. At Flu´ðir the negative excursions are not prominent, although a large peak is seen 54 days before the earthquakes. The Laugaland series has well developed negative deviations 101–149

70

P. Einarsson et al.

Pure appl. geophys.,

prior to the events but positive spikes are not seen, except possibly a high value 58 days before the events. The Kalda´rholt time series is quite irregular and does not display the same pattern as the other stations. Kalda´rholt is located quite close to the source fault of the June 17 earthquake. ¨ xnalækur, Flu´ðir, Bakki, and In summary, the preseismic spikes are observed at O ¨ xnalækur, Selfoss, Laugaland, and possibly at Laugaland. The preseismic low is seen at O ¨ xnalækur, Laugaland, and possibly subdued at Bakki. The coseismic drop is observed at O at Selfoss. At Kalda´rholt and Flu´ðir the sampling was discontinued because of the coseismic pressure drop in the geothermal system. The radon values had returned to normal at all stations about three months after the earthquakes. The pressure in the geothermal systems also returned to normal at about this time.

7. Discussion The changes in radon concentration at our stations occur on a regional scale as shown by the similarity between the time series in Figure 3. The distance between the stations Selfoss and Laugaland is 27 km. This argues for a common cause of the changes. A meteorological cause is considered highly unlikely. The samples are taken from deep geothermal boreholes. No evidence for correlation with precipitation has been found. Seasonal effects are also considered unlikely. Radon time series are available for the same boreholes for the time period 1977–1993. Seasonal effects were only seen at one of the stations (Bakki) and only for a part of the observation period (Jo´NSSON, 1994). All the boreholes are producing geothermal holes, used for house heating. The production is at a maximum during the winter months. Therefore, if a seasonal effect is responsible the radon concentration is expected to be low in the winter and high in the summer. This is not consistent with the changes observed in 2000 (Fig. 3). The most robust result of this study is the demonstration of the coseismic drop in radon concentration and its postseismic return to previous values. Because of its temporal correlation with observed changes in groundwater pressure, we suggest that there is a causal relationship between these parameters. We note, however, that the radon concentration at all stations dropped during the earthquakes, regardless of whether they were located within areas of increasing or decreasing water pressure. It would seem that both increasing and decreasing groundwater pressure leads to a decrease in radon flow from the crustal rocks. This may not be as unreasonable as it seems at first sight. BJo¨RNSSON et al. (2001) and Jo´NSSON et al. (2003) point out that the pressure variations show a spatial pattern consistent with stress changes due to the faulting during the earthquakes. Pressure increases in areas where coseismic volumetric compression occurs in the crustal rocks. Pressure decreases where volumetric expansion occurs. The volumetric changes and water pressure are related through the porosity of the rock which in this case is to a large degree due to fractures. High water pressure is thus caused by the closure of cracks. Since the

Vol. 165, 2008

Radon Changes and the South Iceland Earthquakes of June 2000 J

1200

F M

A M

J

J

F M

A M

J

J

A S

O

N D

O

N D

Öxnalækur

900 J

A S

71

600

2000

Hlemmiskeið

300

1500

0

1000

5000

Kaldárholt

4000

500

3000 0

2000 1000

1200

Laugaland

0

900

6000

Flúðir

600

4000 300

2000

0

0

800

Selfoss 600

300

400

200

200

100

Bakki

0 0

50

100

150

200

N day

250

300

350

0 0

50

100

150 200 N day

250

300

350

Figure 3 Time series of radon activity at the seven sampling stations. The activity is given in counts per hour. 226 counts per hour correspond to 1 Bq/l of radon in the water. The data points are connected by a thin line. The dashed lines show a seven-point running average of the data. Vertical bars with stars on top give the time of significant earthquakes in the area. Their heights are proportional to the excess magnitude of the events, as defined by equation (2) and depend on the earthquake magnitude and epicentral distance to the radon station.

release of radon into the water takes place across fracture walls, the closure of fractures leads to reduced radon concentrations in the water. In the areas where coseismic dilatation takes place, increasing pore volume leads to a drop in pressure which inhibits the flow of water out of the rock. This will also lead to a reduced flux of radon out of the crustal rocks. Chemical components or physical parameters other than radon have not been monitored at our stations. However, a multi-parameter approach is highly desirable for a more meaningful interpretation of the causes of the changes as shown by the work of CLAESSON et al. (2004, 2007) in North Iceland.

72

P. Einarsson et al.

Pure appl. geophys.,

8. Conclusions Four different patterns can be identified in the radon time series within the South Iceland Seismic Zone in association with the earthquake sequence of June 2000: 1. Pre-seismic decrease of radon. Anomalously low values were measured in the period 101–167 days before the earthquakes. 2. Preseismic increase. Positive spikes appear in the time series 40–144 days prior to the earthquakes. 3. Coseismic step. The radon values decrease at the time of the first earthquake. This is most likely related to the coseismic change in groundwater pressure observed over the whole area. 4. Postseismic return to preseismic levels about three months after the earthquakes, probably also linked with the pressure equilibration in the geothermal systems. In view of the positive results of the project, we are developing and testing a new, automatic radon instrument, Auto-Radon, based on the same design that continuously monitors the radon concentration in the geothermal groundwater (THEODo´RSSON and GUDJo´NSSON, 2003; Jo´NSSON et al., 2007). The instruments are located at the drill hole stations, measuring radon four times a day.

Acknowledgements The radon programs in South Iceland have been supported by grants from several agencies, including the Icelandic Research Council, the SEISMIS Project, and the European Union under the projects PRENLAB and PREPARED. Numerous persons have participated in this radon project and measurements. We particular like to mention Gı´sli Jo´nsson and the attendants of the sampling sites in South Iceland, Guðlaugur ´ lafur O ´ lafsson, Valdimar Þorsteinsson, Vilhja´lmur Eirı´ksson, Sveinsson, Stefa´n O Guðru´n Magnu´sdo´ttir, Olgeir Engilbertsson, and Hannes Bjarnason.

REFERENCES ´ RNADo´TTIR, Th., HREINSDo´TTIR, S., GUðMUNDSSON, G., EINARSSON, P., HEINERT, M., and Vo¨LKSEN, C. (2001), A Crustal deformation Measured by GPS in the South Iceland Seismic Zone due to two Large Earthquakes in June 2000, Geophys. Res. Lett. 28, 4031–4033. ´ RNADo´TTIR, Th., Jo´NSSON, S., PEDERSEN, R., and GUDMUNDSSON, G. (2003), Coulomb stress changes in the South A Iceland Seismic Zone due to two large earthquakes in June 2000, Geophys. Res. Lett. 30, doi:10.1029/ 2002GL016495, no. 5. ´ RNADo´TTIR, Th., GEIRSSON, H., and EINARSSON, P. (2004), Coseismic stress changes and crustal deformation on A the Reykjanes Peninsula due to triggered earthquakes on June 17, 2000, J. Geophys. Res. 109, B09307, doi:10.1029/2004JB003130.

