Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining: ISRM 2020 - Volume 2 [1st ed.] 9783030602680, 9783030602697

Citation preview

Lecture Notes in Civil Engineering

Dieu Tien Bui Hai Thanh Tran Xuan-Nam Bui   Editors

Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining ISRM 2020 - Volume 2

Lecture Notes in Civil Engineering Volume 108

Series Editors Marco di Prisco, Politecnico di Milano, Milano, Italy Sheng-Hong Chen, School of Water Resources and Hydropower Engineering, Wuhan University, Wuhan, China Ioannis Vayas, Institute of Steel Structures, National Technical University of Athens, Athens, Greece Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup, WA, Australia Anuj Sharma, Iowa State University, Ames, IA, USA Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science Bangalore, Bengaluru, Karnataka, India Chien Ming Wang, School of Civil Engineering, The University of Queensland, Brisbane, QLD, Australia

Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in Civil Engineering - quickly, informally and in top quality. Though original research reported in proceedings and post-proceedings represents the core of LNCE, edited volumes of exceptionally high quality and interest may also be considered for publication. Volumes published in LNCE embrace all aspects and subfields of, as well as new challenges in, Civil Engineering. Topics in the series include: • • • • • • • • • • • • • • •

Construction and Structural Mechanics Building Materials Concrete, Steel and Timber Structures Geotechnical Engineering Earthquake Engineering Coastal Engineering Ocean and Offshore Engineering; Ships and Floating Structures Hydraulics, Hydrology and Water Resources Engineering Environmental Engineering and Sustainability Structural Health and Monitoring Surveying and Geographical Information Systems Indoor Environments Transportation and Traffic Risk Analysis Safety and Security

To submit a proposal or request further information, please contact the appropriate Springer Editor: – Mr. Pierpaolo Riva at [email protected] (Europe and Americas); – Ms. Swati Meherishi at [email protected] (Asia - except China, and Australia, New Zealand); – Dr. Mengchu Huang at [email protected] (China). All books in the series now indexed by Scopus and EI Compendex database!

More information about this series at http://www.springer.com/series/15087

Dieu Tien Bui Hai Thanh Tran Xuan-Nam Bui •

•

Editors

Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining ISRM 2020 - Volume 2

123

Editors Dieu Tien Bui GIS group, Department of Business and IT University of South-Eastern Norway Bø I Telemark, Norway

Hai Thanh Tran Department of Geology Hanoi University of Mining and Geology Hanoi, Vietnam

Xuan-Nam Bui Hanoi University of Mining and Geology Hanoi, Vietnam

ISSN 2366-2557 ISSN 2366-2565 (electronic) Lecture Notes in Civil Engineering ISBN 978-3-030-60268-0 ISBN 978-3-030-60269-7 (eBook) https://doi.org/10.1007/978-3-030-60269-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

We would like to welcome you to the International Conference on Innovations for Sustainable and Responsible Mining - ISRM 2020, which will be held during October 15–17, 2020, at Hanoi University of Mining and Geology (HUMG), Hanoi, Vietnam. ISRM 2020 is organized by HUMG to celebrate the 55th anniversary of the Department of Surface Mining, Faculty of Mining, HUMG. The conference is the effective cooperation of HUMG and both TU Bergakademie Freiberg (Germany), and Hanoi University of Public Health (Vietnam). Especially, the event is financially supported by Vietnam National Coal-Mineral Industries Holding Corporation Limited (VINACOMIN), Dong Bac Corporation (NECO), and other organizations. The main aim of the ISRM 2020 is to provide a platform for researchers, academicians, and engineers in the field of mining, earth resources, civil engineering, and geospatial technologies to present their recent research results. Besides, the conference provides a setting for them to exchange new ideas, innovative thinking, and application experiences face-to-face, to establish research or business relations, and to find partners for future collaboration. The conference program was organized into four sessions covering topics of sustainable development and responsible mining, earth sciences and geospatial technologies, occupational safety and health, and Industry 4.0 in mining. The ISRM 2020 has received 344 submissions, and among them, 68 high-quality manuscripts were recommended to submit for the section earth sciences and geospatial technologies of this Springer proceedings book for a double-blind peer-reviewing. Herein, each manuscript has been reviewed for its merit and novelty by at least two reviewers by matching the content areas. As a result, a total of 21 papers have been finally selected for this book. We believe that this proceedings book provides a broad overview of recent advances in earth sciences and geospatial technologies for readers. Finally, we would like to express our sincere thanks to the university council of HUMG, the rector and vice-rectors of HUMG, and the International Office of HUMG for their help in administrative works and other supports. Special thanks to Dr. Nguyen Quoc Long, the secretary of the ISRM2020, and Pierpaolo Riva at v

vi

Preface

Springer for help and always responding promptly. We would like to thank all the reviewers for their timely and rigorous reviews of the papers and to thank all the authors for their submissions. October 2020

Dieu Tien Bui Hai Thanh Tran Xuan-Nam Bui

Organization

List of Reviewers Anh Hong Le Ataollah Shirzadi Bach Thao Nguyen Chu Dinh Toi Chuyen Tran Trung Dang An Tran Dieu Tien Bui Duc Minh Do Duc-Tan Tran Duc Van Bui Duong Van Nhiem Duong Thanh Trung Gian Quoc Anh Hanh Hong Tran José Lázaro Amaro-Mellado Khaled Shaaban Kim Anh Nguyen Lam Van Nguyen Loc Huu Ho Loi Huy Doan Luyen Khac Bui

Hanoi University of Mining and Geology, Vietnam University of Kurdistan, Iran Hanoi University of Mining and Geology, Vietnam Hanoi National University of Education, Vietnam Hanoi University of Mining and Geology, Vietnam University of Tsukuba, Japan University of South-Eastern Norway, Norway Vietnam National University, Vietnam Phenikaa University, Hanoi, Vietnam Hanoi University of Mining and Geology, Vietnam Vietnam National University of Agriculture, Vietnam Hanoi University of Mining and Geology, Vietnam Nam Dinh University of Technology Education, Vietnam Hanoi University of Mining and Geology, Vietnam University of Seville, Spain Qatar University, Qatar National Central University, Taiwan Norwegian University of Science and Technology, Norway Nanyang Technological University, Singapore Institute of Transport Science and Technology, Vietnam Curtin University, Australia

vii

viii

Minh Hieu Nguyen Nguyen Quoc Long Nguyen Van Sang Nguyen Hong Quang Omid Rahmati Paraskevas Tsangaratos Quang Tu Pham Romulus-Dumitru Costache Seyda Tasar Thu Trang Le Tien Dat Pham Tien Duc Pham Tran The Viet Viet-Ha Nhu Vu Minh Ngan

Organization

University of Transport and Communications, Vietnam Hanoi University of Mining and Geology, Vietnam Hanoi University of Mining and Geology, Vietnam Vietnam Academy of Science and Technology, Vietnam Agricultural Research Education and Extension Organization, Iran National Technical University of Athens, Greece Thuyloi University, Vietnam National Institute of Hydrology and Water Management, Romania Firat University, Turkey Université Clermont Auvergne, France Florida International University, USA Vietnam National University, Vietnam Thuyloi University, Vietnam Hanoi University of Mining and Geology, Vietnam Hanoi University of Mining and Geology, Vietnam

Contents

Neotectonic Activities and Its Significance to River-Course Evolution: Implication for the Cai River Catchment, Ninh Thuan Province, South-Central Vietnam . . . . . . . . . . . . . . . . . . . . Hai Thanh Tran, Chi Kim Thi Ngo, Hau Vinh Bui, Binh Van Nguyen, Thao Thanh Nguyen, Hien Thi Hoang, Nam Xuan Nguyen, and Tu Do Ngo Hoang Mining-Induced Land Subsidence Detection by Persistent Scatterer InSAR and Sentinel-1: Application to Phugiao Quarries, Vietnam . . . . Bui Xuan Nam, Tran Van Anh, Luyen K. Bui, Nguyen Quoc Long, Thi Le Thu Ha, and Ropesh Goyal Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam) for Geohazard Assessment and Sustainable Development . . . Viet-Ha Nhu Establishing a Tungsten Deposit Group and a Pattern Grid Exploration in the Nui Phao Area, Northeastern Vietnam . . . . . . . . . . . Khuong The Hung, Luong Quang Khang, Pham Nhu Sang, and Hoang Van Vuong Identification of Sensitive Factors for Placement of Flood Monitoring Sensors in Wastewater/Stormwater Network Using GIS-Based Fuzzy Analytical Hierarchy Process: A Case of Study in Ålesund, Norway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lam Van Nguyen, Dieu Tien Bui, and Razak Seidu Evaluating the Service Quality of the First Bus Rapid Transit Corridor in Hanoi City and Policy Implications . . . . . . . . . . . . . . . . . . . Minh Hieu Nguyen

1

18

39

58

79

98

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam Using Multi-criteria Analysis . . . . . . . . . . . . . . . . . . . . . . . . 124 Tuan Anh Pham and My Van Nguyen ix

x

Contents

Modernization of Height System in Vietnam Using GNSS and Geoid Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Ngoc Ha Hoang Seismic Hazard Assessment for South-Central Region, Vietnam . . . . . . 167 Trong Cao Dinh, Bach Mai Xuan, Hung Pham Nam, Tuan Thai Anh, Vuong Trong Kha, and Trieu Cao Dinh Secondary Processes Associated with Landslides in Vietnam . . . . . . . . . 192 Pham Van Tien, Le Hong Luong, Tran Thanh Nhan, Do Minh Duc, Dinh Thi Quynh, Nguyen Chau Lan, Nguyen Quoc Phi, Do Canh Hao, Nguyen Huu Ha, Dang Thi Thuy, and Vu Ba Thao Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional: A Case Study in Sapa, Vietnam . . . . . . . . . . . . . 210 The Viet Tran, Viet Hung Hoang, Huy Dung Pham, and Go Sato Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment of the Cliff at Chau Thoi Quarry . . . . . . . . . . . . . . 230 Nguyen Quang Tuan A Review of Soil Improvement Methods for Tunneling Projects in Urban Areas and Their Application at the Hochiminh Metroline No. 1, Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Minh Ngan Vu, Phuc Lam Dao, Vu Nam Chien Nguyen, and Duc Thinh Ta Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture: A Case of X-well in Vietnam . . . . . . . . . . . . . . . 270 Hai Linh Luong, Hung Van Nguyen, and Nhu Y. Ha High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution at Thi Vai River, Ho Chi Minh City, Vietnam . . . . . . . . . . 290 Cuong Van Anh Le, Man Ba Duong, and Thong Duy Kieu Initial Results of Using Biochar Derived from Spent Coffee Grounds to Remove Pollutants from Livestock Wastewater in Vietnam . . . . . . . . 305 Tran Thi Thu Huong, Nguyen Van Hoang, Vu Ngoc Toan, Nguyen Xuan Tong, Tran Anh Quan, and Vu Kim Thu Assessment the Impact of Climate Change and Sea Level Rise on the Unconfined Aquifer at the Red-River Delta of Vietnam: A Case Study at Thai Binh Province . . . . . . . . . . . . . . . . . . . . . . . . . . . 326 Tran Thi Thanh Thuy, Pham Khanh Huy, Dao Duc Bang, and Pham Hoang Anh Assessment of the Shoreline Evolution at the Eastern Giens Tombolo of France . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 Minh Tuan Vu, Yves Lacroix, and Quoc Hung Vu

Contents

xi

The Evolution of Water Management in the Red River Delta of Vietnam and a Case of Chuc Son, Hanoi City . . . . . . . . . . . . . . . . . . 373 Tuan Anh Pham and Kelly Shannon Building Climate Change Resilience Indicators for the Rural Commune in the Northern Delta, Vietnam . . . . . . . . . . . . . . . . . . . . . . . 396 Toan Duong Thi, Duc Do Minh, and Luu Tran Thi Deriving Attributes of Walking Behavior Using GPS-Based Travel Survey and Fuzzy Logic: A Case Study in Lyon, France . . . . . . . . . . . . 429 Minh Hieu Nguyen and Jimmy Armoogum Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455

About the Editors

Prof. Dieu Tien Bui is currently a full professor at the GIS group, Department of Business and IT, University of South-Eastern Norway (USN), Bø i Telemark, Norway. He is a global highly cited researcher 2019, ranking among the top 1% by citations for Cross-Field in Web of Science. Besides, he is an editorial board member of scientific journals, Geomorphology (Elsevier), Remote Sensing (MDPI), Journal of Mountain Science (Springer), and Vietnam Journal of Earth Sciences (VAST). Also, he is a representative of the National Norwegian IAG Geomorphology Group (GeoNor) members at Norwegian universities. He received the M.Sc. degree in Cartographic Engineering from Hanoi University of Mining and Geology, Hanoi, Vietnam, in 2004, and the Ph.D. degree in Geomatics at Norwegian University of Life Sciences (NMBU), Ås, Norway, in January 2013. He was a postdoctoral researcher at NMBU between 2013 and 2014. From 2004 to 2007, he was a university lecturer at Faculty of Surveying and Mapping, Hanoi University of Mining and Geology. In 2008, he was a geospatial analyst at Ugland IT Group, a geographic information services company in Lysaker, Oslo, Norway. He has more than 200 publications, and out of them, >180 articles were published in Science Citation Index (SCI/SCIE) indexed journals, >10 book chapters published by Springer and Elsevier, and one edited book published by Elsevier. He is a reviewer for more than 30 SCI/SCIE journals. His research interests are GIS and geospatial information science, remote sensing, and applied artificial intelligence and machine learning for natural hazards and environmental problems, such as landslide, flood, forest fire, ground biomass, and structural displacement. Prof. Hai Thanh Tran is currently a professor of Geology at the Department of Geology, Hanoi University of Mining and Geology (HUMG), Vietnam. He received the MSc in Geoscience in 1997 and the Ph.D. degree in 2001, both from the University of Regina, Canada. He has been working for more than 30 years on regional geology, structural interpretation, tectonic evolution, and their relationship to natural resources and geohazards in Vietnam and other countries. He is the

xiii

xiv

About the Editors

author and co-author of nearly 40 international scientific papers, six books, and the presenter of more than 40 international presentations relating to his scientific interests. He is a member of the editorial board for several domestic and international scientific journals. Besides his official duty of the rector at the HUMG, he has also been a member of several professional organizations both in Vietnam and internationally, including vice president and president of the Earth Science Council of the National Foundation for Science and Technology Development of Vietnam (NAFOSTED), vice president (2014, 2017), and president (2015–2016) of the Solid Earth Section (SE) Session of the Asia-Oceana Geological Society.

Prof. Xuan-Nam Bui is currently a full professor at the Surface Mining Department, Faculty of Mining, Hanoi University of Mining and Geology (HUMG), Vietnam. He received the M.Eng. degree in Mining Engineering from HUMG in 2001 and the Dr. Eng. Degree in Mining Engineering from TU Bergakademie Freiberg (Germany) in 2005. He has been working at HUMG since 1996. His research interests are environment-friendly mining technology and engineering, occupational safety and health in the mining industry, and applications of artificial intelligence and machine learning in geoengineering, mining, and environmental issues, such as ground vibration, overpressure, fly rock, air pollution, and slope stability. He is the editor-in-chief of the Journal of Mining and Earth Sciences, HUMG, and an editorial board member of several international scientific journals. He has more than 60 articles published in ISI and Scopus Indexed Journals and Springer chapters. Currently, he is a member of the Society of Mining Professors and vice president of the Vietnam Association of Mining Science and Technology.