Vol. 165, 2008

Radon Changes and the South Iceland Earthquakes of June 2000

73

´ . G., SÆMUNDSSON, K., and EINARSSON, E. M. (2001), Pressure changes in Icelandic BJo¨RNSSON, G., FLo´VENZ, O geothermal reservoirs associated with two large earthquakes in June 2000. In Proceedings to Twenty-Sixth Workshop on Geothermal Reservoir Engineering, Stanford University. CLAESSON, L., SKELTON, A., GRAHAM, C., DIETL, C., Mo¨RTH, M., TORSSANDER, P., KOCKUM, I. (2004), Hydrogeochemical changes before and after a major earthquake, Geology 32, 641–644. CLAESSON, L., SKELTON, A., GRAHAM, C., and Mo¨RTH, M. (in review 2007), The timescale and mechanism of fault sealing and water-rock interaction after an earthquake, Geofluids. CLIFTON, A. and EINARSSON, P. (2005), Styles of surface rupture accompanying the June 17 and 21, 2000 earthquakes in the South Iceland Seismic Zone, Tectonophysics 396, 141–159. EINARSSON, P. (1985), Jarðskja´lftaspa´r (Earthquake prediction, in Icelandic with English summary), Na´ttu´rufræðingurinn 55, 9–28. EINARSSON, P. (1991), Earthquakes and present-day tectonism in Iceland, Tectonophysics 189, 261–279. EINARSSON, P., BJo¨RNSSON, S., FOULGER, G., STEFA´NSSON, R., and SKAFTADo´TTIR, Th. (1981), Seismicity pattern in the South Iceland Seismic Zone. In Earthquake Prediction - An International Review (eds. D. Simpson and P. Richards), Am. Geophys. Union, Maurice Ewing Series 4, 141–151. EINARSSON, P., CLIFTON, A., SIGMUNDSSON, F., and SIGBJo¨RNSSON, R. (2000), The South Iceland earthquakes of 2000: Tectonic environment and effects, Am. Geophys. Union, Fall Meeting, San Francisco, EOS 81, 890. GUDJONSSON, G. I. and THEODo´RSSON, P. (2000), A compact automatic low-level liquid scintillation system for Radon in water by pulse pair counting, Appl. Radiation and Isotopes 53, 377–380. HAUKSSON, E. (1981), Radon content of groundwater as an earthquake precursor: evaluation of worldwide data and physical basis, J. Geophys. Res. 86, 9397–9410. HAUKSSON, E. and GODDARD, J. (1981), Radon earthquake precursor studies in Iceland, J. Geophys. Res. 86, 7037–7054. Jo´NSSON, S. (1994), Radonmælingar a´ Suðurlandi (Radon measurements in South Iceland, in Icelandic), University of Iceland, Faculty of Science, BScThesis, 214 pp. Jo´NSSON, S. and EINARSSON, P. (1996), Radon anomalies and earthquakes in the South Iceland Seismic Zone 1977–1993. In Seismology in Europe (ed. Thorkelsson, B. et al.), European Seismol. Commission, Reykjavı´k, pp. 247–252. Jo´NSSON, S., SEGALL, P., PEDERSEN, R., and BJo¨RNSSON, G. (2003), Post-earthquake ground movements correlated to pore-pressure transients, Nature 424, 179–183. JONSSON, G., THEODORSSON, P., and SIGURDSSON, K. (2007), Auto-radon — a new automatic liquid scintillation system for monitoring Radon in water and air. In: Chalupnik S., Schonhofer, F., and Noakes J, eds. LSC 2005, Advances in Liquid Scintillation Spectrometry, in press. KING, C.-Y. (1985), Gas geochemistry applied to earthquake prediction: An overview, J. Geophys. Res. 91, 12,269–12,281. PAGLI, C., PEDERSEN, R., SIGMUNDSSON, F., and FEIGL, K. L. (2003), Triggered Seismicity on June 17, 2000 on the Reykjanes Peninsula, SW-Iceland Captured by Radar Interferometry, Geophys. Res. Lett. 30, 1273, 10.1029/ 2002GL-015310. ´ RNADo´TTIR, Th., SIGMUNDSSON, F., and FEIGL, K. L. (2003), Fault slip distribution of PEDERSEN, R., Jo´NSSON, S., A two June 2000 Mw 6.4 earthquakes in South Iceland estimated from joint inversion of InSAR and GPS measurements, Earth Planet. Sci. Lett. 213, 487–502. ROELOFFS, E. (1999), Radon and rock deformation, Nature 339, 104–105. SIGMUNDSSON, F., EINARSSON, P., BILHAM, R., and STURKELL (1995), Rift-transform kinematics in South Iceland: Deformation from global positioning system measurements, 1986 to 1992, J. Geophys. Res. 100, 6235–6248. STEFa´NSSON, R., BO¨ðVARSSON, R., SLUNGA, R., EINARSSON, P., JAKOBSDo´TTIR, S., BUNGUM, H., GREGERSEN, S., HAVSKOV, J., HJELME, J., and KORHONEN, H. (1993), Earthquake prediction research in the South Iceland Seismic Zone and the SIL Project, Bull. Seismol. Soc. Am. 83, 696–716. STEFa´NSSON, R. and HALLDo´RSSON, P. (1988), Strain release and strain build-up in the South Iceland Seismic Zone, Tectonophysics 152, 267–276. ´ RNADo´TTIR, Th., BJo¨RNSSON, G., GUðMUNDSSON, G. B., HALLDo´RSSON, P. (2000), The two large STEFa´NSSON, R., A earthquakes in the South Iceland Seismic Zone in June 2000. A basis for earthquake prediction research, Am. Geophys. Union, Fall Meeting, San Francisco, EOS 81, 890. THEODo´RSSON, P. (1996), Improved automatic Radon monitoring in groundwater, In Seismology in Europe (eds. Thorkelsson, B. et al.), European Seismological Commission, Reykjavı´k, pp. 253–257.

74

P. Einarsson et al.

Pure appl. geophys.,

THEODo´RSSON, P. and GUDJONSSON, G. I. (2003), A simple and sensitive liquid scintillation counting system for continuous monitoring of Radon in water, Advances in Liquid Scintillation Spectrometry, 249–252. TRIQUE, M., RICHON, P., PERRIER, F., and AVOUAC, J. P. (1999), Radon emanation and electrical potential variations associated with transient deformation near Reservoir Lakes, Nature 399, 137–140. WAKITA, H. (1996), Geochemical challenge to earthquake prediction, Proc. National Acad. of Sciences, USA, Vol. 93, No. 9 (Apr. 30, 1996), pp. 3781–3786. ZMAZEK, B., ITALIANO, F., ZIVCIC, M., VAUPOTIC, J., KOBAL, I., and MARTINELLI, G. (2002), Geochemical monitoring of thermal waters in Slovenia: Relationships to seismic activity, Appl. Radiat. Isot. 57, 919–930. (Received February 15, 2007, revised October 22, 2007, accepted October 24, 2007) Published Online First: February 1, 2008

To access this journal online: www.birkhauser.ch/pageoph

Pure appl. geophys. 165 (2008) 75–94 0033–4553/08/010075–20 DOI 10.1007/s00024-007-0291-7

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

CO2 Degassing over Seismic Areas: The Role of Mechanochemical Production at the Study Case of Central Apennines F. ITALIANO,1 G. MARTINELLI,2 and P. PLESCIA3

Abstract—Field observations coupled with experimental results show that CO2 can be produced by mechanical energy applied to carbonate rocks becoming an unexpected additional gas source besides that degassed from the mantle or produced by thermometamorphism. The evidence that a large amount of carbon dioxide associated with radiogenic-type helium (R/Ra as low as 0.01–0.08) is released through continental areas, denotes the absence of a contribution from the mantle or from mantle-derived fluids. Data collected during the seismic crisis which struck the Central Apennines in 1997–98 have shown an enhanced CO2 flux not associated with the presence of mantle or thermometamorphic-derived fluids. On the other hand, new experimental results highlight the possibility of producing CO2 by mechanical energy that acts on the calcite crystalline lattice. While the CO2 released over the geothermal areas (e.g., Larderello Geothermal Field) is obviously derived by mantlederived activities, this is not the case of the huge amount of CO2 released over the seismically active areas where the presence mantle-derived products is ruled out. We propose that mechanical energy, e.g., released during seismic events, microseismicity or creeping processes is a possible additional energy source able to produce CO2 and thus could explain the presence of CO2 degassing over tectonic areas where the influence of the mantle is low.

1. Introduction Apart from the water vapor that in high energy magmatic and geothermal systems is by far the most abundant gaseous specie, CO2 is the main residual component when temperatures drop and the water vapor condenses, and is the most common gas released from colder systems over tectonically active areas, where CO2-dominated mofettes are the typical gaseous manifestation. CO2 is also globally generated by the decomposition of organic matter or by microbial activity. The areas where CO2 degasses at the global scale have been outlined by IRWIN and BARNES (1980) who listed the recognized CO2-degassing prone areas as related to active tectonics. The active tectonic area of the Central Apennines is characterized by a widely distributed CO2 gas emission which is released both at seismically-active areas and via geothermal activity. A number of hypotheses 1

INGV Istituto Nazionale Geofisica e Vulcanologia, Sezione di Palermo, Italy. E-mail: [email protected] 2 ARPA Environmental Protection Agency of Emilia Romagna, Via Amendola 2, 42100 Reggio Emilia, Italy. 3 CNR Istituto Studio Materiali Nanostrutturati, Montelibretti RM, Italy.

76

F. Italiano et al.

Pure appl. geophys.,

have been proposed about its origin: PANICHI and TONGIORGI (1976) proposed an origin of CO2 from carbonate hydrolysis based upon both carbon isotope composition and geotemperature estimates. GIANELLI (1985) proposed an origin of the geothermal CO2 due to metamorphic process. MARINI and CHIODINI (1994) and CHIODINI et al. (1995) proposed that CO2 discharging in Central Italy originated by mixing of a mantle and a thermometamorphic component, while MINISSALE (2004) suggested that the CO2 is of mantle origin, considering the correlation between the anomalously high CO2 fluxes and the minimum depth of the Moho. Whereas the various postulated generation mechanisms are probably all presently active along the Apennine chain, to a lesser or greater extent, all of them require a thermal energy source which is available from the mantle or from mantle-derived products. The geothermal area of Larderello is located at the western sector of the Central Apennines and the released CO2 is obviously produced by the above-mentioned mechanisms. However, the Central Apennines are well known because of the active tectonic structures that generate moderate earthquakes (e.g., 1694, Me = 6.9; 1857, Me = 7.0; 1915, Me = 7.1; 1962, Me = 6.1; 1980, Me = 6.7, BOSCHI et al., 1995) and release huge amounts of CO2. Since the released CO2 is associated with a helium component marked by a typical radiogenic signature (ITALIANO et al., 2001; MINISSALE, 2004), it is difficult to believe that, although the two areas are close each other, the main source for the released CO2 over those seismic zones, is the same as for the Larderello geothermal area. The absence of geochemical evidences of mantle-derived products implies also that an additional thermal energy source from the mantle acting as the engine for thermometamorphic processes is not available. Following the experimental results about CO2 production, we provide in this paper an explanation for the presence of CO2 in continental seismic areas starting from the observation that: a) the 1997–1998 seismic crisis (Umbria Region; Central Apennines) induced an increased degassing of CO2 (HEINICKE et al., 2000; ITALIANO et al., 2001; CARACAUSI et al., 2005) that carried radiogenic-type helium; b) there is no evidence of He and CO2 decoupling so CO2 cannot be derived from a mantle source; and c) our experimental results of carbonate rock milling have shown the occurrence of CO2 production from mechanical processes.

2. The Field Work The field work started in October 1997 and is still ongoing. The three main dry vents of the seismic area of the Umbria region (Central Apennines) were repeatedly sampled and their gas output was evaluated during and after the seismic period. The collected samples were analyzed in the laboratory for their chemical composition and the isotopic composition of both carbon and helium. Details of the field sampling and analytical methods are given in Appendix 1.

Vol. 165, 2008

CO2 Degassing over Seismic Areas

77

Figure 1 displays the location of the CO2-dominated venting gases (Umbertide – UB; Montecastello di Vibio – VB; San Faustino – SF; Caprese Michelangelo – CM), besides other CO2-rich sources (bubbling and dissolved gases, CO2 exploited wells) of Central

Figure 1 Simplified map of the main seismogenic sources (top) crossing the Apennine chain (after VALENSISE and PANTOSTI, 2001). The location of the sampling sites is reported on the heat flow map of the investigated area of the Central Apennines (modified after CATALDI et al., 1995). The numbers show the sampling sites’ location as listed in Table 1.

78

F. Italiano et al.

Pure appl. geophys.,

Apennines. Table 1 lists the main geochemical features of the CO2-dominated gases from the Central Apennines. Hereafter, some information on the venting sites are summarized: • The Caprese Michelangelo (CM) degassing area is located to the north of the seismic area and characterized by the presence of several natural degassing vents and a well drilled to a depth of about 5000 m (HEINICKE et al., 2006). The main CO2 reservoir is located at a depth of 3700 m, at a static pressure of 70 MPa and a temperature of 120C. • The Montecastello di Vibio (VB) gas emission is located within Miocenic sediments and is characterized by a CO2 flow rate of about 1 m3/sec. Flow rate appears to be independent of seasonal variations. • The S. Faustino (SF) CO2-dominated gas emission occurs close to Massa Martana. Its flow rate is about 1 m3/sec and is characterized by constancy in time and apparently no sensitivity to seasonal variations. It is also a naturally-sparkling water exploited for bottling. • The Umbertide (UB) CO2-dominated gas emission derives from drilling for hydrocarbon exploration up to a depth of about 4800 m. It is currently the strongest CO2 emission located in a very dangerous depression close to the Umbertide town (Umbria Region, Central Apennines). The venting gas is characterized by a flow rate of about 3 m3/sec. The helium isotopic composition in the samples from UB and CM displays the lowest ratios in the range of 0.026–0.034 Ra (Ra = atmospheric ratio = 1.39 9 10-6) and 0.03–0.05 Ra at UB and CM vents and the well, respectively. Vents sampled at MV and SF have 3He/4He ratio ranging between 0.12–0.18 Ra and 0.24–0.28 Ra, respectively higher by an order of magnitude with respect to the 3He/4He ratio measured at the UB site. The helium concentration averages 50 ppmv at the CM gas vents and 440 ppmv at the CM well; 14.2 ± 1.5 (1r) ppmv at MV, 8.5 ± 0.7 ppmv at SF and 44 ± 8 ppmv at UB.

3. The Fluids Geochemistry and the CO2 Output during the Seismic Crisis The last seismic crisis that hit the Central Apennines had a strong impact on Italian society because it killed and injured people, and destroyed a large number of buildings including masterpieces of Italian cultural heritage. The area was struck by hundreds of earthquakes, starting on September 4, 1997, with a Ml = 4.4 foreshock, followed by two events with Ml = 5.6 and 5.8 on September 26 (MORELLI et al., 2000). Information on the earthquakes during the entire seismic sequence is available on web data bases (C.S.I., 1.1; 2003). The hypocenters of the entire seismic sequence were located at 5–10 km depth in the shallow crust (AMATO et al., 1998), except for one on March 26, 1998 (Mw 5.3), located at a depth of 51 km

Vol. 165, 2008

CO2 Degassing over Seismic Areas

79

Table 1 Main geochemical features for gas sources located along the Central Apennine chain. CO2 concentration, carbon and helium isotopic composition ðexpressed as R/RaÞ are reported. Samples labelled with * are from drilled wells. Geographical references in UTM coordinates. Data source: aÞ This work; bÞ MINISSALE ð2004Þ Site # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 40 41 42 43 44 45 46 47 48 49

Site name

Latitude

Longitude

% CO2

Asciano Acqua Bolle Caprese Michelangelo * Baccanella Piersanti Bruciano Borboi Vagliagli Acqua Borra Il Bagno * Caggiolo * Rimaggio * Grotte S. Stefano San Faustino Umbertine Montecastello di Vibio Stifone Triponzo Parrano Fersenone Fontecchio Rapolano Torrite di Siena Venturina Pienza Bagni S. Filippo S.Albino Zancona Roselle Selvena Torre Alfina Saturnia Pereta S.Martino Fiora Strada Ferento Muralto Bagnaccio Tuscania Monterozzi Terme Cotilia Solfatara Nepi Vaiano Borgo Pantano Caldara Palidoro Tivoli Lavinio Cava di selci

4851292 4842116 4837423 4836167 4834130 4782165 4818901 4814551 4801192 4817118 4798615 4817989 4709216 4709677 4800803 4746322 4708090 4743827 4750135 4749531 4676447 4799675 4786346 4774391 4773372 4757952 4751219 4750256 4746821 4740856 4737198 4728674 4728234 4727675 4709875 4707201 4705009 4703754 4702096 4693510 4678936 4678053 4670899 4663899 4650026 4647013 4595671 4504251

136601 183740 256719 153970 130131 653941 153440 205176 210303 716793 727027 717845 264826 300994 281548 283748 294277 331380 264029 276943 384897 224437 233189 141376 229191 229238 244494 217022 184431 224012 250303 214311 201312 222162 259556 284651 259182 244956 225945 335135 277171 261742 232195 260205 267870 311168 299133 520819

65.6 94.8 94.8 96.4 99.1 91.3 96 94.4 99.2 99.8 97.2 96.3 98.5 94 93 92 23.7 56.2 43.2 48.1 18 99.3 93.1 95.2 94.8 96.1 99.4 94.9 27.3 90.2 98.8 34.7 75.3 99.2 97.9 97.3 99.3 97.5 98.4 95.8 97.6 98.3 98 98 97.7 91.6 94.2 98.1

d13C CO2

R/Ra

Source

-10.3 -6.5 -4.1 -7.0 -6.7 -5.1 -9.3 -5.1 -5.9 -8.0 n.a. -7.2 0.1 -1.7 -3.5 -0.1- + 2.1 -2.5 -4.4 n.d. n.d. n.d. -7.5 -3.8 -13.4 -3.6 -3.3 -5.2 -4.6 -9.5 -3.3 1.0 -6.3 -6.2 0.1 -0.3 -0.9 -1.9 n.a. -0.1 -2.1 -1.3 -0.2 -2.1 -2.4 -1.8 -3.5 -0.5 n.a.

0.09 0.02 0.03 0.04 0.14 2.06 0.07 0.08 0.17 0.03 0.28 0.03 0.03 0.24 0.02 0.14 0.13 0.02 0.08 0.06 0.31 0.13 0.11 0.82 0.21 0.13 0.08 0.45 0.07 0.4 0.4 0.4 0.89 0.69 0.56 0.64 0.54 0.43 0.36 0.11 0.26 0.45 0.31 0.24 0.18 0.6 0.22 1.46

b b a b b b b b b a a a a a a a a a a a a b b b b b a b b b a b b b b b b b b a b b b b b b b a

80

F. Italiano et al.

Pure appl. geophys.,

Table 2 CO2 output from degassing areas, wells and vents of Central Apennines. Gas output data from wells refer to production data. The wells drain gas from shallow ð200–700 m deepÞ and pressurized reservoirs Site Geoth. systems Tuscany Grotte S. Stefano Caggiolo S. Faustino Vibio Stifone S. Albino Umbertide Il Bagno Grotte S. Stefano Caggiolo Rimaggio

Source type Geothermal gases Drilled well Drilled well Venting and bubbling gas Venting and bubbling gas Springs Drilled well Venting and bubbling gas Drilled well Drilled well Drilled well Drilled well

CO2 t/d

R/Ra

Ref.

39.8 64.8 4.8 7.0 10.0 1.0 38.4 20.0 3.6 61 6 6

2.80 0.60 0.30 0.23 0.13 0.10 0.08 0.03 0.03 0.03 0.03 0.03

CHIODINI et al. (1999) This work This work ITALIANO et al. (2001) ITALIANO et al. (2001) CHIODINI et al. (1999) This work ITALIANO et al. (2001) This work This work This work This work

(MORELLI et al., 2000). Focal mechanisms of the main shocks which occurred in 1997 (as well as 1979 and 1984), also confirmed by stress indicators (MONTONE et al., 1997), highlighted active extension processes in the NE-SW and E-W directions (FREPOLI and AMATO, 1997; AMATO et al., 1997). The entire seismogenic process caused crustal deformation with a maximum horizontal co-seismic displacement of 14 ± 1.8 cm and a maximum co-seismic subsidence of 24 ± 3 cm detected by means of SAR differential interferometry and GPS data (STRAMONDO et al., 1999), while a post-seismic long-term deformation process was detected by means of levelling data (BASILI and MEGHRAOUI, 2001). Starting from the beginning of the seismic crisis (September 26, 1997), samples of thermal waters (Bagni di Triponzo, Parrano and Stifone) and gas vents (Montecastello

Figure 2 Schematic representation of the experimental system. The jar with the ring mill is washed by the GC-grade inert gas (argon), then the milling process starts and the mechanochemically-produced gas is collected inside the preevacuated sampling bottle. It is possible to collect several samples simply by switching the three-way valve to replace the bottle.