Neotectonic Activities and Its Significance to River-Course Evolution: Implication for the Cai River Catchment, Ninh Thuan Province, South-Central Vietnam Hai Thanh Tran1(B) , Chi Kim Thi Ngo1 , Hau Vinh Bui1 , Binh Van Nguyen1 , Thao Thanh Nguyen1 , Hien Thi Hoang2 , Nam Xuan Nguyen3 , and Tu Do Ngo Hoang4 1 Department of Geology, Hanoi University of Mining and Geology, Hanoi, Vietnam

[email protected] 2 Department of Geology and Minerals of Vietnam, Division of Radioactive and Rare Minerals,

Hanoi, Vietnam 3 Vietnam Institute of Geosciences and Mineral Resources, Hanoi, Vietnam 4 Faculty of Geography and Geology, Hue University of Sciences, Hue, Vietnam

Abstract. The Cai River catchment of the Ninh Thuan province locates in the South-central portion of Vietnam. This area is underlain by various rock types, including Mesozoic to Cenozoic sedimentary and volcanogenic sequences and a large volume of magmatic complexes. The Cenozoic sedimentary cover comprises thin units that are commonly deposited within small deltaic or fault-controlled alluvial basins, and along discontinuous narrow coastal plains. Pliocene-Quaternary basaltic bodies are locally covered part of the coastal zone or form small offshore volcanic islands. These assemblages are variably deformed and dismembered as consequence of long-lived tectonic movements. Recent tectonic activities are evident by numerous indicators, which include brittle fracture and fault systems straddling Quaternary sedimentary covers and recent weathering profiles, numerous fault scarps, abrupt local land uplift and/or subsidence, disruption of drainage systems, occurrence of earthquake, exposure of hot spring water and mud/sand volcanos. Other morpho-tectonic features such as the presence of Quaternary terraces, exposure and displacement of coral reef, and the development or significant modification of drainage patterns and modern basinal architecture have also resulted from active tectonic activities. Dating of the deformed materials, including fault gauge, uplifted, or subsided sediments reveal that the recent tectonic movements have taken place continually during the Late Holocene to recent. The combination of diverse lithologies, multiple tectonic fracturing and displacement of the rock units, tectonic uplift, and subsidence, in conjunction with the surface processes and sea action, have resulted in complex river course and coastal morphology. Active faulting and local tectonic movements have created long-term effects on the river flow pattern and coastal zone stability, including the shifting of flow direction, the formation of subsided or uplifted structures, sea-shore modification. Lateral tectonic movements, on the other hand, have led to the modification of the flow © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 1–17, 2021. https://doi.org/10.1007/978-3-030-60269-7_1

2

H. T. Tran et al. pattern, which consequently creates local complex flow patterns and associated hazards along the Cai River and its tributary. Keywords: Recent tectonics · Change of river course · Subsidence and uplift · South-central Vietnam

1 Introduction The Cai River catchment area of Ninh Thuan province locates in the south-central portion of Vietnam (Fig. 1). Geological works conducted previously [1–4] show that the basement of area is represented by numerous lithologies of differing composition, ages, and origins. These are overlain by fluvial-deltaic or marine aeolian deposits and subordinated volcanic rocks of the Tertiary-Quaternary in age. Both the basement assemblages and sedimentary cover have undergone poly-phases of neotectonic deformation that have taken place from the late Mesozoic to present. The strong tectonic activities have led to intense fracturing and dismemberment of both basement rocks and sedimentary cover. The neotectonic activities are currently active [5–9], which is indicated by numerous indicators including the occurrence of neotectonic/active faults and fractures, the deformation of Quaternary sedimentary cover, tectonic- controlled terrane uplift and subsidence, modification of morphological features such as abrupt change of flow direction. The interaction of many factors such as the variation in lithologies of the basement, faulting and fracturing, localized tectonic uplift and subsidence in combination with the influence of subaerial processes including climate, weathering, surface water flow, wave and marine actions [5, 6]. The drainage system along the Cai River catchment in the Ninh Thuan province is characterized by a unique and complicated pattern, which is controlled mainly by the tectonic features (Figs. 1, 2). The morphological feature and evolution of a river system have been demonstrated to be the consequence of endogenic elements such as the spatial occurrence and composition different lithological units, tectonic strain regime, crustal movements, coupled with exogenic activities under the heavy influence of tropical climate (e.g., [10–17). In addition, geohazards along the river catchment as well as in the river mouth area are also common. Some studies in this area (eg., Tran, 2015 [5]; 2020 [6]) have pointed out that the endogenic processes such as extensive faulting and eustatic movement are essential factors for the hazards to take place in the study area. Furthermore, the combined effects of ground uplift and subsidence in the context of relative sea-level rise also create additional parameters that shape the morphology and evolution as well as related geohazards in the river catchment, especially in the lower course and coastal areas related to the river basin [11, 15, 18]. However, these important endogenic parameters in the evolution of the river and its related geohazards in the area have largely been ignored or inadequately addressed in previous studies. To obtain a thoroughly understanding of the nature of the recent geological structures and their role in the evolution and related geological hazards in the middle and lower portions of the Cai River catchment area in Ninh Thuan province and establish a basis for

Neotectonic Activities and Its Significance to River-Course Evolution

3

application in assessment of the role and effect of neotectonic activities to other catchment areas along the coastal zone of central Vietnam, a comprehensive interpretation of structural features and their relationship to the evolution of the river must be carried out. This paper presence new evidences for neotectonic movements and assess their roles as regional controlling parameters in the evolution of the Cai River course and its morphological features using the combination of field mapping, structural and tectonomorphological interpretations, and age dating of the Quaternary deposits as well as fault-derived materials.

Fig. 1. A. General position of the Cai River catchment in the Ninh Thuan province, South-central Vietnam; dashed rectangle box is the area of B, area in dotted line is the approximate area of the Ninh Thuan province; small box is the location of this study; B. Generalized tectonic map showing the tectonic position of the study area in south-central Vietnam: rectangle box is the location of this study (modified after [19]); C. Generalized geological map of south-central Vietnam showing major lithologies of the underlying basement of the study area (modified from [4]); the area in the black box is the area discussed in this study.

4

H. T. Tran et al.

2 Neotectonic Features of the Cai River Catchment, Ninh Thuan Province 2.1 Basement Rocks and Sedimentary Cover a. Basement units The Cai River catchment area occurs along the southeastern margin of Da Lat structural block, a tectonically complex terrane (Fig. 1B) comprising predominant Jurassic sedimentary units or intruded variably by numerous types of plutonic complexes (Fig. 1C; [1–4]). These are partly overlain by Cenozoic sediments and local volcanic layers that dominantly cover as well as shape the topography of the lower course and rivermouth portions of Cai River (Fig. 1C). Within the area, oldest sedimentary units exposed are Mesozoic active continental margin assemblages including Lower to middle Jurassic volcanogenic clastic sedimentary units, which exposed in the western part of the area (Fig. 1C). These units are locally unconformably overlain by rhyolitic to dacitic volcanics of Cretaceous in age. In places, small outcrops of Neogene-Quaternary basalts are locally occurred. The Mesozoic rocks are largely intruded by numerous batholith-style felsic intrusive bodies of Late Cretaceous in age, which produced a region-wide basement of the study area (Fig. 1C). b. Quaternary cover Cenozoic sedimentary sequences cover dominantly the central part and along the coastal plain of the study area (Fig. 1C, 2). These consist of mostly Quaternary unconsolidated deposits of various origins and can be classified based of their age and origin [1–3]. Pleistocene deposits comprise marine, fluvial, alluvial and aeolian deposits that expose as the terraces, forming the foot-hill areas in the western part, whereas the aeolianmarine sediments commonly occur as remnants of dunes and plains along the eastern portion, particluarly the coastal zone of the area. Poly-origin Holocene deposits consisting of alluvial, fluvial, lacustrine/lagoon, marine and aeolian sediments overlay the modern alluvial valeys along the Cai River catchment and coastal area (Figs. 1C, 2). 2.2 Structural Features The area has experienced strong neotectonics activities. All lithological units and Quaternary deposits in the area are variably deformed, indicated by fracturing, partly dismembering, and/or displacing that are caused by many phases of regional deformation [5, 6, 19]. Field mapping and structural interpretation have identified numerous paleoand neotectonic structures, including faults and fracture zones of differing orientations and ages (Fig. 2). Many of fault systems apprear to be multiple reactivated and are remained active throughout the Quaternary.

Neotectonic Activities and Its Significance to River-Course Evolution

5

a. Paleotectonic structures The Paleotectonic faults and fractures, which are considered to be predated 5 Ma in age (eg., [6, 20]), are widely documented within pre-Quaternary lithotectonic assemblages. These comprise numerous cross-cutting systems, including northwest-southeast, northeast-southwest, longitudinally, and trending [5, 6]. They can be identified by numerous evidences including large zones of brecciation, slickensides, and systemtic distribution of fractures [5, 6]. Kinematic indicators identified from the faults and fracture zones as well as the cross-cuting relationship between different systems indicate that the movement history along the paleotectonic faults are complex, including reverse, strike-slip or oblique-slip (Fig. 2). Many fault systems are multiple reactivated, which are indicated by the presence of several generations of overprinting and cross-cutting fault-generated products. b. Neotectonic structures In the Cai River catchment area, numerous neotectonic fault and fracture systems, which are postdated 5 Ma in age (eg., [6, 20]), have been recognized (Fig. 2). Field mapping and structural interpretation have recorded many evidences demostrating the occurence of the neotectonic structrues, especially active fault zones. These include the widespread exposure of brittle slickensides, unconsolidated fault gouge, and numerous open-spaced, systematic fractures that straddle and dismember the Quaternary deposits and/or weathering profiles/regoliths (Fig. 3). Furthermore, many tectonogeomorphological markers resulted from neotectonic and active movements such as linear distribution of triangular facets, fault scarps, uneven occurrence of drainage systems, colluvial cones and alluvial fans, or abrupt change of flow direction of flows, local zones of fault-controlled uplift or subsidence (eg., [11, 15, 21] and references herein) are also widely recognized within the study area (Figs. 2, 3, 4 [5, 6, 19]). In many places, the paleotectonic structures are overprinted or reactivated during neotectonic movements, which have locally created zones of multiple displacements during the Quaternary. In such zones, the overprinting of constrasting slip directions identified on the slickensides points towards a unstable neotectonic regime with opposing movements that has strongly affected the architecture as well as the spatial occurence of the Quaternary sediments along the Cai River course and its tributary (Figs. 2, 3, 4). Besides, the existence of active seismic activities [22], which has been recorded systematically recently [6, 8, 9], as well as the evidence for a recent volcanic eruption in nearby areas, both onshore and offshore, and presence of numerous hot springs, are direct accounts for strong active tectonics in the area (see Tran, 2020 [6] for detailed data).

6

H. T. Tran et al.

Fig. 2. Generalized neotectonics map of the middle-lower courses of Cai River, Ninh Thua.n Province constructed based on various data sources (see [1–3, 5, 6]). Insert show the general distribution of fault systems and their approximate relationship to local stress-field during the Quaternary. Note that the main flow of the Cai River was migrated from the southwest to the northeast of the study area during the recent time, indicated by numerous evidences for the abandoned currents of the Cai River that are widespread in the area and can be used to delineate and restore at least parts of the paleo-flows (Fig. 2). The widespread of the bending and sudden change of flow direction is related to the lateral movement of many active faults, whereas subsided areas are commonly related to the fault-controlled, pull-apart style basin (e.g., Dam Nai area, Fig. 2). The dominant northeasterly-directed migration of the flows is considered the consequence of a major uplifting of the crust in the south of the rea (eg., the Ninh Phuoc Uplift).

2.3 Morphological Variation Related to Neotectonics Movements in the Cai River Area The current geomorphology and landscape along the middle and lower courses of the Cai River area in Ninh Thuan has derived from the development of a number type of landform [1, 2] that are controlled by a combination of neotectonic endogenic and exogenic processes [11–16, 21, 23]. This includes the occurence and spatial distribution of various type of lithologies, their internal structures and physical property, tectonic structures, seismicity activities, water activities, sedimentary transportation and deposition, wind and sea/wave actions, weathering and erosion and other climatic elements. Based on the examination of the comprehensive relationship between tectnomorphological features with basement architecture, neotectonic endogenic and subaerial exogenic processes, a number of genetic morphological types have been catergorized in the study area [1, 2, 5, 6], consisting of erosional, fluvial, marine, lagoonal, aeolian, and composite landforms. The landforms related to erosion commonly occur above the exposed lithologies and within the uplifted terranes. The fluvial-derived landforms consist of valleys and plains developed along the drainage systems, which commonly create flood plains, terraces or channels that formed by the combinaiton of subsidence or uplift coupled with the influcence of faulting and fracturing of the basement (Figs. 3 and 4). In many cases, the morphology and spatial distribuiton of the surface flows are governed

Neotectonic Activities and Its Significance to River-Course Evolution

A

B

C

D

7

Fig. 3. Examples of field occurrence of neotectonic structures in the middle and lower course portion of the Cai River, Ninh Thuan province: A. A large fault zone developed in intrusive rocks and Quaternary cover crops out in the Nha Trinh area. Note the white arrows point to the fracture zones as parts of the fault straddle the Quaternary sedimentary units; B. A slickenside as part of a large, longitudinally trending fault exposed at the Dinh Co Pass, northeast of the area (After Tran, 2017 [5]); C. A Riedel-style brittle fault zone in granitic rocks in Nha Trinh Dam area; D. Multiple cross-cutting faults, which produced complete fracture zones and dismemberment of the rock bodies seen at Nha Trinh Dam; E. A NE-SW trending fault crosses cut the NeogeneQuaternary basaltic layers observed along the left bank of Cai River in Nha Ho area; F. An angular unconformity in Quaternary deposits exposed in Nha Ho area: dotted yellow line is the trace of unconformity, dashed white line is trace of bedding; G. A large outcrop of folded and faulted Quaternary deposits seen in My Tan, east of the study area (see Tran, 2020 [6]): dashed yellow line is the trace of folded bedding, dashed red lines are the fault zone; rectangle boxes are location of figs H and I; H. Part of a fault zone in G, red arrow indicates the width of fault zone in the picture; I. part of G being sampled for radiocarbon and TL dating (after Tran, 2020 [6]).

by the underlying neotectronic structures, particular large fault and fracture zones (eg., [11, 15, 21]; Figs. 2, 3, 4). The lagoonal and marine derived landforms including marine terraces, coastal plains, lagoons and beaches occurring along the lower course and rivermouth portion of the sutdy area (Figs. 3, 4). The aeolian landforms are widespread in the eastern portion of the area and comprises both depsotional and ersional types: the earlier comprises dunes, barchan dunes, and locally thin loess layers that are locally formed elongated sand hills adjacent to the shoreline in the southeastern portions of the area (Fig. 2), whereas the later commonly occurs as ventifacts, yardings and pans in the high-relief, bed rock-exposed areas along the coastal zone of study area.

8

H. T. Tran et al.

Fig. 3. (continued)

3 The Controls of Neotectonic Activities on the Flow of the Cai River and Its Tributary In general, neotectonic activities act as a major factor in controlling the morphology of one area, which has been well documented in the literature (eg., [11, 12–17, 22–28]). The neotectonic movements that include uplifting, subsidence or lateral transportaiton along the neotectonic faults, are the most important factors governing the formation or migration of the flow and its morphology, the spatial distribution of drainage networks, evolution of river valley, plain, and terrace and other features. The spatial occurence of flow networks in the study area, on the other hand, could also be the consequence and therefore reflects the nature of neo- and active-tectonic movements (e.g., [15, 25]). In the study area, the the discharges and major flows the Cai River tributary are either oriented along the neotectonic faults or fracture zones or being offset and diverted the flow direction due to the displacement of the active faults or being influenced by the local uplift/subsidence (Figs. 2, 5). In many places, the sudden change in flow direction

Neotectonic Activities and Its Significance to River-Course Evolution

9

Fig. 4. Examples of tectonic morphology in the study area: A. Residual fault scarps with triangular facets of a large scale, NE-SW trending fault zone observed in Khanh Son area (photo looking SW). The dashed red line is the approximate trace of the fault, whereas thick red arrows point towards the hanging wall block. The subsidence of the hanging wall block led to the formation of the flood-plain wetland (the front part of the picture); B. A section of residual fault scarp with triangle facets seen in the NW Son Hai area. The dashed red line indicates the piedmont scarp (after [19]); C. Linear valley developed along a fault zone; D. A trough developed along a fault the wall on the left is a scarp (slickenside) of the fault with slickenline on the surface. The red arrow indicates the direction of the striation. The formation of trough or valley in C and D also indicates strong tectonic uplift, which led to the rising of the base-level and incision to form straight and steep-slope valleys and/or troughs; E. A mountain slope consists of linear, steep-side troughs and exposed granitic rocks, which formed by the erosion of the rocks along the fracture zones and ¸ rapid uplifting of the area; F. Different marine terrace levels seen in An Hai area, the numbers II-IV equivalent to 2nd to 4th level respecterly (after [19]); G. Remnant of the first (I) and second (II) marine terraces observed in Vinh Hai area, in which the 2nd terrace occurs as remnant eroded surface on top of granitic rocks; H. A platformal-type coral reef exposed either due to relative sea level fall or uplifting of the sea bottom seen in Thai An area (Ninh Hai); I. Two levels of coral reefs exposed in the Nui Chua area in which the level (0) is the younger reef that uplifted recently, 1st level (I) is older reef has been uplifted about 5 m above current sea level. Yellow dots indicated the position for radiocarbon dating samples (after Tran, 2015 [5]).

commonly coincides with the location of fault transect, and the shift of flow direction is controlled by the movement of the fault walls (Figs. 2, 5). The cross-cutting and opposing movements of the fault sets have led to the complex zig-zag shape of the flow in many places along the Cai River and its branches (Fig. 2, 5). The uplift due to neotectonic activities commonly results in the base-level rise, which leads to vertical incision and vertical ersoison that consequently straightens the river course and development of V-shape river valleys, angular and antecedent flow networks, exposure of marine or river terraces, and sea-wards expansion of the deltaic

10

H. T. Tran et al.

Fig. 4. (continued)

plain and river mouth (eg., [14, 15, 25]). The uplifting walls of the faults also create systematic residual fault scarps that are abundant in the area (Figs. 2, 3, 4 and 5). In contrast, subsided areas are localized and commonly occurred within the graben or halfgraben caused by pull-apart or horse-tail portion of the strike-slip faults [29], or along the hanging walls of the normal faults (Fig. 4A, B). The subsidence also leads to the rise of base-level, lateral erosion of the riverbanks and formation of meanders, mash, oxbow lake, flood-plain, and lagoon in the lower portion of the river. The general model for interrelationship between neotectonic faulting and movements to river shape and morphology is presented in Fig. 6.

4 Timing of Neotectonic Movements Along the Cai River Area and Environs As described above, the evidence for neotectonics movements is abundant in along the Cai River catchment, including the uplifting or subsidence and movements along the fault. Some works recently have focused on the dating of neotectonic events to precisely define the exact timing of recent tectonic movements [5, 6, 30, 31]. Michelli (2008) [30] had used δ 180 and δ13C isotopes dating for beachrocks, and beach ridge has determined the fluctuations of sea-level from mid-Holocene to recent in which the coastal zone of Ninh Thuan area has experienced a sea level highstand during the period from 6721 to

Neotectonic Activities and Its Significance to River-Course Evolution

11

Fig. 4. (continued)

5869 cal yr BP that followed by a period of sea level rise from 5687 to 5377 cal yr BP. This was continued by a sea-level fall period starting after the 5377 cal yr BP until recent time. Stattegger, et al. (2013) [31] used AMS 14C dating methods for beachrocks, beach ridge, washover and backshore deposits along the coastal zone including Ninh Thuan area have argued that the area was influenced by Holocene sea-level fluctuations that caused by eustatic and isostatic movements, including a final phase of sea-level rise due to deglaciation from −5 to +1.4 m during the period of 8.1 to 6.4 ka, followed by the mid-Holocene, slightly above +1.4 m sea level highstand between 6.7 and 5.0 ka and

12

H. T. Tran et al.

Fig. 5. DEM model and photo-lineament maps showing the relationship between the distribution of the drainage systems and their relationship to the neotectonic lineaments, many are neo- and active faults along the Cai River catchment. Red lines are topographic lineaments, which in many cases, are neotectonic fault traces. The areas within blue dashed lines are strong uplifting; the areas within the purple dashed lines are strong subsided; red lines are defined or extrapolated neotectonics faults that affect the formation or modify the flow pattern of the drainage systems. In many places, the uplifting or subsidence governs the flow pattern (e.g., such as Ninh Phuoc Central Uplift), whereas the faults control the direction and/or the (abrupted) change of the flow direction. The movement along with conjugate, strike-slip fault systems generally lead to the formation of the Z or S shape; zig-zag flows along the river; the development of horse-tail and or pull apart, or imbricate structures along the fault zones commonly lead to the formation of the fault-controlled grabens and/or hosts along the river catchment.

reached to maximum of +1.5 m at 6.0 ka. These followed by a post 5.0 ka sea-level fall linearly below +1.4 m at a rate of 0.24 mm/year, which lasted until 0.63 ka and at + 0.2 m. These studies, however, did not count for the effects of local neotectonic movements on such sea level variation in the studied areas but considered the sea level changes were a consequence of eustatic and globally glaciation. On the basis of the identification of the deformation and/or displacement of the Quaternary deposits or weathering profiles that are widely occurred in the area, Tran (2015, 2020) [5, 6] have attempted to date the Quaternary tectonic movements using several modern methods including (radiocarbon,

Neotectonic Activities and Its Significance to River-Course Evolution

13

Fig. 6. The general model summarizes the interrelationship between flows and neotectonic elements in the middle and lower courses of the Cai River catchment, Ninh Thuan province.

ESR, TL and OSL techniques (eg., [11, 15, 32]) for different materials collected from the study area. Tran (2015) [5] has dated the fault gouge collected from fault zones in Ninh Hai area, east of this study area using ESR method and obtained the ages for lastest phase of fault movement including ca. 13,30 ka and 33.1 ka for a fault gouge zone (eg., at cordinator: 11 42 05832 N and 109 11 2.9256 E) and 32,60 ± 3,20 ka for another sample (at cordinator: 11 42 01908 N and 109 11 23028 E). This indicates that the latest phase of movement along the studied faults could have been at ca. 13 ka (eg., see [5] for detail). In addition, on the basis of identification of the uneven distribution and displacement of the Quaternary terraces that were caused by a large active fault in the Nha Ho area on the left bank and Phuoc Vinh area on the right bank of the Cai River, west of this study (Figs. 2, 7), Tran (2015) [5] has also dated the Holocene units that were truncated and displaced by the fault (Figs. 7A and B) by OSL technique. The results show that the same unit with similar ages occurred on opposing walls are significantly displaced (Fig. 7B). The dating results also show a complex movement history of the fault with opposing sense of movement, in which the earlier phases taken place from ca. 6.3 ka to 3.0 ka had controlled the deposition of sedimentary layers with a NWside down relatively producing beds with greater thickness compare to those in the SE side, whereas the latest phase took place after ca. 2.6 ka, truncated all sediments with a northeast side-down, sinistral oblique movement ( see Figs. 7A and B). Furthermore, on the basis of the discovery of a pronouncedly deformed Quaternary sedimentary unit in the My Tan area, east of this study, which led to the significant folding and faulting of the sedimentary layers (see Fig. 4G–I), Tran (2020) [6] has collected samples for δ13C radiocarbon dating of the remnants of in situ materials (eg., shells and roots collected within the sedimentary layers) and got the ages ranging from ca. 5.9 to 4.89 ka for older and younger beds within the strigraphic sequence, respecterly (Fig. 8). Based on the bedding configuration and ages of the beds within the sedimentary unit, Tran (2020) [6] has extrapolated the youngest age of the sedimentary layers in the studied location is at ca. 2.2 ka. However, as the exposed beds in the outcrop (Figs. 4G, 8) is not the top of the sequence, this means that the top of the seidmentary unit in this area may be younger than 2.0 ka and therefore the deformation of these beds, including folding and subseqent faulting (Fig. 8) must have taken place at least after the age of ca. 2.0 ka (Fig. 8).

14

H. T. Tran et al.

Fig. 7. A. The relationship between terrace, river and fault movement in Nha Ho - Phuoc Vinh area: red dashed line is the trace of a fault straddling the Cai River; arrows indicate senses of strike-slip movement; thick red arrows indicate the vertial movement direction of the uplifted and subsided walls of the fault respectively. Yellow dots and numbers in yellow boxes are sample collection locations for OSL dating in Fig. 7B (after Tran, 2015). View of the picture looking east; B. A northwest-southeast cross-section straddles the left and right banks of the Cai River, from Nha Ho (NW) to Phuoc Vinh (SE) modified after [5, 19], showing the relationship between the active fault and the distribution of Quaternary sediments and their ages. Dating of the maker beds reveals complex movement history along the faults, which resulted in the variation in thickness and height of the sediments in the NW and SE walls of the fault. The latest phase of the fault is oblique, NW side-up sinistral strike-slip. Approximate locations for sample NT305/2 and NT 305/3 are: 11.6293631 N; 108.8830180 E, for samples NT 306/2 and NT 306/3 are: 11.6293631 E and 108.8712300 E. See Tran, 2015 [5] and Tran 2017 [9] for details of sample locations, dating methods and discussion of results.

Neotectonic Activities and Its Significance to River-Course Evolution

15

Fig. 8. Generalized cross-section schetched from the outcrop in Fig. 4G, showing the relationship between deformation and age of the Quaternary sedimentary beds in My Tan area, east of the study area (cordinator: 15.8511014N, 108.3827994E). The age of the beds was obtained by δ13C isotopes dating of the remnants of shells collected within each bed that expressed in year BC (see Tran, 2020 [6] for detail of sampling and dating). The data shows that the dated layers occur in the middle of the dtratigraphic sequene, and the top layer of the sequence has not been dated. However, based on the age data of 3 sucessive layers have been dated (eg., ca. 5960, 5490, 4890 ya, from lower to upper part of the sequence, respectly) and the defined younging direction of the sequence, the average age of the top layer in the section can be extrapolated to be at least ca. 2.2 ka. This indicates that the age of the youngest beds of the sequence (which extend far to the southeast of the outcrop) should be equivalent to at least ca. 2.0 ka. In this case, the pronounced folding and later faulting events in the area had taken place after 2.0 ka.

5 Conclusion Systematic field mapping and structural interpretation of the geological parameters and their interrelationship to the surface flow networks along the middle and lower courses of the Cai River and its tributary, Ninh Thuan province have shown a complicated regional structural pattern that was created by long-lived regional neotectonic evolution, which are remain active. Active tectonism has acted as a controlling factor in producing the geomorphology and landscape of the study area. The neotectonic overprints, caused by many phases of deformation that produce non-penetrative, open-spaced brittle fracture and fault zones, have not only cross-cutted the basement rocks but also disrupted the Quaternary cover. This consequently led to both vertical and horizontal displacements, including local uplift and subsidence, tilting, or dismemberment of both undercover lithologies, the Quaternary deposits as well as the Cai River and its tributaries. The combined effects of all structural elements within a framwork of neotectonic endogenic and exogenic activities have produced the unique morphology of the Cai River and its tributary in the study area. Localized vertical movements within generally fault-controlled block has led to local rise or fall of base level and consequently the formation of braided or meandered flows, uplifted blocks or subsided basins within area of generally regional uplift or subsidence, respectively. In addtion, many basins and and/or uplifting areas along the Cai River are also products of fault movement and the relationship between faults and fractures within a system in the form of extensional fans/pull-apart basins or contractional impricate structures. Moreover, the local uplifting and tilling of the basement has also greatly influenced the abandontment and migration of the river courses from the uplifting, higher base-levels to the areas of lower-base levels. The widespread cross-cutting and lateral displacement along active structures, on the other hand, have led to deformation and redirection of the flow direction, enhancing

16

H. T. Tran et al.

the affects of exogenic processes including vertical and lateral erosion that have resulted in the modification of the riverbanks and its related coastal area of the Cai River in Ninh Thuan. Dating of the neotectonic generated elements have revealed that the study area is currently tectonically activated and the recent movements, both vertical and lateral, contribute significantly to the formation and evolution of the modern morphology including the drainage system in the study area. Thus, neotectonic structures and activities, especially those were formed recently are critical parrametors for the shaping of current morphology, landscape, and spatial distribution of the drainage networks in the middle and lower courses of Cai River catchment in Ninh Thuan province. Therefore, detailed structural interpretation, including adequatly documentation of all structural evements, their relative timing, especially for those have been originated or activated by neotectonism are therefore very important and, as such, must be carried out systematically and detailedly in any geological works aimed to understand the nature of a river course, especially in assessment and prediction of the evolution of river and its associated natural hazards. Acknowledgments. This work is partly supported by the Ministerial Project B2019-MDA-562– 15 to Hai Thanh Tran. We thank two anonymous reviewers for their constructive comments and corrections, which have helped to improve the quality of this work.

References 1. Phuong, H., Nguyen, N.S., Trinh, N.T.: Geological and Mineral Resources map of Phan Rang-Thap Cham Urban area scale 1:25.000. Department Geology and Minerals of Vietnam Achieves, Hanoi (1999) 2. Stepanek, P., Ty, H.T.: Report on mineral geological mapping at Phan Rang-Cam Ranh at 1:50,000 scale. Archived at the Center for Information & Archives of Geology, Hanoi (1986) 3. Thang, N.D. (ed.): Report on surveying and mapping of Geology and Minerals in Phan RangNha Trang area at 1:200,000 scale. Archived at the Center for Information & Archives of Geology, Hanoi (1988) 4. Tran, V.T., Vu, K.: Geology and Natural Resources of Vietnam. Natural Sciences and Technology Publishing House, Hanoi (2011) 5. Tran, T.H.: Research on and evaluation of active tectonics in central Vietnam coastal area and its role for the natural hazards for predicting and mitigation of natural hazards in the climate change condition (Vietnamese). Final Report of National Project BÐKH 42 (2015) 6. Tran, T.H.: Research on the impacts of neotectonic on the modification of main courses of river catchment in Central Vietnam for the protection of river, applied for Vu Gia - Thu Bon catchment (Vietnamese). Final Report of National Project BÐKH 13/16-20 (2020) 7. Pham, V.H.: Characteristics of neoteconic faults in South Central Vietnam. Ph.D. thesis, Geological Institute, Academy of Science and Technology of Vietnam, Hanoi (2002) 8. Nguyen, H.P.: Map of earthquake dangerous rate in Vietnam and the East Sea. J. Sci. Earth 26(2), 97–111 (2004). (in Vietnamese with English abstract) 9. Nguyen, H.P., Pham, T.T.: Probabilistic seismic hazard maps of Vietnam and the East Vietnam Sea. J. Mar. Sci. Technol. 15, 77–90 (2015) 10. Easterbrook, D.J.: Principles of Geomorphology. McGraw-Hill Book Company, New York (1969)

Neotectonic Activities and Its Significance to River-Course Evolution

17

11. National Research Council (NRC): Active Tectonics: Impact on Society. Study in Geophysics (1986) 12. Hurtrez, J.E., Lucazeau, F., Lavé, J., Avouac, J.P.: Investigation of the relationships between basin morphology, tectonic uplift, and denudation from the study of an active fold belt in the Siwalik Hills, central Nepal. J. Geophys. Res. Solid Earth 104, 12779–12796 (1999) 13. Pidwirny, M.: The drainage in concept. In: Fundamentals of Physical Geography. Physical Geography.net (2006) 14. Huggett, R.J.: Fundamentals of Geomorphology. Taylor & Francis E-library (2007) 15. Burbank, D.W., Anderson, R.S.: Tectonic Geomorphology. Wiley, Hoboken (2011) 16. Whitney, B.B., Hengesh, J.V.: Geomorphological evidence for late quaternary tectonic deformation of the Cape region, coastal west central Australia. Geomorphology 241, 160–174 (2015) 17. Lawrie, K., Brodie, R.S., Gibson, D., Halas, L., Magee, J.: Inter-disciplinary approaches to mapping neotectonic intra-plate fault systems & their hydrological effects in Australia’s floodplain landscapes. In: Lawrie, K. (ed.) Neotectonics on the Australian Plate: New Science for Energy, Mineral and Groundwater Systems, and Hazard Assessment (Abs), p. 31 (2016) 18. McGuire, B., Maslin, M.A.: Climate Forcing of Geological Hazards. Wiley-Blackwell, Oxford (2013) 19. Tran, T.H.: Active tectonics in the coastal zone of Central Vietnam and its impacts on the geohazards. Science and Technics Publishing House, Hanoi (2017). 218 pp. 20. Moores, E.D., Twiss, R.J.: Tectonics. Freeman and Company, New York (1995) 21. Molnar, P., Anderson, R.S., Anderson, S.P.: Tectonics, fracturing of rock, and erosion. J. Geophys. Res. 112, F03014 (2007). https://doi.org/10.1029/2005TF000433 22. Keller, E.A., Pinter, N.: Active Tectonics: Earthquakes, Uplift, and Landscape. Prentice Hall, Upper Saddle River (2001) 23. Ritter, M.E.: The physical environment: An introduction to physical geography. http://www. earthonlinemedia.com/ebooks/tpe_3e/title_page.html (2006) 24. Ritter, D.F., Kockel, R.C., Miller, J.R.: Process Geomorphology. Waveland Press, Inc., Long Grove (2002) 25. Schumm, S.A., Dumont, J.F., Holbrook, J.M.: Active Tectonics and Alluvial Rivers. Cambridge University Press, Cambridge (2000) 26. Malik, J.N., Mohanty, C.: Active tectonic influence on the evolution of drainage and landscape: geomorphic signatures from frontal and hinterland areas along the Northwestern Himalaya, India. J. Asian Earth Sci. 29, 604–618 (2007) 27. Keil, M., Neubauer, F.: Neotectonics, drainage pattern and geomorphology of the orogenparallel Upper Enns Valley (Eastern Alps). Geologica Carpathica 62(3), 279–295 (2011) 28. Vita-Finzi, C.: River history. Philos. Trans. R. Soc. A 370, 2027–2263 (2012) 29. Wu, J.E., McClay, K., Whitehouse, P., Dooley, T.: 4D analogue modelling of transtensional pull-apart basins. Mar. Pet. Geol. 26, 1608–1623 (2009) 30. Michelli, M.: Sea-level changes, coastal evolution and paleoceanography of coastal waters in SE-Vietnam since the mid-Holocene. Ph.D. thesis, Christian-AlbrechtsUniversität Kiel, Germany (2008) 31. Stattegger, K., Tjallingii, R., Saito, Y., Michelli, M., Nguyen, T.T., Wetzele, A.: Mid to late Holocene sea-level reconstruction of Southeast Vietnam using beachrock and beach-ridge deposits. Glob. Planet. Change 110(Part B), 214–222 (2013) 32. Walker, M.: Quaternary Dating Methods. Wiley, Hoboken (2005)

Mining-Induced Land Subsidence Detection by Persistent Scatterer InSAR and Sentinel-1: Application to Phugiao Quarries, Vietnam Bui Xuan Nam1(B) , Tran Van Anh2 , Luyen K. Bui3 , Nguyen Quoc Long4 , Thi Le Thu Ha4 , and Ropesh Goyal5 1 Department of Surface Mining, Hanoi University of Mining and Geology, Hanoi, Vietnam

[email protected] 2 Department of Photogrammetry and Remote Sensing, Hanoi University of Mining and

Geology, Hanoi, Vietnam 3 Department of Geodesy, Hanoi University of Mining and Geology, Hanoi, Vietnam 4 Department of Mine Surveying, Hanoi University of Mining and Geology, Hanoi, Vietnam 5 Indian Institute of Technology Kanpur, Kanpur 208016, Uttar Pradesh, India