Vol. 165, 2008

CO2 Degassing over Seismic Areas

81

di Vibio, Umbertide, San Faustino) were collected on a weekly basis during the period October 1997 to July 1998 and at longer intervals thereafter (twice a week, then monthly and then seasonal during 2001–2002). The data collected during a seismically quiet period (1999–2007) has allowed us to identify the background values for some geochemical parameters that could characterize the study area. Some previously published results have pointed out that: a) The thermal waters are CaSO4-enriched due to their circulation in the deep evaporitic basement and have changed their mixing proportions with cold waters typically equilibrated in shallow carbonate systems (FAVARA et al., 2001); during the seismic period, deep-originating waters contaminated shallow aquifers exploited for human purposes (ITALIANO et al., 2004); b) the gases dissolved in the thermal waters, CO2-dominated with the presence of CH4 and an excess of N2 in respect to the atmosphere, displayed various influxes of both deep-originating and atmospheric-derived components (ITALIANO et al., 2004); the 3He/4He ratios of the venting gases showed that although the region is located in a typical crustal environment, a slight contribution of a mantle-derived component could be detected (ITALIANO et al., 2001); the CO2 flux had shown modifications during the occurrence of the seismic crisis (HEINICKE et al., 2000; ITALIANO et al., 2001). A significant fluctuation of the 3He/4He ratio was observed at the MV site in concomitance with the strongest seismic events of 1997 and with the March 26, 1998 deep event. As an example, Figure 3 displays the results of the long-term monitoring at the gas vent of Umbertide. The data are useful for evaluating the temporal variations of CO2 and He contents in addition to 3He/4He ratios. The earthquakes recorded in the area are also reported (starting from the foreshock of the seismic crisis) and filtered as to be higher than Ml 2.5 and to have occurred within a radius of 40 km to the respect of the sampling site. The geochemical variations record changes in crustal permeability due to deformations associated with the subduction processes that characterize the Apennine chain. The most relevant geochemical changes were apparently linked to 5 events characterized by Ml C 5.6 and limited to the period September 1997 to April 1998. Conversely, the long-term geochemical trends indicate the persistence of long-lasting degassing processes which have no relationships with the seismic shocks. The CO2 released over the area affected by the seismic crisis was probably already stored in relatively shallow reservoirs whose existence is demonstrated by the large number of wells drilled for CO2 exploitation up to a depth of 5000 m (e.g., Caprese Michelangelo-CM and Umbertide – UB sites). Field measurements and observations carried out during the seismic crisis revealed that the broadly higher CO2 degassing occurred over the entire seismic area and induced a pH-drop in several springs with an increase in flow rate of about 50% at all the main gas vents (ITALIANO et al., 2001). All of the released CO2-dominated gases carried helium with a typical radiogenic signature. Only during the 1997–1998 seismic crisis were small additions of mantle-derived helium detected in the venting gases (ITALIANO et al., 2001) and

82

F. Italiano et al.

Pure appl. geophys.,

Figure 3 Temporal variations of CO2 (vol. %) and He (ppm vol.) concentration and 3He/4He ratio (R/Ra) values at the venting site of Umbertide. The earthquakes which occurred within a range of 40 km and with a magnitude above 2.5 are reported.

interpreted as due to normal faulting that temporarily enhanced the vertical permeability, allowing deep-originated fluids to mix with the shallow/crustal-derived main components.

Vol. 165, 2008

CO2 Degassing over Seismic Areas

83

4. Laboratory CO2 Production from Mechanical Energy The origin of CO2 in the absence of a degassing mantle or magmatic products is an already faced problem since early 1960s. Experimental results and theoretical approaches provided data and models about the possibility of generating CO2 as a result of hydrolysis processes (KISSIN and PAKHOMOV, 1969) or through decarbonation metamorphic processes (GIANELLI, 1985). All of the proposed models require a thermal energy source: KISSIN and PAKHOMOV (1969, 1975) deal with experimental water-rock interactions at boiling temperatures up to 250C; GIANELLI (1985) proposed theoretical results obtained in a PT range of 500–2000 bar (hydrostatic) and 350–520C with specific mineral assemblages. All of the previous results provide ways to produce CO2 in a geothermal environment, when enough thermal energy is available at a depth in the range of 2–8 km (lithostatic pressure, assuming a rock density of 2.6 g/cm3). The investigated area of Central Apennines is characterized by low heat flux with a geothermal gradient of about 30C/km. As such all of the proposed thermometamorphic temperatures can be available in the natural environment only at depths in the range of 11–17 km, namely at lithostatic pressures of 3–4.5 kbar, well above that in equilibrium for CO2 generation. The experimental activity aimed to produce CO2 by mechanochemical energy was developed in the laboratory using a commercial ring mill, modified to improve its gastightness and to allow the gas sample collection from the jar. Figure 2 schematically shows the experimental apparatus. A weighted amount of pure calcite powder was milled and the powder after the milling procedure was also carefully recovered and weighted. The produced gas was analyzed by gas chromatography. All of the experimental conditions are given in Table 3, while Table 4 lists the analytical results of the collected gas phase. Details of the experimental procedures are given in Appendix 2. The experimental results provide the information that mechanical energy is able to destroy the crystalline lattice allowing the calcite dissociation (CaCO3 = CaO + CO2). The analytical results reported on Table 4 show that the gas collected from the jar is always a mixture between air and an additional component mainly composed of CO2, where H2, CO and methane are also present in variable concentrations. We always adopted the same experimental procedure: to load the jar with the sample (5 g), wash the system by pure argon (washing time = 10 min.), milling (20 min.) and gas sampling. Finally we recovered the milled powder. The variable gas compositions we found Table 3 Results from the experiments on synthetic matter and natural rock ðCalcare Massiccio; Umbria region, ItalyÞ. The table displays the initial weight ðI.W.Þ, the final weight ðF.W.Þ after milling and the milling time Exp #

Composition

I.W. g

F.W. g

Grinding time minutes

A B C

Synthetic CaCO3 Natural CaCO3 Synthetic CaCO3 + Clay

5 5 5

2.5 n.d 3.9

20 10 20

84

F. Italiano et al.

Pure appl. geophys.,

Table 4 Analytical results of the collected gases. The sampling interval shows how many minutes after the beginning of the milling process the sampling bottle was opened ðfirst numberÞ and how long the sampling lasted ðsecond numberÞ. < = below the detection limit Exp. #

Sample #

H2 ppm

O2 vol.%

N2 vol.%

CO ppm

CH4 ppm

CO2 vol. %

Sampling interval

A A B B C C

1 2 5 6 9 10

< 177 72 142 < 428

19.5 19.3 0.4 5.9 19.9 4.3

78.3 78.0 2.0 23.6 77.2 17.7

< < 31 58 122 1.3

2.4 25 31 64 2.1 58

0.07 0.05 0.46 0.55 0.13 10.40

5–15 1–20 1–4 6–15 5–5 10–10

(Table 4) seem to be related to the different carbonate products (Table 3). The amount of CO2 produced by pure, synthetic calcite, is the lowest we measured, ranging from a little higher to double of the atmospheric content. The CO2 increased by one order of magnitude with the second set of experiments, when natural samples were milled (limestone from Central Italy known as Calcare Massiccio). The highest CO2 content, up to 10.4 vol%, was measured in gas samples collected when kaolinite was added to the synthetic calcite. To gain a quantitative estimation of the CO2 produced during the experiments, the gas analyses were recalculated and compared to the estimated volume of the milling system (3000 ml). Since the sampled gas was in an argon matrix, we restored the CO2 concentration assuming the total gas volume as composed of oxygen, nitrogen and carbon dioxide (Table 5). The restored CO2 concentration (Table 5) was then calculated by Ar removal and the amount of produced CO2, expressed as moles in Table 5, was estimated scaling the volume of the experimental system to the restored CO2 concentration. The recalculated CO2 concentrations show the presence of a significant amount of CO2 in most of the experiments, with concentrations up to 32%. The theoretical amount of CO2 produced by the mechanochemical process can be quantitatively estimated considering a total dissociation of the solid CaCO3; starting from Table 5 Recalculated CO2 data. Starting from the CO2 concentration data ðTable 4Þ. The table lists the recalculated data considering: The total gas concentration ð1Þ namely O2 + N2 + CO2 concentration; the Ar concentration ð100% - O2 + N2 + CO2Þ; the restored CO2 concentration is then calculated from the CO2 content ðTable 3Þ after Ar removal. The volume of CO2 is estimated considering the volume of the experimental apparatus ð3000 mlÞ and the CO2 concentration Sample #

Total gas (1)

Ar vol. %

Restored [CO2]