Abstract. Mining activities can cause mining hazards, among them land subsidence occurring in mine areas and their vicinities. Mining-induced surface subsidence is often quantified by traditional geodetic surveying with an advantage of high accuracy, but it is usually applied in small areas. With the development of radar technology, there have been many studies applying radar interferometry technique to determine surface subsidence over wide areas at a few millimeters accuracy. In this paper, 24 Sentinel-1 SAR images are used with the persistent scatterer InSAR method to determine land subsidence of the Phugiao quarries and surrounding areas. The results are compared with average annual subsidence of 13 surveying points at the Phugiao1 mine and 16 surveying points at the Phugiao2 mine measured by global navigation satellite system (GNSS) technology from January 2018 to March 2020. The correlation coefficients of annually average land subsidence of the two methods are 0.93 and 0.82 at the two areas (Phugiao1 and Phugiao2), respectively, indicating the feasibility of applying InSAR Sentinel-1 data with the persistent scatterer InSAR method to determine surface deformation in quarries in Vietnam. Keywords: InSAR · Persistent scatterer · Deformation · Sentinel-1 · Quarry · Mine · Vietnam

1 Introduction Land subsidence due to exploration of natural resources such as groundwater or minerals is common in many areas of the world, e.g., in Texas [1], California (United States) [2], Bangkok (Thailand) [3], and Jakarta (Indonesia) [4]. This so-called mining hazard causes adverse impacts, which in turn result in damage to constructions [e.g. 5, 6]. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 18–38, 2021. https://doi.org/10.1007/978-3-030-60269-7_2

Mining-Induced Land Subsidence Detection by Persistent Scatterer

19

Mining-induced surface subsidence is often quantified by traditional geodetic surveying using total stations, global navigation satellite system (GNSS) receivers or terrestrial laser scanners with an advantage of high accuracy [e.g. 7, 8, 9]. However, this method is only suitable for small and accessible areas. With the development of radio detecting and ranging (radar) technology, the interferometric synthetic aperture radar (InSAR) technique can be applied effectively to determine surface subsidence over broad areas at a few millimeters accuracy [10]. Differential InSAR (DInSAR) is an InSAR technique, which was first applied to Seasat satellite imagery to study small elevation changes over an area of 50 km2 in the Imperial Valley, California, USA [11]. This method uses at least two SAR images acquired at different times of the same location on the Earth’s surface to detect the displacement by measuring the phase difference between acquisitions. However, this method has limitations because it is unable to reduce some error and noise sources, e.g., atmospheric artefacts, orbital errors [e.g. 12]. Multi-temporal InSAR (MT-InSAR) methods were therefore proposed to reduce these error and noise sources [e.g. 13]. These methods work by analyzing a network of multiple acquisitions to derive the deformation time series and thus deformation rate [e.g. 14]. Small baseline subset (SBAS) [e.g. 15, 16– 21] and permanent or persistent scatter (PS) InSAR [e.g. 22, 23, 24] are among the most commonly used MT-InSAR methods. SBAS makes use of a network of multiple master images linking multiple interferograms limited in their temporal and/or perpendicular baselines, and thus increases its redundancy, particularly in case of low signal to noise ratio [25]. Persistent scatterer InSAR adopts a network of a single master image with no multilooking applied [e.g. 22, 23]; hence, it is able to detect deformation of single scatterers, and thus small objects. As a result, it is particularly appropriate for areas with highly coherent natural or man-made objects, e.g., rocks, buildings, roads [10]. The basic principle behind the PSI method is the use of a series of SAR images in the same area to extract PSs to determine land deformation. Persistent scatterer InSAR has been widely used showing good results of land deformation detection at an accuracy of up to a few millimeters [10]. For example, López-Quiroz, Doin, Tupin, Briole and Nicolas [18] applied 38 Envisat Advanced Synthetic Aperture Radar (Envisat ASAR) images to determine subsidence in Mexico City. Liu, Luo, Chen, Huang and Ding [26] used 26 European remote-sensing satellite 1/2 (ERS1/2) images to determine subsidence in Shanghai, China. In Indonesia, radar technology was applied for the first time in land subsidence determination over Jakarta in 2001 [27]. In this study, 17 Japan Earth Resources Satellite-1 (JERS-1) SAR images acquired from February 1993 to September 1998 were used to form 41 interferometric pairs relying on the PSI method with spatial baselines of 1000 m or less. The research shows that, during the period of 1993–1995, Jakarta was subsided ten centimeters, whilst from 1995 to 1998 subsidence was six centimeters. In addition to using radar imagery to identify urban subsidence, InSAR has also been applied in mining-induced land subsidence [e.g. 28, 29–32]. Baek, Kim, Park, Jung, Kim and Kim [33] applied JERS-1 images with the small baseline subsets (SBAS) method to determine surface subsidence of Gangwon-do coal mines, Korea with the standard deviation of subsidence estimated being 7.8 mm. He, Liu and Yue [34] determined land subsidence of the Hebei mine, China by DInSAR. Diao, Wu, Hu, Li and Zhou [35] used

20

B. X. Nam et al.

a series of SAR images to identify subsidence of Hainan coal mines, China. The abovementioned studies applied different MT-InSAR methods. For instance, Diao, Wu, Hu, Li and Zhou [35] utilized the advantage DInSAR method with probability integration to detect the settlement on a large scale, while Ma, Cheng, Yang, Zhang, Guo and Zou [36] applied the SBAS method to determine land subsidence of Bu’ertai Mine, Shendong coalfield, China. In Vietnam, the application of InSAR to determine land subsidence are mainly limited to urban areas, e.g., Hanoi and Ho Chi Minh cities. For example, Tran, Tran, Nguyen, Ho, Tran, Nguyen and Luong [37] used 27 Italian space agency (ASI)’s constellation of small satellites for Mediterranean basin observation (COSMO-SkyMed) images to determine subsidence of Hanoi urban areas due to groundwater extraction. Le and Nguyen [38] determined subsidence in Ho Chi Minh City using Advanced Land Observing Satellite Phased Array type L-band Synthetic Aperture Radar sensor (ALOS PALSAR) images by persistent scatterer InSAR. However, to the best of the authors’ knowledge, InSAR has not been applied to determine mining-induced land surface deformation in Vietnam to date. This study therefore focuses on ground subsidence detection in Phugiao quarries, Vietnam and surrounding areas by the PSI method using Sentinel-1 data from 2018 to 2020. Sentinel-1 are selected in this study thanks to its free accessibility with regularly repeated acquisition at a 12-day interval, which makes it a great data source for monitoring surface deformation in mining areas. GNSS measurements observed at some locations surrounding the study area at the same time of InSAR acquisitions are adopted for validation.

2 Study Area and Data 2.1 Study Area Phugiao quarries are located in Binh Duong province, which is of many different types of land and suitable for industrial plantations, civil construction and industrial development. Until now, there have been 35 companies for mining with more than 1400 hectares in Binh Duong province [39]. In addition, Binh Duong has many special mineral resources, e.g., magmatic or weathered rocks, sediments. These mineral resources provide the province’s traditional industries such as ceramics, construction materials and mining. The study area of Phugiao quarries is located in Phugiao district, Binh Duong province with the coordinates ranging from 11o 11 to 11° 29 north latitude, and 106° 38 to 106° 57 east longitude, bounded by the pink box in Fig. 1. There are 17 quarries, which are being exploited at different depths in which the deepest ones are the Chang Tan III and Phugiao IV mines, both currently at approximately −90 m. According to the provincial mineral exploitation plan, the quarries will be exploied down to −150 m.

Mining-Induced Land Subsidence Detection by Persistent Scatterer

21

Fig. 1. The study area referred by the pink box. The red box indicate Sentinel-1 extension with three sub-swaths

2.2 Data Used The C-band Sentinel-1 data (∼5.6 cm wavelength) are utilized in this study owing to its free accessibility. Sentinel-1 satellites operate with four working modes, which are Stripmap (SM), Interferometric Wide swath (IW), Extra-Wide swath (EW), and Wave mode (WV) with different spatial resolutions. Sentinel-1 consists of two platforms currently operating in the orbit, including Sentinel-1A (launched in April 2014) and Sentinel-1B (launched in April 2016). For land surface subsidence studies applying the PSI method, images at the single look complex (SLC) format are normally adopted. In this study, SLC images at the IW mode are used. The data were downloaded from the Alaska Satellite Facilities (ASF) website of NASA [41]. In total, 24 Sentinel-1B dual polarization (VV + VH) images captured in the study area in the descending orbit, at the path number 18 and the row number 554 have been downloaded. The images were selected according to the criteria of a good weather day, no rain, and combined with the master image creates a short spatial base lines. Then the data are processed to separate the VV polarization. Details on the Sentinel-1B data used in this study, including acquired dates and perpendicular baselines with respect to the master image are shown in Table 1, with their frame extension shown in Fig. 1 by the red box.

22

B. X. Nam et al.

Table 1. Sentinel-1B data used in this study. Perpendicular baseline indicates those with respect to the master image chosen as 14-Dec-2018 Image order

Acquired data (D/M/Y)

Perpendicular baseline (m)

Image order

Acquired data (D/M/Y)

Perpendicular baseline (m)

1

12-Jan-2018

64

13

13-Apr-2019

49

2

05-Feb-2018

52

14

19-May-2019

96

3

25-Mar-2018

−45

15

12-Jun-2019

7

4

30-Apr-2018

76

16

18-Jul-2019

34

5

05-Jun-2018

−64

17

23-Aug-2019

65

6

04-Aug-2018

38

18

28-Sep-2019

138

7

09-Sep-2018

42

19

22-Oct-2019

44

8

03-Oct-2018

90

20

15-Nov-2019

118

9

08-Nov-2018

26

21

21-Dec-2019

40

10

14-Dec-2018

Master

22

26-Jan-2020

107

11

07-Jan-2019

102

23

19-Feb-2020

73

12

08-Mar-2019

106

24

26-Mar-2020

82

3 Research Method Persistent scatterer InSAR uses multi-temporal SAR images, which relies on the DInSAR method. Assuming there is a point P located on the ground and there are two images acquired at different times (see Fig. 2), the phase difference between the two images can be extracted to determine the land deformation as: ϕint = ϕs − ϕM =

SP  − MP λ 4π

+ ϕscatt_s − ϕscatt_M

(1)

where ϕ int is the phase difference (i.e., the interferometric phase) between the phase of the master image (ϕM ) and that of the slave image (ϕS ), M and S are the satellite positions of the master and the slave images, respectively (see Fig. 2), MP and SP’ are the distances from the sensor to the monitoring position at the two acquisitions, respectively, λ is the radar wavelength, ϕscatt_M and ϕscatt_S are the phase change generated during the interaction between the radar wave and the target. The interferometric phase shown in Eq. (1) includes contributions associated with topography component, surface deformation, and various error and noise sources. Therefore, to determine the land deformation, it is necessary to remove the topography component from the interferometric phase, which leads to the DInSAR method [40]: ϕDint = ϕint −ϕTopo_simu = ϕDispl + ϕTopo_res + ϕAtm_s − ϕAtm_m + ϕOrb_s − ϕOrb_s + ϕnoise + 2.k.π

(2)

Mining-Induced Land Subsidence Detection by Persistent Scatterer

23

M S B

P Deformation Δr

P

Fig. 2. DInSAR deformation measurement

where ϕ DInt is the DInSAR signal, ϕTopo_simu is the topographic component, ϕDispl corresponds to surface deformation, ϕTopo_res is the residual topographic error (RTE) component caused by error in digital elevation model (DEM) used, ϕAtm_M and ϕAtm_S are the atmospheric delay while the radar signal propagates through the atmosphere at the master and slave acquisitions, respectively, ϕOrb_M and ϕOrb_S are the phase components due to the orbital errors of the two images, ϕnoise is the phase noise, k is an integer termed as phase ambiguity corresponding to the number of full wavelengths. The DInSAR technique aims at deriving surface deformation from DInSAR signal, which leads to the separation of surface deformation from other phase components in Eq. (2). An essential condition for doing so is to analyze pixels with small phase noise, which often involve objects that have strong and constant scattering characteristics over time (thus called permanent or persistent scatterers − PSs) or objects having constant scattering characteristics over time but come from small scattering objects (thus called distributed scatterers − DSs). The biggest limitation of the DInSAR method is the decrease in correlation as the time separation between master and slave acquisitions, i.e., the temporal baseline increases and the noise resulted from atmospheric delay. The PSI method represents an improved DInSAR method, which uses multiple SAR images acquired in the same area to generate an interferogram network with a single master image. This method provides a suitable data analysis and processing procedures to separate the phase component corresponding to surface displacement from other phase components shown in Eq. (2) as well as eliminate the influence of atmospheric phase screen (APS). A detail of DInSAR processing steps used in this study will be described in Sect. 4.

24

B. X. Nam et al.

4 Images Processing Data processing is implemented by the Sentinel Application Platform (SNAP) [42] and the Stanford Method for Persistent Scatterer (StaMPS) [10] software. SNAP is a common architecture for all Sentinel toolboxes, which is developed jointly by Brockmann Consult, SkyWatch and C-S [42]. StaMPS is a software built to process a series of SAR scenes applying the PS and SBAS methods. StaMPS was developed originally at Stanford University, U.S but subsequent versions were developed by research groups at the University of Leeds, UK, University Iceland and Delft University of Technology, Netherlands [43]. Data processing applying the PSI method by the two software consists of two independent tasks: (i) processing DInSAR for the master image and preparing slave images by SNAP, and (ii) persistent scatterer InSAR analysis by StaMPS. Fig. 3 depicts a persistent scatterer InSAR data processing flow chart using SNAP and StaMPS. Details on these steps will be described in the following sections. a. Preparation of the Master Image In the first step, the master image is selected from the dataset in such a way to reduce the overall temporal and perpendicular baselines (see Table 1). The image acquired on 14-Dec-2018 is selected as the master image with the interferogram network being shown in Fig. 4. This master image is imported into SNAP. The subswath 1 (IW1) that fully covers the study area is then selected (see Fig. 1 for the subswath that is selected in this study). And accurating for the orbit of the Sentinel-1 image by Graph Builder function in SNAP which ability to run the command automatically. These steps are important as it will help in optimizing the subsequent processing in terms of the time consumption and resources [44]. b. Preparation of Slave Images In this step, the remaining images other than the selected master image are sorted by their acquired dates, and also checks and reduces the name of the original image file. From this step, to enable processing in batch mode, a Graph Processing Tool (GPT) is used. The Python 2.7 code used to run the processing is available on Github [45]. c. Image Cropping and Orbit Update Slave images are separated based on the part and the polarization, which has been determined from the master image preparation step. Bursts between master and slaves will automatically be extracted relying on the defined area of interest (AOI) region. The orbital information of slave images are subsequently updated by automatically downloading precise orbital vectors at the ESA website [46]. d. Co-registration and Interferogram Formation This step conducts co-registration between the master image and each of the slave images in sequence, which requires a lot of calculations. The interferograms corresponding to each pair of master and slave images are then generated prior to removing the flat earth phase, i.e., the phase associated with the ellipsoid. Debursting for both the SLC images and the differential interferograms is applied to remove horizontal stripes. The topographical phase is then simulated using the 3-s Shuttle Radar Topography Mission (SRTM) Digital Terrain Model (DTM), which is downloaded automatically by SNAP. This topographical component is then removed from the interferograms. In this step, the research area that have selected before is subseted.

Mining-Induced Land Subsidence Detection by Persistent Scatterer

SNAP workflow

Sentinel-1 dataset

25

StaMPS workflow Estimation of coherence

Selection of the master image and subswaths

PS selection

Preparation of slave image files

Unwrapping

Slave image cropping to the selected master

Atmospheric correction and deformation

Coregistration and interferogram Fig. 3. Persistent scatterer InSAR data processing flow chart using SNAP and StaMPS.

Fig. 4. Interferogram network with the image acquired on 14-Dec-2018 chosen as the master image indicated by the red square.

26

B. X. Nam et al.

Finally, folders will be created to save resulted interferograms. They consist of folders used to store single images of SLC views for all SAR scenes, folders in which interferograms are saved, and a folder storing the coordinates of the master image and the cropped DEM of the study area. e. Estimation of Coherence Before importing data into STaMPS, data processed from SNAP will be prepared. An important factor to include is the value for selecting the Candidate PS pixels from the interferograms [43]. The criteria for the selection of the Candidate PS pixels in time series is called amplitude dispersion (DA ) (Eq. 3) [22]. In our region, the DA value selection was 0.4. σA DA = (3) mA where σA is the standard deviation of backscatter intensity and mA is the mean of the backscatter intensity. After converting the data into the format of MATLAB for PS processing, the next step is to perform the coherence calculation which means calculating the spatial correlation for each candidate PS pixel. Coherence is estimated by cross correlation operation between the complex images and is performed over a local window surrounding each pixel. The complex coherence (complex correlation coefficient) between two complex SAR images u1 and u2 is defined as [47, 48].  u1 u2 γ =  (4)  | u 1 |2 | u2 |2 where u1 and u2 are corresponding complex values from two images. f. PSs Selection The selection of PSs based on the probability, by comparing the results for data with random phase. Then pixels selected are weeded and dropping those that are due to signal contribution from neighbouring ground resolution elements and those deemed too noisy [43]. g. Phase Unwrapping The interferograms processed previously by SNAP are wrapped, indicating that their values vary from −π to π ; so, they need to be unwrapped by the so-called phase unwrapping, which is the most difficult step. In StaMPS, phase unwrapping can be implemented either by the 2D or 3D Minimum Cost flow (MCF) method [43]. The 3D MCF method is chosen in this study because it has been proven to be highly accurate [43]. The unwrapped interferograms are shown in Fig. 5. h. Atmospheric Correction and Deformation Estimation The heterogeneity of the atmosphere (ionosphere and troposphere) and its variation over time and space cause the changes in signal’s speed in the direction from the antenna and the target on the Earth’s surface. This results in the phase delay termed as Atmospheric Phase Screen (APS) [43]. In this step, APS needs to be removed first, then the deformation rates are calculated from deformation time series as following: Suppose subsidence of an interest pixel is defined by d = [d1 , d2 , . . . , dn ], n being the number of the

Mining-Induced Land Subsidence Detection by Persistent Scatterer

Fig. 5. Unwrapped interferograms

27

28

B. X. Nam et al.

Fig. 5. (continued)

image acquisitions with their corresponding acquired times being T = [T1 , T2 , . . . , Tn ]. The deformation rate is estimated relying on weighted least squares with the weights defined by mean square errors of the interferometric phase. The following equation is used to estimate the subsidence rate [49]: ν = (T T PT)−1 T T Pd

(5)

where υ is the subsidence rate, P is the vector of weights of M interferograms, which are defined by: P = diag (σ1 , σ1 , .. , σm )

(6)

where σi is the mean square error of the i th interferogram. P = diag (σ 1 , s2 ,...sM ) sk

5 Results and Discussion 5.1 Deformation Rate Map The deformation rates estimated by Eq. (5) from Sentinel-1 data are displayed by the ArcGIS pro 2.5 software (Link: https://esrivn.com/arcgis-pro/) and shown in Fig. 6

Mining-Induced Land Subsidence Detection by Persistent Scatterer

29

many subsidence points have been detected with the average annual rate being less than 45 mm/year. The number of points with high rate of subsidence are about 40 mm/year while the majority of subsidence is smaller than 30 mm/year. However, the sites around the mine are located on roads or houses that are our interest because these sites will directly affect people’s lives. 5.2 Result Validation In order to validate land subsidence determined from Sentinel-1 SAR data, 20 GNSS points located throughout the entire area at locations that can represent the deformations of the Phugiao quarries are utilized. Because of its large area, the measurement method chosen was the GNSS high accuracy. The monitoring stations are connected with a 1st -grade national elevation benchmark. Before each monitoring cycle, this benchmark is measured using static mothod relying on a network connecting with two national elevation benchmarks to assess its stability in the vertical. The results indicate that it was not displaced in both measuring cycles in Feb 15–17, 2018 and March 2–4, 2020. The static method and five CHC X91B receivers [50] are then chosen to conduct the GNSS measurement. Duration of each measument is a 180-min. The antenna heights with respect to the station marks are read five times by stainless steel tapes. Post-processed data processing is implemented by the Trimble Business Center v3.5 (TBC 3.5) software. The root mean square error (RMSE) of the weakest point in the network are 5 mm and 4 mm in 2018 and 2020 measuring cycles, respectively. The InSAR and GNSS subsidence results are then compared. GNSS measurement data at the monitoring points in January 2018 and December 2019 were used to calculate the average subsidence of 1 years. These are considered standard values for comparison with the settlement results from the images of period 1 / 2018–3 / 2020. The subsidence values and the deviations between the two methods are calculated according to Eq. (7) and are presented in Table 2.  = (1)/2 − (2)(mm)

(7)

where: 1 – Subsidence measurement by GNSS (mm); 2 - Subsidence computation from Sentinel-1 images (mm/y). 5.3 Discussions a. In Mining Areas – The subsidence points from Sentinel-1 images are quite concentrated in the mine area. However, because it is an open pit mine, the surface of the mine topography is constantly changing with the exploitation process, so that the mining pits have not been surveyed in the field by GNSS.

30

B. X. Nam et al.

Fig. 6. Deformation rate map of the Phugiao 1 (a) and Phugiao 2 (b) quarries. The circles indicate coherent scatters and the triangles refer to GNSS stations.

Mining-Induced Land Subsidence Detection by Persistent Scatterer

31

Table 2. Subsidence measurement values by GNSS and Sentinel-1 images (Phugiao1) Number Station name GNSS subsidence 1 InSAR subsidence 2 Deviation  (mm/y) (mm) (mm/y) 1

A10

−15.8

−11.6

3.7

2

A11

−26.2

−12.2

−0.9

3

A13

−30.5

−19.5

4.25

4

A14

−35.2

−21.1

3.5

5

A16

−35.1

−18.8

1.25

6

A43

−14.2

−4.7

7

A45

−15.6

−10.3

2.5

8

A73

−66.3

−35.1

1.95

9

A6

−27.2

−13

10

A32

−49.2

−25.9

1.3

11

A69

−30.3

−19.1

3.95

12

A94

−64.7

−35.5

3.15

13

A41

−48.6

−25.2

0.9

−2.4

−0.6

– Although there is not any GNSS measurement data to validate subsidence at the mine pits, from the comparison results in Table 2 which is the subsidence survey sites by GNSS and the PS points located near the quarry or on roads close to the open pits. It can be seen that the subsidence values from the images are quite similar to the result from the GNSS measurement, with accuracy ± 2.6 mm at Phugiao1 and ± 3.7 mm at Phugiao2. Therefore, the results of subsidence at the open pit sites are considered accurate. With the determined values of subsidence in the mine area, we can note some locations with large values of subsidence, such as at locations A, B (Phugiao1) (Fig. 7) and C, D (Phugiao2) (Fig. 8) which are concentrated on the pits and the transport routes. – For A position, the maximum value of subsidence is about −40 mm/year with red PS points, orange PS points with subsidence values less than −20 mm/year. At B position, the number of distribution subsidence points are more than the A position but the subsidence values are not as high as those in A position (see the chart in Fig. 6). Subsidence values are mostly less than −30 mm/year. At C position, the average of subsidence is about −30mm / year, there are some PS points with subsidence close to −50mm/year but almost the PS points are from −10mm/year to −30mm/year. At D position, the majority of subsidence ranges from −10 mm/year to −25 mm/year. In this site there are some positive points also. The reason for the occurrence of PS points with both positive and negative values in the mining pits is partly due to the drilling and blasting operations and the work of heavy trucks are also a cause of the positive deformation.

32

B. X. Nam et al. Table 3. Subsidence measurement values by GNSS and Sentinel-1 images (Phugiao2)

Number

Station name

GNSS subsidence 1 (mm)

InSAR Subsidence 2 (mm/y)

Deviation  (mm/y)

1

A12

−58.2

−31.3

2.2

2

A13

−70.2

−33.6

−1.5

3

A11

−32.1

−17.4

1.35

4

A50

−37.5

−20.5

1.75

5

A49

−29.2

−16.3

1.7

6

A74

−24.3

−10.3

−1.85

7

A20

−62.3

−33.1

1.95

8

A80

−48.1

−21.9

−2.15

9

A34

−53.2

−33.7

7.1

10

A36

−50.6

−27.1

1.8

11

A60

−8.1

−7.4

3.35

12

A59

−18.5

−5.3

−3.95

13

A19

−40.5

−22.6

2.35

14

A17

−30.8

−20.1

4.7

15

A58

−58.8

−20.4

−9

16

A73

−34.6

−15.5

−1.8

b. In Locations Adjacent to Open-Pit Mines At the positions located on the roads or residential areas surrounding open-pit mines, we conducted assessment 13 points at Phugiao 1 and 16 points at Phugiao2 respectively which concentrated locations of PS points with the distribution near or outside within a radius of about 20 m. These points are listed in Table 2 and Table 3 above. The reason for the selection of points within this 20 m radius is due to the scattering properties of the radar image when the Radar wave reaches the surface of the object, the reflected scattering rays may not return immediately but it may be corner backscattering or volume scattering before returning, thus the checkpoints shown on the image may be slightly shifted from its actual position. Based on Table 2, the GNSS measurements have 2 years interval while the measurements from 24 Sentinel-1 image processing are averaged over the period from January 2018 to March 2020. So the values of subsidence measurements using GNSS will have to be divided by an average of 2 years. The correlation of these 20 points was evaluated for the purpose of a preliminary assessment of how the subsidence points measured by the PSI method achieved compared to the high precision GNSS measurement method. Below is the distribution of ground settlement values of the 13 locations shown in Table 2 and 16 locations in Table 3.

Mining-Induced Land Subsidence Detection by Persistent Scatterer

33

A Position 10 0 Subsidence value (mm/y)

B

A

-10 -20 -30 -40 -50 0

5

10

15

B position 10 5 Subsidence value (mm/y)

0 -5 -10 -15 -20 -25 -30 -35

0

10

20

30

40

50

60

70

80

Fig. 7. Subsidence in A, B (a) and C, D (b) with the charts of subsidence values

Based on the chart in Fig. 9, it is to recognize that the settlement values made from the PSI method tend to be higher than the settlement values measured from GNSS high accuracy method, however, the correlation is quite good with R2 reached 0.93 and 0.82 at two sites. This can also be explained easily because the number of survey points is not much and in addition, the GNSS and PS points from the image are not completely overlapping. Although the number of checkpoints is not much (31 points), the measured values also reflect the subsidence situation around the Phugiao open-pit mines in the period from January 2018 to March 2020.

34

B. X. Nam et al.

C Position (Phugiao2) 10 Subsidence value (mm/y)

C

D

0 -10 -20 -30 -40 -50 0

20

40

60

80

D Position (Phugiao2)

10

Subsidence value (mm/y)

5 0 -5 -10 -15 -20 -25

0

10

20

30

40

50

60

70

Fig. 8. Subsidence in A, B (a) and C, D (b) with the charts of subsidence values

80

Mining-Induced Land Subsidence Detection by Persistent Scatterer

35

Fig. 9. The chart of the correlation of subsidence results made by Sentinel-1 radar and GNSS measurement at Phugiao1 (a) and Phugiao2 (b)

6 Conclusions With a data set of 24 Sentinel-1B images collected between January 2018 and March 2020, the PSI method was applied with the combination of ESA SNAP and StaMPS software. The results of determining the land subsidence of Phugiao open pit and surround area indicate the following: The PSI method is suitable for mining areas with few trees and as many constructions as this study area. Sentinel-1 image with a resolution of 3.5 m in range and 22 m in azimuth, large cover and high frequency of image acquisitions (12 days), is very appropriate for studies related to land subsidence. This is the first time, the issue of land subsidence in mining areas in Vietnam has been investigated by using a combination of radar intermerometric method and GNSS method.

36

B. X. Nam et al.

With a large number of images, the rate of land subsidence was evaluated within a period of approximately 2 years and 3 months. In the quarry area there were many subsidence PS points, but these subsidence points were mostly located on the road of pits or the slope of the mining pits. The biggest land subsidence rate at mines was smaller than −45mm/year. In the area around the quarry such as roads, edge of pits or houses near the mines have been surveyed by GNSS. Persistent scatterer InSAR subsidence points were compared to GNSS measurement points from January 2018 to December 2019. However, the number of GNSS measurement points and PS points from radar images did not completely coincide, we used 31 locations around the mine area where PS subsidence points coincided with those measured by GNSS in radius of 20 m. Correlation of these two data types reached 0.93 and 0.82 at Phugiao1 and Phugiao2 respectively, proving the ability to determine land subsidence by the Sentinel-1 image sequence. Acknowledgments. This work is supported financially by Research Institute for Mining ElectroMechanics of Hanoi University of Mining and Geology. The authors also thank Alaska Satellite Facilities (ASF) for providing Sentinel-1 InSAR data and colleagues from Department of Mine Surveying for providing the GNSS data.

Conflicts of Interest. The authors declare no conflict of interest.

References 1. Bawden, G.W., Johnson, M.R., Kasmarek, M.C., Brandt, J.T., Middleton, C.S.: Investigation of land subsidence in the Houston-Galveston region of Texas by using the global positioning system and interferometric synthetic aperture radar, 1993–2000. US Geological Survey (2012) 2. Tizzani, P., Berardino, P., Casu, F., Euillades, P., Manzo, M., Ricciardi, G., Zeni, G., Lanari, R.: Surface deformation of Long Valley caldera and Mono Basin, California, investigated with the SBAS-InSAR approach. Remote Sens. Environ. 108, 277–289 (2007) 3. Aobpaet, A., Cuenca, M.C., Hooper, A., Trisirisatayawong, I.: InSAR time-series analysis of land subsidence in Bangkok, Thailand. Int. J. Remote Sens. 34, 2969–2982 (2013) 4. Ng, A.H.M., Ge, L., Li, X., Abidin, H.Z., Andreas, H., Zhang, K.: Mapping land subsidence in Jakarta, Indonesia using persistent scatterer interferometry (PSI) technique with ALOS PALSAR. Int. J. Appl. Earth Obs. Geoinf. 18, 232–242 (2012) 5. Yang, K., Yan, L., Huang, G., Chen, C., Wu, Z.: Monitoring building deformation with InSAR: experiments and validation. Sensors 16, 2182 (2016) 6. Akcin, H., Kutoglu, H., Deguchi, T., Koksal, E.: Monitoring subsidence effects in the urban area of Zonguldak Hardcoal Basin of Turkey by InSAR-GIS integration. Nat. Hazards Earth Syst. Sci. 10, 1807–1814 (2010) 7. Gao, J.X., Chao, L., Jian, W., Li, Z.K., Meng, X.C.: A new method for mining deformation monitoring with GPS-RTK. Trans. Nonferrous Met. Soc. China 21, 659–664 (2011) 8. Lian, X., Hu, H.: Terrestrial laser scanning monitoring and spatial analysis of ground disaster in Gaoyang coal mine in Shanxi, China: a technical note. Environ. Earth Sci. 76, 287 (2017) 9. Tiwari, A., Narayan, A.B., Dwivedi, R., Swadeshi, A., Pasari, S., Dikshit, O.: Geodetic investigation of landslides and land subsidence: case study of the Bhurkunda coal mines and the Sirobagarh landslide. Surv. Rev. 52, 134–149 (2020) 10. Hooper, A., Bekaert, D.P.S., Spaans, K., Arıkan, M.: Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics 514–517, 1–13 (2012)

Mining-Induced Land Subsidence Detection by Persistent Scatterer

37

11. Graham, L.C.: Synthetic interferometer radar for topographic mapping. Proc. IEEE 62, 763– 768 (1974) 12. Tran, A.V., Masumoto, S., Raghavan, V., Shiono, K.: Spatial distribution of subsidence in Hanoi detected by JERS-1 SAR interferometry. Jpn. Soc. Geoinf. 18, 3–13 (2007) 13. Hooper, A.: A multi-temporal InSAR method incorporating both persistent scatterer and small baseline approaches. Geophys. Res. Lett. 35, L16302 (2008) 14. Shanker, P., Casu, F., Zebker, H.A., Lanari, R.: Comparison of persistent scatterers and small baseline time-series InSAR results: a case study of the San Francisco bay area. IEEE Geosci. Remote Sens. Lett. 8, 592–596 (2011) 15. Berardino, P., Fornaro, G., Lanari, R., Sansosti, E.: A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Trans. Geosci. Remote Sens. 40, 2375–2383 (2002) 16. Cavalié, O., Doin, M.P., Lasserre, C., Briole, P.: Ground motion measurement in the Lake Mead area, Nevada, by differential synthetic aperture radar interferometry time series analysis: probing the lithosphere rheological structure. J. Geophys. Res. Solid Earth 112, 1–18 (2007) 17. Hetland, E.A., Musé, P., Simons, M., Lin, Y.N., Agram, P.S., DiCaprio, C.J.: Multiscale InSAR time series (MInTS) analysis of surface deformation. J. Geophys. Res. Solid Earth 117, B02404 (2012) 18. López-Quiroz, P., Doin, M.P., Tupin, F., Briole, P., Nicolas, J.M.: Time series analysis of Mexico City subsidence constrained by radar interferometry. J. Appl. Geophys. 69, 1–15 (2009) 19. Lundgren, P., Usai, S., Sansosti, E., Lanari, R., Tesauro, M., Fornaro, G., Berardino, P.: Modeling surface deformation observed with synthetic aperture radar interferometry at Campi Flegrei caldera. J. Geophys. Res. Solid Earth 106, 19355–19366 (2001) 20. Schmidt, D.A., Bürgmann, R.: Time-dependent land uplift and subsidence in the Santa Clara valley, California, from a large interferometric synthetic aperture radar data set. J. Geophy. Res. Solid Earth 108 (2003) 21. Usai, S.: A least squares database approach for SAR interferometric data. IEEE Trans. Geosci. Remote Sens. 41, 753–760 (2003) 22. Ferretti, A., Prati, C., Rocca, F.: Permanent scatterers in SAR interferometry. IEEE Trans. Geosci. Remote Sens. 39, 8–20 (2001) 23. Hooper, A.: Persistent Scatter Radar Interferometry for Crustal Deformation Studies and Modeling of Volcanic Deformation. Proquest Dissertations Publishing, Ann Arbor (2006) 24. Hooper, A., Segall, P., Zebker, H.A.: Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcán Alcedo, Galápagos. J. Geophys. Res. Solid Earth 112 (2007) 25. Bui, L.K., Featherstone, W.E., Filmer, M.S.: Disruptive influences of residual noise, network configuration and data gaps on InSAR-derived land motion rates using the SBAS technique. Remote Sens. Environ. 247, 111941 (2020) 26. Liu, G., Luo, X., Chen, Q., Huang, D., Ding, X.: Detecting land subsidence in Shanghai by PS-networking SAR interferometry. Sensors 8, 4725–4741 (2008) 27. Hirose, K., Maruyama, Y., Murdohardono, D., Effendi, A., Abidin, H.Z.: Land subsidence detection using JERS-1 SAR Interferometry. In: 22nd Asian Conference on Remote Sensing, pp. 5–9. (Year) 28. Yang, Z., Li, Z., Zhu, J., Feng, G., Wang, Q., Hu, J., Wang, C.: Deriving time-series threedimensional displacements of mining areas from a single-geometry InSAR dataset. J. Geodesy 92, 529–544 (2018) 29. Gee, D., Bateson, L., Sowter, A., Grebby, S., Novellino, A., Cigna, F., Marsh, S., Banton, C., Wyatt, L.: Ground motion in areas of abandoned mining: application of the intermittent SBAS (ISBAS) to the Northumberland and Durham Coalfield, UK. Geosciences 7, 85 (2017)