Volume moles CO2

1 2 5 6 9 10

97.84 97.37 2.93 30.03 97.28 32.41

2.16 2.63 97.07 69.97 2.72 67.59

0.07 0.05 15.70 1.83 0.14 32.09

9.29 1.04 2.60 2.84 2.35 4.80

9 9 9 9 9 9

10-5 10-4 10-2 10-3 10-4 10-2

Volume ml CO2 2.08 2.34 581.47 63.59 5.25 1074.40

Vol. 165, 2008

CO2 Degassing over Seismic Areas

85

5 grams (5 9 10-2 moles) of solid matter, the process might produce the same number of moles of both CO2 and CaO. The maximum volume of CO2 is thus fixed in 1120 ml. Table 5 shows the calculated amount of CO2 produced by the various experiments. The maximum amount (sample 10, Table 5) is close to the theoretical value. The powders from experiment A resulted to be 2.5 and 2 g (samples 1 and 2, respectively) after the weight, showing that a large amount of gas was missed during the experiments. Similar results came from the experiment C, sample 9: a final weight of 3.9 g (Table 4) should provide a CO2 amount of 115.4 ml, well above the measured 5.25 ml, showing that a better gas-tight system has to be developed. Although the experimental apparatus needs improvements and the experimental conditions can be considered distantly to approach the natural systems, the evaluation of the experimental results show that the production of CO2 is possible without any thermal energy as the primary energy source, and data from the induced pressure, had shown that they fit some conditions already determined on natural faults. In fact, the evaluation of pressure induced on samples by grinding was made by the analysis of residual stress on CaCO3 crystals, using the method of two peaks (MARTINELLI and PLESCIA, 2004). The mean stress was calculated to be: ðr1 þ r2 Þ ¼ ðE=rÞ  Dd=d; where (r1 + r2) are the main components of the stress, E is the Young module, r is the Poisson ratio, d is the interplanar distance at normal condition and Dd is the difference induced by the stress on the interplanar distance. The maximum value of residual stress was about 6.2 kbar and the mean value of 2.2 kbar. The numerical simulations of the stress in a fault system (ZOBAK et al., 1987), provided a shear stress of the order of 2 kbar, assuming a Mohr-Coulomb relation to calculate the normal stress in a fault system at lithostatic pressure (3 kbar at a depth of 10 km). A more detailed tensorial analysis provided a depth-averaged shear strength for faults in the brittle continental crust in the range of 1.5 kbar for thrust faulting, 0.6 kbar for strike-slip faulting, and 0.3 kbar for normal faulting (HICKMAN, 1991).

5. Discussion To infer the provenance of a natural gas mixture, a useful approach is to couple the isotopic features of carbon and helium. The various CO2 sources of a natural environment are marked by a different d13C ratio that is considered a powerful tool to investigate the origin of CO2 in a gas mixture (e.g., FAURE, 1977; JAVOY et al., 1986; ROLLINSON, 1993); the d13C data listed in Table 1, however, show that the isotopic composition of carbon deriving from a mantle-derived source (geothermal area of Larderello, R/Ra in the range of 0.89–2.06) and from other, radiogenic type sources (R/Ra in the range of 0.03–0.2), span over the same wide range of d13C values (between +1% and -7%). Such a wide range of isotopic composition highlights that gases coming from the same geochemical

86

F. Italiano et al.

Pure appl. geophys.,

environment do not identify any typical source. Since modifications of the d13C values are related to a wide spectrum of possibilities (e.g., isotopic fractionation, isotopic equilibrium, mixing of different sources etc.) the isotopic composition of carbon from CO2 does not provide useful information to constrain the origin of the gases released in the Central Apennines. In contrast, helium comes from three different and well distinguishable sources (mantle, crust and air), each marked by different isotopic ratios (e.g., OZIMA and PODOSEK, 1983). Moreover, due to its chemical inertness, the 3He/4He ratio is only sensitive to mass fractionation and mixing processes. It is well documented that helium isotopic ratios can be useful for evaluating a variety of geophysical and geological environments (MAMYRIN and Tolstikhin, 1984; OZIMA and PODOSEK, 1983). For example, in subduction zones, there is a clear geographical contrast in the 3He/4He ratio between lower values in the frontal arc and values higher in the volcanic arc (e.g., Southern Italy; SANO et al., 1989). In continental settings, it has long been established that the presence of mantle-He correlates well with tectonic and magmatic activity (POLYAK and TOLSTIKHIN, 1985), and that the mantle-He input occurs in areas with young volcanism (O’NIONS and OXBURGH, 1983; MARTY et al., 1989). As such, He-isotope ratios have been used in several tectonically active regions to identify mantle-derived products intruded at shallow levels in the crust (e.g., ITALIANO et al., 2000). The fluids released through the seismic and geothermal areas of the Central Apennines are a CO2-dominated mixture containing helium with isotopic ratios ranging from a typical radiogenic of 0.03 Ra to a mantle-derived signature of 2.8 Ra (Table 1). The presence of the mantle, or mantle-derived products, at shallow levels may account for both higher helium isotope values and CO2 generation either by direct degassing or by thermometamorphic processes. However, although this happens in the volcanic and geothermal peri-Tyrrhenian areas, there is no clear evidence of mantlederived products in other CO2-dominated gases released over the seismic areas of the Central Apennine chain. The existence of the small, Pleistocene, volcanic structures of San Venanzo (STOPPA and SFORNA, 1995) in the vicinity of Montecastello di Vibio and Colle Fabbri (STOPPA, 1988) in the vicinity of Massa Martana gives clues about possible tectonic movements that might have induced temporary, large increases in the vertical permeability, allowing the passage of tiny amounts of mantle-derived volatiles. This short-lived effect might allow some magmas to intrude to shallower crustal layers and to be erupted, giving rise to small volcanic edifices (at a reduced scale, this is the case of the 3He anomalies recorded during the 1997–1998 seismic crisis; ITALIANO et al., 2001). Considering the geochemical features of the fluids circulating at present in the area (e.g., Parrano and Fersenone springs with CO2-dominated dissolved gases and He isotope ratio of 0.08 and 0.06 Ra, respectively) we cannot consider those magmatic products as the origin of the presentlyreleased CO2, neither in terms of gas availability nor in terms of thermal energy. The evidence of the absence of a shallow thermal energy source fits also with the general map of the heat flow (HF) over the Italian peninsula (CATALDI et al., 1995) that highlights the

Vol. 165, 2008

CO2 Degassing over Seismic Areas

87

Figure 4 He vs. 4He for the three sampled vents of Umbertide (UB), San Faustino (SF) and Montecastello di Vibio (VB) are shown (modified after ITALIANO et al., 2001). The plotted data originate from the long-term geochemical monitoring and show that the isotopic ratio is almost constant with the time even on a ten-years scale. The CO2 flux is reported in tons per day units. 3

low HF values along the Apennine chain which is a ‘‘cold’’ area. Figure 4 shows the calculated 3He and 4He contents for various mixing proportions between the ‘‘local’’ mantle-derived source and the radiogenic source. The 3He/4He ratio of 4.48 9 10-6 is the highest value measured in the Central Apennines (Larderello geothermal field), and adopted as the ‘‘local’’ mantle-derived end-member for our calculations. The mixing lines are for pure radiogenic component and for the addition of 1, 2, 4 and 10% of mantlederived helium. Using those end-members, the UB site may have an addition of 1% of mantle-derived helium at most, while gases from SF display the highest mantle-derived helium content, reaching a value close to 10%. Significantly, the amount of degassed CO2 is lower where the helium isotopic ratio is higher. Moreover, scaling the helium concentration with the 3He/4He isotopic ratio for all the sampling sites, we could calculate that, although the measured helium isotopic ratios differ by an order of magnitude, 3He content is similar at all the sites (UB = 1.8 ± 0.4 9 10-6, MV = 2.8 ± 0.3 9 10-6 and SF = 3.1 ± 0.3 9 10-6 lmol/mol). It is also worth noting that the higher the helium concentration, the lower the 3He/4He isotopic ratio. The abovementioned observations suggest that similar amounts of 3He are supplied on a regional scale by the deep mantle-derived source, and 4He is added in different proportions at shallow depth together with CO2.

88

F. Italiano et al.

Pure appl. geophys.,

Figure 5 XRD of ‘‘as is’’ sample (natural limestone, black spectrum) and of the recovered powder (red spectrum) after 20’ grinding. The calcite peaks disappeared and the red line shows an amorphous powder marked by a higher background and the presence of only quartz as a crystalline product.

The occurrence of travertine rocks in the vicinity of many CO2 gas emissions (BARBIER and FANELLI, 1976; MINISSALE et al., 2002), highlights how the CO2 ascent is a massive and long-lasting phenomenon. The CO2 flux from the investigated vents (Table 2) can be considered as a massive degassing if compared, for example, to the CO2 degassing rate from the fumarolic field of the volcanic island of Vulcano (Aeolian Islands) estimated to be in the range of 50–200 tons per day as a function of the magmatic activity (ITALIANO et al., 1994). A possibility to support such a large degassing rate over a sedimentary area might be a shallow-depth magmatic intrusion as reported for the gas emission of Mefite d’Ansanto (Southern Apennines, ITALIANO et al., 2000) where helium isotope ratios reveal a clear mantle-derived signature (2.54 Ra). In those conditions, a CO2 production from any mantle-derived contribution lacks, and different or additional production mechanisms have to be hypothesized. Among them, the mechanochemical process is proposed as the energy feeder for CO2 production and degassing, for which experimental results fit with observations from the natural environment.

6. Conclusions The huge amount of CO2 released along the seismic area of the Central Apennines is considered to be originated by processes involving mantle degassing and/or thermometamorphism. Even though this is true for the geothermal areas (e.g., Larderello geothermal field), it does not look like the case for the seismic areas mainly located in the Umbria and Marche regions.