38

B. X. Nam et al.

30. Zheng, M., Deng, K., Du, S., Liu, J., Liu, J., Feng, J.: Joint probability integral method and TCPInSAR for monitoring mining time-series deformation. J. Indian Soc. Remote Sens. 47, 63–75 (2019) 31. Zhu, Y., Ding, X., Li, Z., Luo, Y.: Monitoring of surface subsidence of the mining area based on SBAS. JCP 9, 1177–1184 (2014) 32. Ng, A.H.M., Ge, L., Du, Z., Wang, S., Ma, C.: Satellite radar interferometry for monitoring subsidence induced by longwall mining activity using Radarsat-2, Sentinel-1 and ALOS-2 data. Int. J. Appl. Earth Obs. Geoinf. 61, 92–103 (2017) 33. Baek, J., Kim, S.W., Park, H.J., Jung, H.S., Kim, K.D., Kim, J.W.: Analysis of ground subsidence in coal mining area using SAR interferometry. Geosci. J. 12, 277–284 (2008) 34. He, J., Liu, G., Yue, H.: Monitoring ground subsidence in mining area using spaceborne InSAR technology. In: 2009 Joint Urban Remote Sensing Event, pp. 1–6. IEEE (2009) 35. Diao, X., Wu, K., Hu, D., Li, L., Zhou, D.: Combining differential SAR interferometry and the probability integral method for three-dimensional deformation monitoring of mining areas. Int. J. Remote Sens. 37, 5196–5212 (2016) 36. Ma, C., Cheng, X., Yang, Y., Zhang, X., Guo, Z., Zou, Y.: Investigation on mining subsidence based on multi-temporal InSAR and time-series analysis of the small baseline subset—case study of working faces 22201-1/2 in Bu’ertai mine, Shendong coalfield, China. Remote Sens. 8, 951 (2016) 37. Tran, A.V., Tran, C.Q., Nguyen, A.D., Ho, D.M.T., Tran, A.T., Nguyen, H.N., Luong, L.T.T.: Application of PSInSAR method for determining of land subsidence in Hanoi city by CosmoSkymed imagery. In: Proceedings of GIS IDEAS (2016) 38. Le, T.V., Nguyen, C.T.: Assessment of capacity of using sentinel-1 images in monitoring land subsidence in Ho Chi Minh City. Sci. Technol. Dev. J.-Sci. Earth Environ. 2, 19–25 (2019) 39. https://btv.org.vn/tin-tuc/khai-thac-khoang-san-di-doi-voi-bao-ve-moi-truong-23304.html 40. Gabriel, A.K., Goldstein, R.M., Zebker, H.A.: Mapping small elevation changes over large areas: Differential radar interferometry. J. Geophys. Res. Solid Earth 94, 9183–9191 (1989) 41. https://search.asf.alaska.edu 42. https://step.esa.int/main/toolboxes/snap/ 43. Hooper, A., Spaans, K., Bekaert, D., Cuenca, M.C., Arıkan, M., Oyen, A.: StaMPS/MTI manual. Delft Inst. Earth Obs. Space Syst. Delft Univ. Technol. Kluyverweg 1, 2629 (2010) 44. Delgado Blasco, J.M., Foumelis, M., Stewart, C., Hooper, A.: Measuring urban subsidence in the Rome metropolitan area (Italy) with sentinel-1 SNAP-StaMPS persistent scatterer interferometry. Remote Sens. 11, 129 (2019) 45. Foumelis, M., Blasco, J.M.D., Desnos, Y.-L., Engdahl, M., Fernández, D., Veci, L., Lu, J., Wong, C.: ESA SNAP-StaMPS integrated processing for sentinel-1 persistent scatterer interferometry. In: IGARSS 2018–2018 IEEE International Geoscience and Remote Sensing Symposium, pp. 1364–1367. IEEE (2018) 46. https://qc.sentinel1.eo.esa.int/ 47. Lee, J.S., Hoppel, K.W., Mango, S.A., Miller, A.R.: Intensity and phase statistics of multilook polarimetric and interferometric SAR imagery. IEEE Trans. Geosci. Remote Sens. 32, 1017– 1028 (1994) 48. Bamler, R., Just, D.: Phase statistics and decorrelation in SAR interferograms. In: Proceedings of IGARSS’93-IEEE International Geoscience and Remote Sensing Symposium, pp. 980– 984. IEEE (1993) 49. Jiang, Y., Liao, M., Wang, H., Zhang, L., Balz, T.: Deformation monitoring and analysis of the geological environment of Pudong international airport with persistent scatterer SAR interferometry. Remote Sens. 8, 1–18 (2016) 50. https://nguyenkimjsc.vn/may-gps-rtk-chc-x91b.html

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam) for Geohazard Assessment and Sustainable Development Viet-Ha Nhu(B) Hanoi University of Mining and Geology, Hanoi, Vietnam [email protected]

Abstract. Geotechnical characteristics in three-dimensional form with the full thickness of each layer are an essential tool for analyzing and visualizing the underground conditions, which are useful for civil engineering projects, geologic hazard assessment, and sustainable development. The objective of this study is to propose and verify a procedure for building a high-resolution 3D geotechnical model in Hanoi city (Vietnam). For this regard, a total of 1,386 borehole logs with attributes of 10,278 soil samples and 16,626 in-situ tests during the last 20 years in the city were collected and processed. These data were interpreted and analyzed in a systematic way to integrate into the 3D model. Then, a high-resolution 3D geotechnical model was constructed and visualized the real geotechnical system with 21 Quaternary geotechnical layers, which integrated the attribute of 19 soil parameters and tectonic activities. It is also feasible as a powerful tool for the reproduction and analysis when it allows extracting spatial distribution of any layer (or point, column) in elevation (or depth, volumetric) and any 2D geotechnical map at different elevation or depth. Finally, the results demonstrated the feasibility of the high-resolution 3D geotechnical model of Hanoi city, which provides valuable information for subsurface and geohazard assessment as well as sustainable development study mitigation. Keywords: 3D geotechnical model · Urban planning · Urban geohazard · Hanoi · Vietnam

1 Introduction Geotechnical conditions and properties play an essential role in dealing with the complicated nature of subsurface as urban development and urban geohazard mitigation. Within the last decades, it has been receiving the attention of many researchers through building high-resolution three-dimensional (3D) geotechnical models [1–3]. Within the framework of urban development, geotechnical characterizations of the foundation are essential information for the urban planning and geohazard assessments of cities. Herein, geotechnical parameters are highly necessary to determine underground spaces accurately for constructions, which helps to establish a safe urban extension and sustainable planning [4, 5]. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 39–57, 2021. https://doi.org/10.1007/978-3-030-60269-7_3

40

V.-H. Nhu

Literature review shows that urban geohazards involving ground instability can be costly, dangerous, and affect many people. Some urban geohazards such as land subsidence and liquefaction have become a global problem that many parts of the world have to face. They cause various damages to roads, bridges, buildings, underground infrastructures, and may alter groundwater flows and rivers [6]. For this regard, information on the high-resolution geological-geotechnical maps plays critical roles in geohazard predictions and mitigations. The geotechnical characterization is typically investigated, stored, and represented in (one-dimensional - 1D) borehole logs, cross-sections, or (two-dimensional - 2D) ground surface maps associating with soil properties. Therefore, several geotechnical characteristics are still not so well estimated in these traditional representations, especially for complicated problems as land subsidence and liquefaction [7–10]. From the development of information technology in recent years, high-resolution 3D geotechnical models have been strengthened research and development. Initially, they were developed as 3D geo-models that mainly used to solve problems in hydrological geology, resource protection, or in the oil and gas industry to simulate and predict the distribution of oil resources, then expanded to serve for other fields such as construction engineering, geotechnics, numerical analysis. Up to now, most of the efforts in the world have focused on the accuracy and reliability improvements of 3D geo-models, including the development of 3D geotechnical models with high-resolution. Many optimization analysis algorithms have been developed and applied over the years. These are essential tools for increasing the quality of modeling of surface structures in underground space. The high-resolution 3D geotechnical models have the potential to apply quite diversified, but the actual number of development models is still limited. The main reason is that they need a large and standardized database. The slow development of data digitization, modeling, and high computerization applications in the field of geosynthetics surveys is a significant obstacle to the cost of modeling [3, 11–13]. The objective of this study is to propose and verify a procedure for production a high-resolution 3D geotechnical model of Hanoi city (Vietnam) using the variogram and kriging interpolation techniques. Then, the model was established using 10,278 soil samples and 16,626 in-situ tests from 1,386 borehole logs in the city during the last 20 years. Finally, discussions are given.

2 Study Area The study area is the Hanoi city (Fig. 1), which located between longitudes 105°45 E and 105°55 E, and latitudes 20°56 N and 21°06 N. It covers an area of about 161 km2 that consists of 13 districts comprising urban and suburban areas. The study area is bounded by the Red River in the northeast, and two smaller rivers (the Nhue and Tolich Rivers) border the study area in the west and the south. The altitude ranges from 2.2 m to 12.8 m a.s.l where the high-altitude areas are the dykes along the Red River (Fig. 1). The Red River as the main river which provides water for the study area. Its width varies from 480 m to 1,440 m and its riverbed elevation changes from −5.6 m to −8.3 m a.s.l [14]. Geologically, the Hanoi city is situated in the upper part of the Red River delta, which was formed by the sinking of a former mountain region and was filled by alluvial

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

41

Fig. 1. Location of the study area

deposits of the Red River. The area has undergone periods of active tectonics during the Neogene and the Quaternary. The Neogene sediments are characterized by a thickness of more than 250 m and are mainly composed of sand and gravel. The Quaternary sediments comprise the Pleistocene and the Holocene sediments with a total thickness of about 80–100 m. The Pleistocene sediments consist of gravel, sand, and clay. The upper contact with the Holocene formation is conformable while the lower contact of the formation with Neogene formation is angularly unconformable. The Holocene sediments covering the Hanoi city were deposited as recent alluvium. The composition consists of fine-grained sediments such as sand, silt, and clay, in which there are sporadic occurrences of marine fossils. The formation is a mix of fluvial, lacustrine, and marshy sediments with the dominance of an alluvial origin. On the top are the youngest sediments distributed in the interior side of the Red River dikes and their affluent that are deposited as riverbed-riverbank sediments and anthropogenic embankments. Along the large rivers, it is composed mainly of brownish clayey silt, clay, and sand. While in their affluent, these deposits are yellowish-grey pebble, gravel, and sand mixed with brownish silt and clay (Fig. 1) [15–19]. The Quaternary sediments are a multi-aquifer system consisting of two aquifers of the Holocene and Pleistocene and separated by confining interlayers [20–24]. These consist of six geological formations from old to young: Lechi, Hanoi, Vinh Phuc, Hai Hung, Thai Binh, and anthropogenic deposits. Each geological formation was separated into geological beds in which fining-upward cycles could be recognized in the particle size distribution. A total of 18 geological beds with lithological attributes were delineated in these geological formations (Fig. 2).

42

V.-H. Nhu

Fig. 2. The lithostratigraphic table of Hanoi city [19]

The faults system in the Hanoi city was mainly tectonically active before the Quaternary but is still having an impact on the Quaternary sediments. Some evidence of faulting

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

43

activity has been observed on the ground surface, but interruptions of the deposits are seldom observed in the borehole logs [10, 25]. Based on the geological maps, fault systems were identified comprising the main fault of Vinhninh with the NW-SE trend and the two smaller faults of Nghiado-Tulien and Thanhxuan-Gialam with a NE-SW trend. All of them are normal faults. The Vinhninh fault goes through the center of the Hanoi city and has a length of 18–20 km, displacement of 2–3 km, and a brecciated fault zone of 400–500 m. The fault has been tectonically active since the Paleozoic-Mesozoic and played an important role in the Cenozoic. The Thanhxuan-Gialam fault is more than 15 km long and has a brecciated fault zone of 200–300 m. The Nghia Do-Tu Lien fault is 20–25 km long, has a 600–700 m wide brecciated fault zone, and lies in the north of the Hanoi city (Fig. 6). The Hanoi area is situated in the monsoonal region, with hot, rainy and dry seasons. During the last decades, the coldest month is January with an average temperature of 16.8 °C, whereas the hottest month is July with an average temperature of 29.3 °C. The maximum evaporation occurs in June with an average value of 99.1 mm, while the minimum occurs in February with an average value of 52.5 mm. The rainy season is generally from May until October and is characterized by high temperatures and humidity. Rainfall in the rainy season occurs with high frequency and intensity and accounts for 80 to 90% of the annual rainfall [26].

3 Data and Materials Data for high-resolution 3D geotechnical modeling focus on the topographic surface, geological settings, borehole logs, and property attributes [2]. Among them, the most important is probably the borehole logs, which are generally obtained from directly subsurface observation. They supply all information about coordinates, address, starting elevation, down-hole depth, and lithological attributes. The lithological attributes comprise material descriptions and properties from in-situ and laboratory tests. The topographic surface data are the upper limit of the models and are critical for modeling if ground elevations are missing in the borehole logs. The geological settings are essential data to identify sedimentary deposits, geotechnical layers, and tectonic activities. Based on the availability and knowledge of the Hanoi city, the data was used including (i) national topographic maps at a scale of 1:2,000 [27]; (ii) Hanoi maps of Quaternary sediments, geology, and lithology at a scale of 1:50,000 [16, 19, 28]; (iii) 1,386 borehole logs with attributes of 10,278 soil samples; and (iv) 16,626 in-situ tests. These borehole logs were coming from 348 geotechnical investigation projects during the last 20 years at Hanoi city.

4 Methodology All available traditional geotechnical collected data was and analyzed in a systematic way by an elaborate data analysis to build a high-resolution 3D geotechnical model off Hanoi city, that included three main components as ground surface, subsurface, and attributes.