Vol. 165, 2008

CO2 Degassing over Seismic Areas

89

An unavailable thermal (mantle or mantle-derived) source along the Central Apennines, considered also a ‘‘cold’’ area on the basis of the low HF values besides the isotopic signature of helium that denotes a typical radiogenic (crustal) origin of the circulating fluids, rules out the origin of CO2 from mantle-derived or thermometamorphic processes. The mechanochemical production of CO2 appears by the process that is able to supply enough CO2 by a diverse source from the mantle and to be a comporary fit of the collected experimental and field data. We propose that over the investigated area, mantle-derived fluids are available at the main geothermal area of Larderello (Tuscany Region, western sector). The high heat flux denotes also the availability of thermal energy capable of producing CO2 from thermometamorphism that mixes to mantle-derived CO2 as already noted by other authors. The fluids from that area move through the tectonic discontinuities and can be spread throughout the Central Apennines. The mechanical energy available as the friction produced by fault movement and/or by deformation can be the other energy source able to produce CO2 and to release the radiogenic products trapped in rocks. Temporal modifications of the gas composition, including modifications of the helium isotopic composition, are interpreted as modifications of the mixing ratios between shallow and deep components.

Acknowledgments The authors are grateful to the Ph.D. student Sonia Pizzullo who supported the laboratory work proposed here and is still producing new data on the mechanochemical production of greenhouse gases. The authors wish to acknowledge D. Hilton, M. Kurtz, J. Heinicke, N. Perez, O. Vaselli, H. Woith, G. Yuce and M. Zimmer for their interesting and extensive discussions pertaining on this new topic during the Ninth International Conference on Gas Geochemistry, ICGG9, Taipei Conference. The paper benefited from the revision by David Hilton and Jens Heinicke, who greatly improved the earlier version of the manuscript. The work was partially supported by funds from the INGV-DPC-V5 project, RU9-Italiano.

Appendix 1 The gas samples were collected in pyrex flasks with valves at both ends. The bubbling gases were collected using a funnel to carry the bubbles toward the sampling bottle. The whole sampling apparatus, including the sampling bottle, was submerged and filled with water to prevent any air contamination. The venting gases were collected using a pipe or a funnel, depending on the site conditions. A long tygon tube connected the pipe to the sampling system made of a three-way valve joining together the sampling bottle and a syringe. The sample was collected after a cleaning procedure to purge the sampling bottle

90

F. Italiano et al.

Pure appl. geophys.,

of air. The gas was slowly aspired by the syringe and then pushed hardly to the bottle several times. The cleaning procedure was repeated until the aspired gas volume was 10 times larger than the volume of the sampling bottle. In the laboratory, the chemical analyses, the carbon isotopic ratios and the helium isotopic ratios were determined by gas chromatography and mass spectrometry, respectively, using gas aliquots extracted from the same sampling bottle. Chemical analyses were carried out using a Perkin Elmer 8500 gas chromatograph. An aliquot of 0.3 cm3 at a known pressure was admitted to the allmetal purification UHV line for helium isotope analysis. Helium was purified following three steps in isolated sections of the system: First, the reactive gases were absorbed into a charcoal trap held at 77 K; then two SAES getters worked simultaneously to absorb residual nitrogen at 250C and hydrogen at room temperature, and finally, after measuring the 4He/20Ne ratio and checking for residual 40Ar on an in-line quadrupole mass spectrometer, was to completely separate helium from neon by a charcoal trap held at 40 K. The isotopic analyses of the purified helium fraction were performed by a modified static vacuum mass spectrometer (GVI5400TFT) that allows for the simultaneous detection of 3He and 4He-ion beams, thereby causing the 3He/4He measurement error to drop to very low values. Typical uncertainties in the range of low-3He (radiogenic) samples are within ±5%. The gas output estimates were carried out following the simple method of water displacement. The gas was captured by a funnel or a specific device able to convey all the released gas to a container filled with water. The time the gas took to displace the water out of the container provided the output of the venting gas. For wide gas emissions, all the venting surface was covered by the adopted device. The uncertainty of the method (in the range of ±10%) was estimated by repeated measurements on the same vent.

Appendix 2 For the simulation of friction and mechanical compression over a large surface, we used a ring mill, which is the most efficient grinding system available at present for laboratory work (PLESCIA et al., 2003; MARTINELLI and PLESCIA 2004). The ring mill consists of a reinforced jar to contain the material and a grinding body made of a steel ring and solid steel roller. The ring mill operates basically through nonhydrostatic compression impact on the material particles and through friction rotation forced by the grinding body on the particles and between the particles and the jar walls. We account for some experiments carried out using a weighted amount of solid matter (5 grams) milled for 10–20 minutes (see Table 3). The system was connected to a bottle of high-pure argon to purge the jar before milling and to a pre-evacuated sampling bottle. An inlet and an outlet hole on the jar cover (Fig. 2) were allowed to flush before the beginning of the milling process to minimize possible reaction due to the presence of air. Then a pre-evacuated (internal pressure = 10-4 mbar) pyrex gas sampling bottle was connected at the outlet hole and

Vol. 165, 2008

CO2 Degassing over Seismic Areas

91

gas samples were collected at fixed time intervals during the 20’ milling (5, 10, 15 and 20 minutes after beginning the milling). Table 3 displays the experimental conditions for every single laboratory experiment. The collected gases were analyzed for their chemical composition. Chemical analyses were performed by a Perkin Elmer gas chromatograph equipped with a 4-m Carboxen 1000 column, using Ar as carrier and a double detector HWD and FID plus a methanizer. The detection limits are: 2 ppm vol. for He and H2, 1 ppm vol. for CH4 and CO2 and 500 ppm vol. for N2 and O2. The analytical errors are ± 5% for He and ± 3% for the other gaseous species. The analytical results are listed in Table 4. At the end of the milling process the pulverized sample was carefully collected by cleaning the jar, and all the grinding parts and the powder were weighted to evaluate the weight loss. XRD and TGA analyses were carried out before and after the milling process to evaluate the main composition of the solid matter. The solid matter milled during our experiments consisted of a pure synthetic calcium carbonate for analyses, a natural limestone and a calcium carbonate-kaolinitic clay mixture. The maximum theoretical amount of CO2 produced by the process can be calculated considering a total dissociation of the solid CaCO3, therefore starting from 5 grams (5 9 10-2 moles) of solid matter, the process might produce the same number of moles of both CO2 and CaO. The maximum volume of CO2 is thus fixed in 1120 ml, however the amount of milled powder (namely pure CaO) we extracted after the process is lower than the theoretical weight by 10%, probably because of manual errors in collecting all of the powder from the jar. Since the boundary conditions of the experiment were far to match a close system, losses of the produced gases have to be considered. Although the adopted system had many technical limitations, all the gas chromatographic analyses show the presence of CO2 concentration anomalies. In particular the slight CO2 anomaly found in the first experiment (synthetic CaCO3), becomes a significant anomaly (Table 4) with the second experiment (natural limestone). The last experiment (C; Tables 3 and 4) was carried out with some improvements to minimize the leaks from the jar, and the collected CO2 approached the theoretical value. The gas-chromatographic analyses of the gases generated during the milling of a mixture 2.5 + 2.5 grams of CaCO3 + Kaolynite (‘‘C’’ experiment, Table 3) display the largest CO2 concentration besides a significant amount of H2 and CH4. Although the considerations of the release of H2 and carbon from the steel jar are still valid, dehydroxylation processes of the clay fraction provide a large amount of H+ and OH- ions that can take part to the chemical reactions. The protonization event coincides with dehydroxylation of the clay during grinding. It has been clearly demonstrated by different authors (AGLIETTI et al., 1986; PLESCIA et al., 2003; KAMEDA et al., 2004) that phyllosilicates are particularly sensitive to mechanochemical action, which easily leads to the elimination of the OH groups and to a completely amorphous structure. The final gas composition is closely related to the mineralogical composition of the milled material, allowing the production of CH4, CO,

92

F. Italiano et al.

Pure appl. geophys.,

H2 besides the large amount of CO2. The produced gas phase exhibits a complex chemical composition that, however, has already been observed as H2 and hydrocarbon generation during grinding processes (e.g., AGLIETTI et al., 1986; KHOMENKO and LANGER, 1999; KAMEDA et al., 2004). Figure 3 shows a notable difference between the XRD spectra of the as-is and the milled samples, where the calcite peak is clearly present before the milling, while quartz is the only crystalline phase in the milled sample. A notable difference can be observed between gas samples collected a short time after the beginning of the milling process and gas samples collected at the end of the process. Good results were also obtained for the natural carbonate sample (Calcare Massiccio from the Central Apennines): we had the lowest air contamination and a reasonable amount of CO2 in the range of about 0.5% vol. Even though this paper focuses attention on the mechanochemical production of CO2 from pure calcium carbonates, it is valuable to note the significant H2 and CH4 generation during the milling process of CaCO3Kaolinite mixtures.