44

V.-H. Nhu

For the ground surface, a Digital Elevation Model (DEM) is considered as the most useful for representation. In addition, high-resolution DEMs are always critical to modeling when ground elevations are missing in the borehole logs. Thus, the high-resolution DEM of the 3D geotechnical model was generated from the national topographic maps on the scale of 1:2000 [27]. The raster elevation contours must first be converted to vectors to be “tagged” with their corresponding elevation values. These contour lines were used as input to build a TIN and then rasterized to a DEM, which can be accomplished by GIS tools. Subsurface models are typically constructed based on a number of measured points obtained from subsurface depths. When the number of measured points is large enough and regularly distributed, high-resolution subsurface models can be constructed by directly connecting the observations [29–31]. If there are few measured points or if they are scattered, realistic subsurface models can only be constructed by interpolation. The subsurface was constructed by spatial analysis supplemented by a cross-validation step for every geotechnical layer (GL). The spatial database was done using Microsoft Excel 365, spatial analysis and interpolation were done using ArcGIS 10.3. The spatial datasets for subsurface were extracted from measured points as a top elevation of the different GLs in borehole logs. Before that, this data was cleaned, inspected of consistency. Both consistency of the coordinates and the starting elevations of the borehole logs needed to be verified to ensure the quality of data for geospatial analysis. Every GL in every drilling record was then interpreted. A GL was defined as a group of observed lithologies that satisfy stratigraphic conditions [32, 33], including (i) same geological formation or bed, (ii) similar composition, (ii) same compactness, or consistency, and (iv) same stratigraphic order. Considering the stratigraphic conditions, the lithology of borehole logs and the stratigraphy of the Quaternary sediments were matched. Along with each drilling record, every GL thickness was also determined based on its top and base elevation, the top was defined as the base of the layer above. These values were derived from the ground elevation minus the depths of down-holes. Missing GLs between interpreted ones were assigned a zero thickness so that the top and base elevation have the same value. Similarly, all in-situ tests and soil samples within each GL were used to construct spatial datasets of soil properties. They were also used as supplementary information for GL interpretation. The spatial datasets of every GL are filtered to serve for subsurface interpolations using geostatistical analysis which supplemented by a cross-validation step [34, 35]. The geostatistical analysis used the stochastic and structural spatial variability of natural phenomena and was based on the theory of spatial random variables. The spatial distribution of the attributes of the spatial datasets was to be analyzed by variogram analysis [36, 37]. This interpolation was predicted unknown values based on the statistical relationships among the measured points, including the fitted model from the variography analysis (Fig. 3). Prior to the geostatistical analysis, spatial trends were identified and removed prior to semivariogram analysis, which is performed on the residuals. For a stochastic variable V(z), if the cumulative distribution of V(zi ) is equal to the cumulative distribution of V(zi + b), the variable satisfies the strict stationarity hypothesis. Fz1 ,z2 ,......,zn (V1 , V2 , . . . . . . , Vn ) = Fz1 +b,z2 +b,......,zn +b (V1 , V2 , . . . . . . , Vn )

(1)

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

45

Fig. 3. The flowchart of the geostatistical analysis in this research

where z is the vector of spatial coordinates, V(z) is the variable under consideration as a function of spatial location and b is the lag vector representing a separation between two spatial locations. The spatial variable normally cannot comply with this very strict stationarity hypothesis. Thus, geostatistics was employed to the estimation of a spatial variable V(z) if the second-order stationarity hypothesis can be satisfied that consists of the following two conditions in the research area: (1) The mathematical expectation of V(z) exists as a constant as bellowing: E[V(z)] = E[V(z + b)] = m

(2)

(2) In the research area, the covariance of V(z) exists and depends only on the lag vector b, not on the location z, as belows: Cov{V(z), V(z + b)} = E[V(z)V(z + b)] − E[V(z)]E[V(z + b)] = E[V(z)V(z + b)] − m2 = C(b)

(3)

when b = 0 → C(0) = Cov{V(z), V(z)} = Var[V(z)]. where m is a constant value, E[V(z)] is the mathematical expectation of V(z), Cov{V(z), V(z + b)} is the covariance function of V(z) and V(z + b), C(b) is the covariance of distance b, Var[V(z)] is the variance of V(z). The Cov{V(z), V(z + b)} is assumed to exist in the second-order stationarity hypothesis. In fact, stochastic variables do not have a covariance, but their variogram can be found. Thus, an even weaker stationarity hypothesis called the intrinsic hypothesis can be satisfied if the following two conditions are met as bellowing:

46

V.-H. Nhu

(1) The mathematical expectation [V(z) − V(z + b)] is equal to zero as: E[V(z) − V(z + b)] = 0

(4)

(2) The variance of [V(z) − V(z + b)] exists and is stationarity: Var[V(z) − V(z + b)] = E[V(z) − V(z + b)]2

(5)

Semivariogram analysis is fitting a theoretical model through the experimental semivariogram function (Eq. 6), for whole pairs of locations separated by a distance b as: 1 γ(b) = Var[V(z) − V(z + b)] 2 1 = E[V(z) − V(z + b)]2 − {E[V(z) − V(z + b)]}2 2

(6)

Under the condition of the intrinsic hypothesis, E[V(z) − V(z + b)] = 0, the variogram is defined as: 1 γ(b) = E[V(z) − V(z + b)]2 2 N(b) 1  ≈ [V(zi ) − V(zi + b)]2 2N(b)

(7)

i=1

where N(b) is the number of data pairs separated by b.

Fig. 4. Relationship between covariance and semivariogram under condition of second-order stationarity

Under the second-order stationarity hypothesis γ(b) = Cov(0) − Cov(b), which is illustrated in Fig. 4, the semivariance γ(b) increases, and covariance Cov(b) decrease when the distance b increases. However, if b is larger than a value a, both the γ(b) and Cov(b) will level off. The variogram γ(b) reaches the maximal value of Cov(0)

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

47

Fig. 5. Experimental semivariogram and fitted theoretical model

and Cov(b) reaches zero. Here the value reflects the autocorrelation influence spectrum, which is called the range of the variogram. At separation distances between zero and the spatial variable is auto-correlated. The experimental semivariogram is only indirectly used in the geostatistical estimation procedure. It is only used for fitting by theoretical models because the experimental variograms cannot leak the semivariance values at a lag distances continuous series. Therefore, continuous theoretical variogram models are defined, which are described by their range, sill, and nugget (Fig. 5). The total sill (sill + nugget) is value of the semivariance when the semivariogram levels off. It corresponds to the maximum autocorrelation distance among the measured points. The nugget value reflects the variable changing at two very close points, that can sometimes be related to measurement errors or insufficiency of data pairs at the smallest lag distances. The range value as the distance at which the semivariogram value reaches the total sill value, where data pairs are still auto-correlated. Equations of common theoretical models of the Nugget, Spherical, Exponential, Gaussian, and Power are illustrated in Eq. 8.  0 if b = 0 + Nugget: γ(b) = ; c otherwise      3  if b ≤ a c · 1.5 ba − 0.5 ba + Spherical: γ(b) = c otherwise

−3b + Exponential: γ(b) = c · 1 − exp a

2 −3b + Gaussian: γ(b) = c · 1 − exp a2 w + Power: γ(b) = c · b 0 < w < 2, c is the sill value (8) Kriging interpolation weights the nearby known values to get a prediction for an unmeasured location (Eq. 9). There are several Kriging interpolation methods. However,

48

V.-H. Nhu

the Ordinary Kriging one is suitable for prediction less trending variable and stationarity hypothesis [38]. n λi · V (xi ) (9) V (x0 ) = (i=1)

where V(x0 ) is the predicted value, λi and V(xi ) are measurement and the weight of the nearby location. Minimal variance and no bias of prediction errors are two requirements for λi selection as: + Optimal: Var[V(xi ) − V(xi )] = min + Un - bias: E[V(xi ) − V(xi )] = 0

(10)

Cross-validation is performed as a statistical calculation of the predicted error diagnostics that indicate whether the adopted theoretical models are reasonable for prediction [39]. Cross-validation removes one point at a time from the dataset of sample, and then the other ones are used to predict the removed sample via theoretical models. Every sample will have both actual and predicted values for every sample point, and the error is defined as the difference between them when through doing the same procedure piecewise as above. Finally, the interpolated results then assessed based on statistical evaluation metrics with Correlation Coefficient (R), Mean Error (ME) and Root Mean Square Error (RMSE) as shown in Eq. 11, 12 and 13, respectively. The best-interpolated model will be the one for which R-value is close to one, ME value is close to zero, and RMSE value is minimal. The ME value is near zero when the prediction errors are unbiased. The model result is underestimated if ME value is positive, whereas it is considered to be overestimating if its value is negative [40–44]. i=1 n 

R=

i=1

n 

  (xi − x) yi − y (11)

(xi − x)2 ·

i=1 n 

yi − y

i=1 n 

xi − yi

ME =

RMSE =

2

 (12)

n



i=1

n 2



xi − yi  n

(13)

where x1 , x2 , ..., xn are the predicted values, y1 , y2 , ..., yn are the actual values, x and y are the mean of predicted and actual values, respectively. The attributes were considered as additional information about the geotechnical behavior of the different layers, such as soil properties and tectonic activities. The soil

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

49

properties comprise the physico-mechanical parameters that are obtained from both laboratory and in-situ tests. The spatial datasets of soil properties, the in-situ tests, and soil samples are analyzed per geotechnical layer. A total of 19 parameters are considered as soil property attributes for each geotechnical layer: The particle size limits of gravel, sand, silt, and clay are >2 mm, 0.06–2 mm, 0.002–0.06 mm, and 1, the void ratio >1, the angle of internal friction 0.1 cm2 kg1 [10, 45, 47, 48]. Tectonic activities were interpreted based on the fault system. However, these geological properties could not be fully represented in 2D by the strike and dip only. Based on the volumetric body of the high-resolution 3D geotechnical model, the planes of faults could be reconstructed using ArcScene through associating polygonal lines by direct triangulation of digitized faults from the geological map. The dipping direction of faults is simplified by having a constant slope. The volumetric body of the high-resolution 3D geotechnical model is constructed by putting the DEM and all the interpolated top elevations of geotechnical layers in stratigraphic order.

5 Results and Discussion Since exhaustive geotechnical characterization for 3D modeling, all available traditional geotechnical data was collected and analyzed in a systematic way to integrate into

50

V.-H. Nhu

the model by an elaborate data analysis. The domain of the model was taking into consideration the overall objective and available geotechnical data in the Hanoi city and surroundings. Within this domain, the model had been based on the interpolation of the real geotechnical system. The high-resolution 3D geotechnical model was constructed with three main components, including (i) ground surface, (ii) subsurface, and (iii) attributes. From the sedimentary conditions and the geotechnical characteristics and available borehole logs, the modeling area was delineated as a rectangular area, including the Hanoi city or area of interest. The domain stretches from longitudes 105°41 E to 105°56 E and from latitudes 20°53 N to 21°07 N, it has an area of about 652 km2 . In the vertical direction, the domain is bounded on top by the ground surface so that it includes the Quaternary sediments. The bottom limit of the domain consists of the Neogene sediment bedrock. Since the Hanoi city is a the Red River delta part, stratification is defined as conformity with continuous sediment deposition (Fig. 1). The ground surface of the 3D model was constructed as a DEM with 30 × 30 m pixel size. Its altitude ranged from 2.2 to 13.8 m a.s.l and gradually inclines towards the southeast. The highest altitude areas were also present along the main rivers as dykes (Fig. 1). The subsurface of the 3D model included 21 geotechnical layers that were defined and interpreted. They were assigned from 1 to 21 by descending stratigraphic order and are correlated to the formations of geology. The spatial database for these geotechnical layers was constructed as a database in Microsoft Excel with four separate files (borehole logs, geotechnical layers, in-situ tests, and laboratory test) that related to each other by the key field of “drilling record name”. Since the large variety in the depth of the borehole logs, the amount of information from borehole logs decrease with depth. Therefore, for the deep geotechnical layers, less data points with elevation data were available. After trend identification elimination, variograms of the residual datasets of the 20 geotechnical layers were calculated and interpolated by Kriging. The final experimental and modeled variograms for these top surfaces are listed in Table 1 (Figs. 7 and 8). The values of R, RMSE, and ME of the actual versus the predicted elevations correspond to (0.81–0.94), (-0.001–0.081 m), (0.59–3.92 m) for all 20 geotechnical layers are satisfactory. It should be noted, however, the predicted elevations were slightly underestimated for geotechnical layers 11, 12, & 21 and overestimated for geotechnical layer 20. These errors can probably be attributed to the scattered spatial distribution of drilling record locations and the complexity of the geological system. The prediction error indices from cross-validation and the interpolated elevations of the geotechnical layers are shown in Table 1 and Fig. 6, respectively. Based on visual inspection of the results, the predicted elevations of all layers are gradually decreasing from the north (mountain region) to the south of the Hanoi city. Geotechnical layer 20 is the thickest with predicted top and base elevations of (−20 m, −48.8 m) and (−50.1 m, −70.8 m). The base surface of this geotechnical layer is also the base surface of the Quaternary sediments or the top surface (geotechnical layer 21) of the Neogene bedrocks. The thinnest geotechnical layers are layer 18 and two groups of layers comprising layers 2 to 7 and layers 10 and 11, respectively.

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

51

Table 1. Final theoretical variogram models and the prediction error indices through crossvalidation of top elevation values of the geotechnical layers (GL) GL Final theoretical model

Sill (m2 )

Nugget Major (m2 ) range (m)

Minor range (m)

Major Lag R direction (km) (deg)

RMSE ME (m) (m)

2

Exponential

4.99 0.06

19.560 29.340

38.1

2.45 0.93 0.69

−0.001

3

Gaussian

0.92 0.12

1.096

93.0

0.14 0.92 0.80

−0.005

4

Exponential

4.88 0.13

20.475 30.713

33.2

2.56 0.89 0.80

−0.004

5

Spherical

3.89 0.31

8.587

12.880

31.3

1.07 0.85 0.99

−0.002

6

Gaussian

1.59 0.13

1.610

1.815

175.8

0.20 0.94 0.59

0.005

7

Gaussian

1.49 0.32

1.224

1.016

104.9

0.15 0.90 0.72

0.006

8

Spherical

8.63 1.48

685

916

29.5

0.07 0.83 1.59

0.015

9

924

Exponential

20.31 0.00

803

1.154

14.6

0.10 0.81 3.92

−0.021

10

Gaussian

16.65 3.07

1.429

1.210

106.9

0.18 0.86 2.30

−0.013

11

Gaussian

19.46 2.99

1.391

1.571

4.4

0.17 0.90 2.40

−0.053

12

Gaussian

24.20 2.95

1.594

1.945

26.4

0.20 0.91 2.80

−0.057

13

Gaussian

26.83 4.66

1.388

1.524

13.2

0.17 0.91 2.80

−0.036

14

Gaussian

25.74 4.61

1.682

2.523

19.9

0.21 0.92 2.77

0.016

15

Exponential 118.46 2.50

21.537 15.710

85.3

2.69 0.91 2.95

−0.010

16

Gaussian

27.24 5.89

1.667

1.422

101.3

0.21 0.91 2.45

−0.027

17

Gaussian

21.77 4.49

1.635

1.305

65.0

0.20 0.89 2.56

0.012

18

Gaussian

19.80 5.63

1.562

1.163

60.5

0.19 0.90 2.52

0.019

19

Gaussian

19.00 4.74

1.534

1.026

60.8

0.19 0.94 1.39

−0.005

20

Spherical

7.20 0.00

1.088

622

51.2

0.91 0.92 2.34

0.081

21

Gaussian

15.06 1.61

3.780

4.111

173.7

0.47 0.91 2.80

−0.057

Except for Neogene bedrock, the attributes of the soil properties were defined from in-situ and laboratory tests performed within each GL. The soil properties are described here for four different types of material, such as gravel, sand, clay, and soft (or organic) soils. The gravel soil is found only in GL20 with a composition consisting of cobbles and pebbles with sand. Its properties range as 8.0 ÷ 13.0 MPa in elastic modulus and 25.9 ÷ 101.3 md−1 in hydraulic conductivity of 25.9 ÷ 101.3 md−1 . The sandy soils were defined as being composed of fine to coarse sands and are found in GLs of 6, 9, 16, and 17. Their properties as 5 ÷ 27 MPa in modulus and 3.0 ÷ 33.1 md−1 in hydraulic conductivity. The clayey soils are found in GLs of 1, 3, 4, 5, 8, 13, 14, 15, 18, and 19 with compositions such as clay, sandy clay and clayey sands. The properties of these soils as 25.5 ÷ 35.6% in moisture content, 30.9 ÷ 44.7 in the liquid limit, 21.6 ÷ 25.2% in the plastic limit, 0.77 ÷ 0.99 in void ratio, 9.5 ÷ 8.2*10–4 md−1 in hydraulic conductivity, and 1.5 ÷ 100.0*10–6 cmkg−1 in volume compressibility coefficient. The organic soils,

52

V.-H. Nhu

Fig. 6. The high-resolution 3D geotechnical model of Hanoi city with 20 predicted top elevations (m a.s.l) of geotechnical layers and attributions

including silt, mud, peat, and clayey soil, are found in GLs of 2, 8, 11, 12, and 13. Their properties range as 36.3 ÷ 63.7% in moisture content, 42.9 ÷ 62.8% in the liquid limit, 26.8 ÷ 38.2% in the plastic limit, 1.04 ÷ 1.90 in void ratio, and 2.3 ÷ 2.7*10–6 cmkg−1 in volume compressibility coefficient. Tectonic activities were interpreted based on the fault system. However, these geological properties could not be adequately represented in 2D by the strike and dip only. Based on the volumetric body of the high-resolution 3D geotechnical model, the planes of faults were reconstructed by direct triangulation through associating polygonal lines of digitized faults from the geological map. The dipping direction of faults is simplified by having a constant slope.

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

53

Fig. 7. The high-resolution 3D geotechnical model of Hanoi city with 20 predicted top elevations (m a.s.l) of geotechnical layers and attributions (1st continued)

54

V.-H. Nhu

Fig. 8. The high-resolution 3D geotechnical model of Hanoi city with 20 predicted top elevations (m a.s.l) of geotechnical layers and attributions (2nd continued)

6 Conclusion In this study, a high-resolution 3D geotechnical model of Hanoi city was build using maps (topography, Quaternary sediments, geology), geotechnical data (1,386 borehole logs with attributes of 10,278 soil samples and 16,626 in-situ tests), and geostatistical analysis supplemented by cross-validation. The high-resolution 3D geotechnical model of Hanoi city provided an exhaustive geotechnical characterization with a total of 21 volumetric GLs with the attribute of 19 soil parameters and tectonic activities. The findings of them were satisfactory from statistical evaluation metrics. It is also feasible as a powerful tool for the reproduction and analysis when it allows extracting spatial

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam)

55

distribution of any layer (or point, column) in elevation (or depth, volumetric) and any 2D geotechnical map at different elevation/depth. From the resultant model, challenges were solved such as (1) the complexity of geotechnical conditions have been transited to a computational representation; (2) the collected and archived data have been increasing quantity, their increased sophistication could be used to improve the variogram analysis results with a better interpolation; (3) the complexity of subsurface was constructed by systematic geostatistical analysis; and (4) the high-resolution 3D geotechnical model could allow performing complex analyses and computations on the geo-framework for different usage. By this study, it is the first-time geotechnical characterization of Hanoi city was reproduced with high-resolution in comparison with previous studies [7, 8, 10, 21, 49– 51]. It is possible that, demonstrated feasibility of the high-resolution 3D geotechnical model of Hanoi city using the proposed methodology. From it, valuable information could be provided for dealing with the complicated environment of subsurface as urban development and geohazard mitigation for the Hanoi city. Acknowledgments. The author would like to thank Prof. Marijke Huysmans from VUB-HYDR, who provide the language and help.