REFERENCES AGLIETTI, E.F., LOPEZ, P.J.M, and PEREIRA, E. (1986), Mechanochemical effects in kaolinite grinding, Int. J. Min. Proc. 16, 135–146. AMATO, A. and 18 others (1998), The 1997 Umbria-Marche, Italy, earthquake sequence: A first look at the main shocks and aftershocks, Geophys. Res. Lett. 25, 15, 2861–2864. BARBIER, E. and FANELLI, M. (1976), Main fractures of Italy from Earts satellite images and correlations with thermal springs, volcanoes and earthquakes. In (Aoki, H. and Iizuka, S., Eds.). BASILI, R. and MEGHRAOUI, M. (2001), Coseismic and postseismic displacement related with the 1997 earthquake sequence in Umbria-Marche (Central Italy), Geophys. Res. Lett. 28, 14, 2695–2698. BOSCHI, E., GUIDOBONI, E., FERRARI, G., MARIOTTI, D., VALENSISE, G., and GASPERINI, P. (2000), Catalogue of Strong Italian Earthquakes from 461 BC to 1197- Introductory text and CDrom. Annali di Geofisica 43, 4, 609–868. CARACAUSI, A., ITALIANO, F., MARTINELLI, G., PAONITA, A., and RIZZO, A. (2005), Long-term geochemical monitoring and extensive/compressive phenomena: Case study of the Umbria region (Central Apennines, Italy), Annals of Geophys. 48, 1, 43–53. CATALDI, R., MONGELLI, F., SQUARCI, P., TAFFI, L., ZITO, G., and CALORE, C. (1995), Geothermal ranking of Italian territory. Geothermics 24, 115–129. CATALOGO DELLA SISMICITa` ITALIANA (2003), CSI 1.1 1981–2002, http://legacy.ingv.it/CSI/. CERLING, T., QUADE, J., YANG, W., and BOREMAN, J. (1989), Soil and paleosols as ecologic and paleoecologic indicators, Nature 341, 138–139. CHIODINI, G., FRONDINI, F., and PONZIANI, F. (1995), Deep structures and carbon dioxide degassing in Central Italy, Geothermics 24, 81–94. CHIODINI, G., FRONDINI, F., KERRIK, D.M., ROGIE, J., PARELLO, F., PERUZZI, L., ZANZARI, A.R. (1999), Quantification of deep CO2 fluxes from central Italy. Examples of carbon balance for regional aquifers and of soil diffuse degassing. Chem. Geol. 159, 205–222. FAURE, G., Principles of Isotope Geology (J. Wiley, New York 1977). FAVARA, R., ITALIANO, F., and MARTINELLI, G. (2001), Earthquake-induced chemical changes in thermal waters of Umbria region during the 1997–1998 seismic swarm, Terra Nova, 13-3, 227–233. FREPOLI, A. and AMATO, A. (1997), Contemporaneous extension and compression in the Northern Apennines from earthquake fault-plane solutions, Geophys. J. Int. 129, 368–388.

Vol. 165, 2008

CO2 Degassing over Seismic Areas

93

GIANELLI,, G. (1985), On the origin of geothermal CO2 by metamorphic processes, Boll. Soc. Geol. Ital. 104, 575–584. HEINICKE, J., ITALIANO, F., LAPENNA, V., MARTINELLI, G., NUCCIO, P.M. (2000), Coseismic geochemical variations in some gas emissions of Umbria region, Central Italy, Phys. Chem. Earth 25, 289–293. HEINICKE, J., BRAUN, T., BURGASSI, P., ITALIANO, F., and MARTINELLI, G. (2006), Gas flux anomalies in seismogenic zones in the Upper Tiber Valley, Central, Geophys. J. Int. 167, 794–806. HICKMAN, S. (1991), Stress in the lithosphere and the strength of active faults, Rev. Geophys. 29, 759–775. IRWIN, W.P. and BARNES, I. (1980), Tectonic relations of carbon dioxide discharges and earthquakes, J. Geophys. Res. 85 (B6), 3115–3121. ITALIANO, F., NUCCIO, P.M., and PECORAINO, G. (1994), Fumarolic gas output at La Fossa di Vulcano Crater, Acta Vulcanolog. 6, 39–40. ITALIANO, F., MARTELLI, M., MARTINELLI, G., and NUCCIO, P.M. (2000), Geochemical evidences of melt intrusions along lithospheric faults of Irpinian Apennines (Southern Italy): Geodynamic and seismogenetic implications, J. Geophys. Res. 105, B6, 13569–13578. ITALIANO,, F., MARTINELLI, G., and NUCCIO, P. M. (2001), Anomalies of mantle-derived helium during the 1997– 1998 seismic swarm of Umbria-Marche, Italy, Geophys. Res. Lett. 28, 5, 839–842. ITALIANO, F., MARTINELLI, G., and RIZZO, A. (2004), Seismogenic-induced variations in the dissolved gases of the thermal waters of the Umbria region ðCentral Apennines, ItalyÞ during and after the 1997–1998 seismic swarm. G-Cubed 5, 11, doi:10.1029/2004GC000720. JAVOY, M., PINEAU, F., and DELORME, H. (1986), Carbon and nitrogen isotopes in the mantle, Chem. Geol. 57, 41–62. KAMEDA, J., SARUWATARI, K., and TANAKA, H. (2004), H2 generation during grinding of kaolinite, J. Colloid and Interface Sci. 275, 225–286. KANAMORI, H. (1994), Mechanics of earthquakes, Ann. Rev. Earth Planet. Sci. 22, 207–237. KHOMENKO, V.M. and Langer, K. (1999), Aliphatic hydrocarbons in structural channels of cordierite: A first evidence from polarized single-crystal IR absorption spectroscopy, Am. Min. 84, 1181–1185. KISSIN, I.G. and PAKHOMOV, S.I. (1967), The possibility of carbon dioxide generation at depth at moderately low temperature, Dokl. Akad. Nauk SSSR 174, 451–454. KISSIN, I.G. and PAKHOMOV, S.I. (1969), A contribution to the geochemistry of carbon dioxide in deep zones of underground hydrosphere. Geokhimiya, 4, 450–471 (in Russian). KISSIN, I.G. and PAKHOMOV, S.I. (1975), Some features of the geochemistry of thermal water in platform areas from experimental data. Proceedings of the Grenoble Symposium, August 1975, IAHS publication 119, 7–15. MAMYRIN, B. A. and TOLSTIKHIN, I. N. (1981), Helium Isotopes in Nature (Energoizdat, Moscow) [in Russian]. MARINI, L. and CHIODINI, G. (1994), The role of carbon dioxide in the carbonate-evaporite geothermal systems of Tuscany and Latium ðItalyÞ, Acta Vulcanol. 5, 95–104. MARTINELLI, G. and PLESCIA P. (2004), Mechanochemical dissociation of calcium carbonate: laboratory data and relation to natural emissions of CO2, Phys. Earth Planet. Int. 142, 3–4, 205–214. MARTY, B., JAMBON, A., and SANO, Y. (1989), Helium isotopes and CO2 in volcanic gases of Japan. Chem. Geol, 76, 25–40. MINISSALE, A. (1991), Thermal springs in Italy: Their relation to recent tectonics, Appl. Geochem. 6, 201–212. MINISSALE, A., (2004), Origin, transport and discharge of CO2 in Central Italy. Earth Sci. Rev. 66, 89–141. MINISSALE, A., EVANS, W., MAGRO, G., and VASELLI, O. (1997a), Multiple source components in gas manifestations from northcentral Italy, Chem. Geol. 142, 175–192. MINISSALE, A., KERRICK, D.M., MAGRO, G., MURRELL, M.T., PALADINI, M., RIHS, S., STURCHIO, N.C., TASSI, F., and VASELLI, O. (2002), Geochemistry of Quaternary travertines in the region north of Rome ðItalyÞ : structural, hydrologic and paleoclimatic implications, Earth Planet. Sci. Lett. 203 709–728. MONTONE, P., AMATO, A., FREPOLI, A., MARIUCCI, M.T., and CESARO, M. (1997), Crustal stress regime in Italy, Annali di Geofisica XL, 3, 741–757. MORELLI,, A., EKSTROM, G., and OLIVERI, M. (2000), Source properties of the 1997–1998 Central Italy earthquake sequence from inversion of long-period and broad-band seismograms, J. Seismol. 4, 365–375. O’NIONS, R.K. and OXBURGH, E.R. (1983), Heat and helium in the Earth, Nature 306, 429–431. OZIMA, M. and PODOSEK, F.A. Noble Gas Geochemistry. (Cambridge University Press, Cambridge, 1983) 286 pp. PANICHI, C., and TONGIORGI, E. (1976), Carbon isotopic composition of CO2 from springs, fumaroles, mofettes and travertines of central and southern Italy: A preliminary prospection method of geothermal areas, Proc.

94

F. Italiano et al.

Pure appl. geophys.,

2nd U.N. Symp. on the Develop. and Use of Geotherm. Energy, San Francisco, USA., 20–29 May 1975, pp. 815–825. PLESCIA, P., GIZZI, D., BENEDETTI, S., CAMILUCCI, L., FANIZZA, C., and PAGLIETTI, F. (2003), Mechanochemical treatment to recycling asbestos containing waste, Waste Managem. 23, 209–218. POLYAK, B.G., TOLSTIKHIN, I.N. (1985), Isotopic composition of the Earth’s helium and the problem of tectogenesis. Chem. Geol. 52, 9–33. ROLLINSON, H., Using Geochemical Data (Longman Group, London 1993). SANO, Y., WAKITA, H., ITALIANO, F., and NUCCIO, P.M. (1989), Helium isotopes and tectonics in southern Italy, Geophys. Res. Lett. 16, 6, 511–514. STOPPA, F. (1988), L’eurimite di Colle Fabbri ðSpoletoÞ: un litotipo ad affinita` carbonatitica in Italia, Boll. Soc. Geol. It. 107, 239–248. STOPPA, F. and SFORNA, S. (1995), Geological map of the San Venanzo volcano ðCentral ItalyÞ: Explanatory notes, Acta Vulcanologica 7, 85–91. STRAMONDO et al. (1999). ZOBACK, M.L., ZOBACK, V., MOUNT, J., EATON, J., and HEALY et al. (1987), New evidence of the state of stress of the San Andreas fault zone, Sci. 238, 1105–1111. (Received October 16, 2007, revised October 28, 2007, accepted October 30, 2007)

To access this journal online: www.birkhauser.ch/pageoph

Pure appl. geophys. 165 (2008) 95–114 0033–4553/08/010095–20 DOI 10.1007/s00024-007-0281-9

 Birkha¨user Verlag, Basel, 2008

Pure and Applied Geophysics

Changes in the Diffuse CO2 Emission and Relation to Seismic Activity in and around El Hierro, Canary Islands ELEAZAR PADRO´N, GLADYS MELIA´N, RAYCO MARRERO, DA´CIL NOLASCO, JOSE´ BARRANCOS, GERMA´N PADILLA, PEDRO A. HERNA´NDEZ, and NEMESIO M. PE´REZ

Abstract—Significant changes in the diffuse emission of carbon dioxide were recorded in a geochemical station located at El Hierro, Canary Islands, before the occurrence of several seismic events during 2004. Two precursory CO2 efflux increases started thirteen and nine days before two seismic events of magnitude 2.3 and 1.7, which took place near El Hierro Island, Canary Islands, on March 23 and April 15, reaching a maximun value of 51.1 and 46.2 g m-2 d-1, respectively, five and eight days before the two seismic events. Other similar increases started thirteen and five days before the occurrence of two seismic events of magnitude 1.3 and 1.5 which took place on October 15 and 21 respectively, reaching the maximum values four and one day before the earthquakes. These changes were not related to variations in atmospheric or soil parameters. The Material Failure Forecast Method (FFM), which analyzes the rate of precursory phenomena, was successfully applied to forecast the first seismic event that took place in El Hierro Island in 2004. Key words: El Hierro Island, precursors, Material Failure Forecast Method, diffuse degassing, carbon dioxide.

1. Introduction El Hierro Island (278 km2) is one of the youngest and the southwesternmost of the Canary Islands and rises 4000 m above the sea floor (Fig. 1). The main characteristics of El Hierro consist of a truncated trihedron shape and three convergent ridges of volcanic cones. The older subaerial rocks of El Hierro have been dated at 1.12 Ma (GUILLOU et al., 1996) and there is only one questionable report of a single volcanic eruption in El Hierro Island in the last 500 years, Lomo Negro volcano, in 1793 (HERNA´NDEZ PACHECO, 1982). The volcanic evolution of El Hierro can be divided into three successive volcanic edifices: Tin˜or volcano, El Golfo volcano and Rift Volcanism (GUILLOU et al., 1996; CARRACEDO et al., 1997). The island has been covered in the last 37 ka by lavas erupted by the last stage of the volcanic evolution and deep embayment has been produced by giant landslides between the three rift zones, being the most recent El Golfo failure on the

Enviromental Research Division, Instituto Tecnolo´gico y de Energı´as Renovables (ITER), 38611 Granadilla, S/C de Tenerife, Spain. E-mail: [email protected]

96

E. Padro´n et al.

Pure appl. geophys.,

northwest flank of El Hierro, which occurred approximately 15 ka ago (MASSON, 1996; GEE et al., 2001). Since fumarolic activity is absent at the surface environment of El Hierro, the study of the evolution of diffuse CO2 emissions becomes an ideal geochemical tool for monitoring its volcanic activity (CHIODINI et al., 1998; HERNA´NDEZ et al., 2001a, b, c, 2003, 2006; SHIMOIKE et al., 2002; FRONDINI et al., 2004; NOTSU et al., 2005, 2006; GRANIERI et al., 2006). CO2 is, after water vapor, the major gas species in basaltic magmas (BARNES et al., 1988), and it is a good geochemical tracer of subsurface magma degassing, since its low solubility in silicate melts at low and moderate pressure (GERLACH and GRAEBER, 1985). Natural emissions of CO2 have different sources: mantle, carbonate metamorphism, descomposition of organic matter and biological activity (IRWIN and BARNES, 1980) and active faults favor gas leaks because they are preferential paths for crustal and subcrustal gases (IRWIN and BARNES, 1980; SUGISAKI et al., 1983; KLUSMAN, 1993; GIAMMANCO et al., 1998; KING, 1996; KING et al., 2006). Areas with high CO2 discharges can indicate high pore pressure at depth and might be a tool to identify potential seismic regions (ROJSTACZER et al., 1995; CASTAGNOLO et al., 2001; SPICAK and HORALEK, 2001). Relatively high CO2 fluxes correlate with areas that show deep fractures or faults with emissions of CO2 from magmatic reservoirs or decarbonation processes (TOUTAIN and BAUBRON, 1999) and increases of diffuse CO2 emissions related to seismic events and volcanic activity have been reported (HERNA´NDEZ et al., 2001b; ROGIE et al., 2001; SALAZAR et al., 2002; CARAPEZZA et al., 2004; PE´REZ et al., 2005). In order to improve the volcanic surveillance program of El Hierro Island and to provide a multidisciplinary approach, a continuous geochemical station to measure CO2 efflux was installed on September 2003 in Llanos de Guille´n, the interception center of the three volcanic-rift zones of the island, with the aim of detecting changes in the diffuse emission of CO2 related to the seismic or volcanic activity. Monitoring of CO2 efflux has demonstrated to be a useful tool to forecast precursory signals of volcanic eruptions and seismic events. HERNA´NDEZ et al., (2001b) reported an increase from 120 to 240 t/d on the CO2 efflux six months before the volcanic eruption of the Usu volcano, Japan, which occurred in 2000. CARAPEZZA et al. (2004) observed a significant increase of nearly double the maximum CO2 efflux values measured previously by an automatic geochemical station one week before the 2002 Stromboli eruption, Italy. SALAZAR et al. (2002) observed anomalous changes in the diffuse emission of carbon dioxide before some of the aftershocks of the 13 February 2001 El Salvador earthquake. PE´REZ and HERNA´NDEZ, 2005 and PE´REZ et al., 2006 have reported significant increases in a CO2 efflux time series prior to seismic events, as the increase observed from approximately 16 g m-2 d-1 to 270 g m-2 d-1, in the carbon dioxide efflux values measured in an automatic geochemical station nine days before the January 2002 short temp unrest occurred at San Miguel volcano, El Salvdor. In the last 15 years, the Instituto Geogra´fico Nacional (IGN) has reported the occurrence of several seismic events in and around El Hierro Island. Figure 2 shows the number of earthquakes registered in and around El Hierro Island since 1993. An

Vol. 165, 2008

Changes in the Diffuse CO2 Emission

97

anomalous increase in the seismic activity occurred in 2004. Unfortunately, no mechanism information is available for these earthquakes due to the characteristics of the seimic network of IGN in the Canary Islands.

2. Procedures and Methods The automatic geochemical station (EHI01) to measure the CO2 efflux was installed at Llanos de Guille´n, in the center of El Hierro Island (Latitude: N 27 420 58.2@; Longitude: W 18 010 8.8@) on September 25, 2003. Previous CO2 efflux surveys covering the entire island indicated that the selected location for the automatic station shows one of the highest CO2 efflux values measured in El Hierro Island (MART´ıNEZZUBIETA, 2001; PADRo´N et al., 2006). Moreover, the place is located at the interception center of the three volcanic rifts of the El Hierro Island. The station measures on an hourly basis the CO2 and H2S efflux, the CO2 and H2S air concentrations, the soil water content and temperature and the atmospheric parameters: wind speed and direction, air temperature and humidity and barometric pressure. The meteorological parameters together with the air CO2 concentration are measured 1 m above the ground and the soil water content and soil temperature are measured 40-cm deep, and recorded contemporaneously with CO2 efflux. On October 5, 2004, a rain gauge was also installed in the geochemical station. Both CO2 and H2S diffuse fluxes are estimated according to the accumulation chamber method (PARKINSON, 1981) by means of a nondispersive specrtophotometer (LICOR Li-820) with a 2000 ppm span cell and a ¨ GER Polytron II, respectively. The geochemical station is powered by a solar cell DRA panel and a battery. Each CO2 and H2S efflux measurement starts when the open side of the chamber is placed onto a fixed collar in the soil surface. A pump allows the air contained in the chamber to circulate through the NDIR spectrophotometer and then back into the chamber. To verify the performances and the reliability of this method, several calibration tests were made in the laboratory and the accuracy was estimated to be ±10%. Each hour the station also measures the soil temperature and water content and the meteorological parameters. All the data are stored on flash memory and radio-telemetered to ITER.

3. Results and Discussion During 2004 a total of thirteen seismic events were registered by the seismic network of IGN in and around El Hierro Island. The locations of these seismic events are shown in Figure 3. A time series of the total 6,385 measured data of CO2 efflux, wind speed, soil water content and temperature, air humidity and temperature and barometric pressure during 2004 is shown in Figure 4. A 48-hour moving average is also plotted for CO2 efflux, wind speed, air humidity and temperature and barometric pressure time series.

E. Padro´n et al.

98

Pure appl. geophys.,

Due to instrumental and telemetry problems, the time series has a 27.1% of missing data, with the main lag of data occurring between June 8 to July 29. Table 1 summarizes the results of the total recorded data. The CO2 efflux ranged between nondetectable values to 53.1 g m-2 d-1, with an average value of 12.5 g m-2 d-1. The detection limit of the automatic station has been estimated to be 0.5 g m-2 d-1. During the period of study, the H2S efflux values were always below the detection limit of the instrument (