References 1. Azaronak, N.: Building 3D models from geotechnical data (2015) 2. Dong, M., Neukum, C., Hu, H., Azzam, R.: Real 3D geotechnical modeling in engineering geology: a case study from the inner city of Aachen. Germany. Bull. Eng. Geol. Environ. 74(2), 281–300 (2015) 3. Lelliott, M., Bridge, D.M., Kessler, H., Price, S., Seymour, K.: The application of 3D geological modelling to aquifer recharge assessments in an urban environment. Q. J. Eng. Geol. Hydrogeol. 39, 293–302 (2006) 4. El May, M., Dlala, M., Chenini, I.: Urban geological mapping: geotechnical data analysis for rational development planning. Eng. Geol. 116, 129–138 (2010) 5. Haeri, S.: The role of geotechnical engineering in sustainable and resilient cities. Sci. Iran. 23, 1658–1674 (2016) 6. Galloway, D., Burbey, T.: Review: regional land subsidence accompanying groundwater extraction. Hydrogeol. J. 19, 1459–1486 (2011) 7. Dai, N.D.: Report on the results of the Hanoi Urban Geological Survey. Federation II GeoHydrology, Department of Geology of Vietnam, Hanoi (1996) 8. Federation II - GeoHydrology: Report on Geological Investigation of Hanoi - Hai Phong Quang Ninh Economic Corridor. Federation II - GeoHydrology, Department of Geology of Vietnam, Hanoi (1996) 9. Thang, L.T.: Studying the types of soft ground structures in Hanoi area and assessing their ability to use them in construction. vol. Ph.D. Hanoi University of Mining and Geology (1995) 10. Phuong, N.H., Thinh, T.D., Ty, P.V., Tuong, D.T., Hong, N., Tinh, N.V., Vu, T.X., Hai, H.V., Bat, D.V., Long, N.H., Quang, N.H., Phong, N.V., Phuc, D.D., Binh, T.T., Tuan, P.M., Huong, P.T., Minh, P.V., Nga, P.T.V., Long, N.T., Nu, N.T., Hung, N.V., Hai, P.H., Hung, N.M., Thong, V.D.: Collection and verify data for supplement studying of soft soil zonation mapping in Hanoi area (TC-ÐT/06-02-03) (2004)

56

V.-H. Nhu

11. Robins, N., Davies, J., Dumpleton, S.: Groundwater flow in the South Wales coalfield: historical data informing 3D modelling. Q. J. Eng. Geol. Hydrogeol. 41, 477–486 (2008) 12. Roy, D., Robinson, K.E.: Surface settlements at a soft soil site due to bedrock dewatering. Eng. Geol. 107, 109–117 (2009) 13. Lelliott, M., Cave, M., Wealthall, G.: A structured approach to the measurement of uncertainty in 3D geological models. Q. J. Eng. Geol. Hydrogeol. 42, 95–105 (2009) 14. CWRCT: Report of geodesy investigation of Red River at scale of 1:5000. Center for Water Resources Consultant and Technology Transfer (CWRCT) (2010) 15. Tanabe, S., Saito, Y., Lan Vu, Q., Hanebuth, T.J., Lan Ngo, Q., Kitamura, A.: Holocene evolution of the Song Hong (Red River) delta system, northern Vietnam. Sediment. Geol. 187, 29–61 (2006) 16. VIGMR: Hanoi geological map at scale of 1:50000. Vietnam Institute of Geosciences and Mineral Resources (VIGMR) (2010) 17. Mathers, S., Zalasiewicz, J.: Holocene sedimentary architecture of the Red River delta, Vietnam. J. Coast. Res. 15, 314–325 (1999) 18. Funabiki, A., Haruyama, S., Van Quy, N., Van Hai, P., Thai, D.H.: Holocene delta plain development in the Song Hong (Red River) delta, Vietnam. J. Asian Earth Sci. 30, 518–529 (2007) 19. VIGMR: Hanoi quaternary deposit map at scale of 1:150,000. Vietnam Institute of Geosciences and Mineral Resources (VIGMR) (1995) 20. Nguyen, D.D.: Report on investigation of urban geology for Hanoi city. Department of Hydrogeology 2 (1996) 21. Thu, T.M., Fredlund, D.G.: Modelling subsidence in the Hanoi City area, Vietnam. Can. Geotech. J. 37, 621–637 (2000) 22. Giao, P.H., Ovaskainen, E.: Preliminary assessment of Hanoi land subsidence with reference to groundwater development. Lowl. Technol. Int. 2, 17–29 (2000) 23. VIGMR: Hanoi hydrological map at scale of 1:50000. Vietnam Institute of Geosciences and Mineral Resources (VIGMR) (2010) 24. Jusseret, S., Tam, V.T., Dassargues, A.: Groundwater flow modelling in the central zone of Hanoi, Vietnam. Hydrogeol. J. 17, 915–934 (2009) 25. Hai, H.V.: Some new discovery about new tectonic in Hanoi area and surroundings. J. Geol. A, 42–49 (2007) 26. NHMS: Report of metorological data in range of 1980–2013 at Hanoi hydro-metorological station. National hydro-metorological service (2014) 27. NARENCA: Topographic maps at scale of 1:2000. Viet Nam Publishing House of Natural Resources, Environment and Cartography (NARENCA) (2000) 28. VIGMR: Hanoi lithological map at scale of 1:50000. Vietnam Institute of Geosciences and Mineral Resources (VIGMR) (2010) 29. Pollard, D.D., Fletcher, R.C.: Fundamentals of Structural Geology. Cambridge University Press, Cambridge (2005) 30. Delaunay, B.: Sur la sphere vide. Bull. Acad. Sci. USSR 7, 793–800 (1934) 31. Peucker, T.K., Fowler, R.J., Little, J.J., Mark, D.M.: The triangulated irregular network. In: American Society of Photogrammetry. Proceedings of the Digital Terrain Models Symposium, p. 532 32. Bondarik, G.K.: Dinamitreckoe i xtatitreckoe zondirovanie gruntop Vinginemoi geolopi. M. Nedra (1964) 33. RockWare: RockWorks15 Manual. RockWare, Inc. (2008) 34. Matheron, G.: Principles of geostatistics. Econ. Geol. 58, 1246–1266 (1963) 35. Krige, D.G.: A statistical approach to some basic mine valuation problems on the Witwatersrand. J. Chem. Metall. Min. Soc. South Afr., 201–223 (1951)

Building a High-Resolution 3D Geotechnical Model of Hanoi City (Vietnam) 36. 37. 38. 39. 40. 41.

42.

43. 44. 45. 46. 47. 48. 49. 50. 51.

57

Webster, R.: Quantitative spatial analysis of soil in the field. Adv. Soil Sci. 3, 1–70 (1985) Isaaks, E.H., Srivastava, R.M.: An introduction to applied geostatistics. AMC 10, 12 (1989) Pyrcz, M.J., Deutsch, C.V.: Geostatistical Reservoir Modeling. OUP USA (2014) Erdogan, S.: A comparision of interpolation methods for producing digital elevation models at the field scale. Earth Surf. Proc. Land. 34, 366 (2009) Schlumberger: Visual Modflow user’s manual. Waterloo Hydrogeologic, Waterloo, ON (2010) Feng, S., Huo, Z., Kang, S., Tang, Z., Wang, F.: Groundwater simulation using a numerical model under different water resources management scenarios in an arid region of China. Environ. Earth Sci. 62, 961–971 (2011) Schoups, G., Addams, C.L., Gorelick, S.: Multi-objective calibration of a surface watergroundwater flow model in an irrigated agricultural region: Yaqui Valley, Sonora, Mexico. Hydrol. Earth Syst. Sci. Discuss. Discuss. 2, 2061–2109 (2005) Witten, I.H., Frank, E.: Data Mining: Practical Machine Learning Tools and Techniques, 2nd Edition. Morgan Kaufmann, Burlington, 2005 (2011) Ormsby, T., Napoleon, E., Burke, R., Groessl, C., Bowden, L.: Getting to Know ArcGIS Desktop. Esri Press Redlands, Redlands (2010) TCVN: Vietnamese standards for soil classification 5747:1993, vol. 5747:1993, p. 10 (1993) Smoltczyk, U.: Geotechnical Engineering Handbook. Ernst & Sohn, Berlin (2003) TCXD: Ground Improvement by Prefabricated Vertical Drain (PVD), vol. 245-2000, p. 10 (2000) 22TCN: Instruction of investigation and design automobil embankment on soft soil foundation, vol. 262–2000, p. 51 (2000) Manh, N.D., Lan, N.N.: Geo-environment with efficient exploitation and use of Hanoi urban underground space (2010) Yonezawa, G.: Developing of 3D urban modeling of Hanoi. Intriguing ASIA 113, 168–174 (2008) Yonezawa, G.: 3D topographical analysis of Hanoi, Vietnam. Southeast Asian Stud. 46, 519– 531 (2009)

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration in the Nui Phao Area, Northeastern Vietnam Khuong The Hung1(B) , Luong Quang Khang1 , Pham Nhu Sang1,2 , and Hoang Van Vuong3 1 Hanoi University of Mining and Geology, 100000 Hanoi, Vietnam

[email protected] 2 State Key Laboratory of Marine Geology, Tongji University, Shanghai 20092, China 3 Song Da 5 Joint Stock Company, 100000 Hanoi, Vietnam

Abstract. Nui Phao area (northeastern Vietnam) has a high potential for tungsten resources. The tungsten ore bodies mainly appear in lens-shaped within granitic rocks of the Da Lien complex in the studied area. Minerals accompanying tungsten deposits are composed of fluorite, native gold, native bismuth, chalcopyrite, a lesser amount of allanite, cassiterite, and rare molybdenite. Based on collecting, synthesizing, geological processing data, and the mathematical method, studied objects of the exploration process are quantitatively described. Tungsten oxide (WO3 ) contents of the major ore body vary from 0.20 to 1.11% with a coefficient of variation (Vc ) of 91.2% (unevenly). Generally, the tungsten oxide contents compliance with rules of the standard lognormal distribution. Major tungsten orebody average 56.7 m in thickness and its coefficient of variation (Vm ) of 61.2% (unstable). Quantitative calculations reveal the Nui Phao tungsten deposit belongs to the III-type of mining exploration group. To explore this type of minerals, a linear grid pattern should be applied. Results show that the appropriate pattern grid exploration for reserve level 122 is 50 ÷ 60 × 30 ÷ 35 m; these values can be applied to other deposits occurring in similar geological settings. Keywords: Mining deposit group · Pattern grid exploration · Tungsten ore · Nui Phao area · Vietnam

1 Introduction The beginning of any mineral explorations is often applied with the orientation exploration grid, following the principle of geological similarity [2, 4, 5, 8, 9, 17]. However, applications of this principle have some certain issues due to ore bodies are frequently not the same in size, thickness, and internal orebody structure, so on. Therefore, the application of geological math modes in the research-exploration data processing is evaluated as the most efficient method for selecting suitable exploration grid of each specific mineral object and mining deposit groups [8, 14, 15, 18]. Many projects have © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 58–78, 2021. https://doi.org/10.1007/978-3-030-60269-7_4

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

59

achieved many successes in geological research by applying for statistical methods in setting up an exploration grid [2, 12, 15, 19, 20]. This indicates that this step plays an essential role in evaluating the effectiveness and reliability of mineral exploration. Based on the complexity, size, and shape of mineral deposits, the Circular of Vietnam Ministry of Natural Resources and Environment (VMNRE) [17] classified deposits into four groups as follows: (1) Group I comprising large deposits that are simple in geological structure, with the least variation in thickness and grades, thus, the highest level of resources are 121 by applying normal grid of drill holes; (2) Group II is composed of relatively large deposits having more complicated structure, more variable thickness, and significant grade variability. By applying a normal grid of exploration, only up to reserves of level 121 might be defined; (3) Group III consists of highly complicated structure deposits. The ore thickness varies significantly, with very uneven grade distribution. Deposits that belong to this group are often smaller in size and minerals distribute unevenly. Category of reserve 122 can be established by a normal grid exploration; and (4) Group IV deposits are extreme complexity in geological structure, grade distribution, and thickness. They are usually small deposits or ore-pocket with very complicated shapes. To establish category reserve 122, dilling in combination with underground works must be applied. It is documented that Northeastern Vietnam is abundant in mineral resources for developing mining and metallurgical industry. Although some tungsten ore deposits have been investigated during geological mapping in northeastern Vietnam, most of them are evaluated as small to medium deposits. However, the tungsten ore in the Nui Phao area is an exception [4, 6, 9]. Up to date, there have been no adequate and systematic studies on the geological aspects in combination with tungsten mineralization, especially the mining exploration group and exploration grid in the area. Thus, the results of this work will be important in mineral exploration and mining for the future. The purposes of this study are to establish a tungsten mining group in the Nui Phao area based on applying statistical models to determine a pattern grid exploration of estimated tungsten orebody parameters. However, these methods can be used or solving problems from other areas having the same geological conditions.

2 Geological Features of the Nui Phao Area Northeastern Vietnam belonged to the South China plate, and it is separated from northwestern Vietnam by the Song Hong (or Red River) shear zone, which is one of the main tectonic structures of Vietnam [7, 11, 16]. In which geological strata and igneous rocks have been found dating from the early Paleozoic to the Quaternary. The Nui Phao area is located near the town of Dai Tu in northeastern Vietnam, approximately 80 km north of Hanoi (Fig. 1a). The Nui Phao area coincides with a strong, WNW-trending, positive magnetic anomaly, which extends for over 2 km in length and 400 m to 500 m width. Drilling has confirmed polymetallic mineralization along 1.3 km of this strike length, but the zone is open along strike in both directions. The lithology of the Nui Phao area is composed mainly of clay shale muscovitebearing quartzitic sandstone, chert; sandstone interbeds of coaly shale rocks that were suggested as Ordovician-Silurian age and named Phu Ngu formation [9]. In the northern

60

K. T. Hung et al.

study area, intrusive rocks of granite are exposed in mass-shaped, termed as Da Lien block (Fig. 1b). Quaternary sediment mainly exposes along valleys and/or lowlands.

Fig. 1. General geological diagram of the southwestern South China plate and Northeastern Vietnam (adapted from [3, 13, 21]) (a); A general geological map of the Nui Phao area, Thai Nguyen province (adapted from [4, 6]) (b)

There are three fault systems recorded in this study which were well documented in the previous researches [4, 6]. They are composed of northeast-southwest (NE-SW), northwest-southeast (NW-SE), and near west-east (W-E) trending systems, of which the NW-SE one is the major system, controlling the structure of the Nui Phao area. Most of the orebodies are associated with the fault system in the Da Lien and Nui Phao granitic massives [1, 6].

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

61

3 Features of the Tungsten Orebody at the Nui Phao Area 3.1 The Major Characteristics of the Tungsten Orebody The Nui Phao deposit contains several tungsten orebodies, in which the major orebody is a prospective one, and is being explored and exploited by Tiberon Minerals Company. Mineralization in the studied area has the following characteristics. The major tungsten orebody occurs as greisens formed both internally and externally contact between the Da Lien granite and Phu Ngu sedimentary rocks (Fig. 2). This orebody is extended about 2 km along with strike line (from east to west), and ranging from 200 m to 400 m in width. The thickness of this orebody can be up to 159 m in the east and 43 m in the west. The high point of the Da Lien granite massive named "granite blade" that plays as a key point of the major orebody. To the east of the granite blade, the major tungsten orebody plugs to the east and is related to a metamorphic zone of granite rocks of up to 50 m thickness (mainly internal greisenized granitoids). To the north and south, the major tungsten orebody is controlled by the granite boundary. To the west of the granite blade, the orebody is inclined to the west and thinned, but its scale may be larger (up to 450 m of width).

Fig. 2. Geological cross-section along line No. 568790E of the Nui Phao area (adapted from [4, 6])

The upper part of the major tungsten body is strongly oxidized, forming a gossan zone, which is rich in quartz and iron. The part of the orebody is developed to northwest-northwest direction; it is showing the surface of polymetallic skarn/greisen major tungsten zone. The orebody is exposed to the area with dimensions of 850 m × 200 m × 10 m (length × width × depth). The lithological units of the major mineralized zone are composed of the interchanged combination of the products of thermal metamorphic processes, scarification, albitization, and greisenization of dike and granite. They are surrounded by the Nui Phao and Da Lien granite massifs and are covered by the clay and weather materials of 20 to 40 m in thickness. The major tungsten orebody is spilled from the skarn and greisen orebodies by the ridge of the Da Lien granite massive. Sometimes, metasomatic rocks compose mainly of pyroxene-(garnet) skarn, amphibole-biotite-(danalite) skarn, calc-silicate hornfels, marble, and magnetite-(danalite) skarn. Dike rocks of granitoids intrude sedimentary rocks of the Phu Ngu formation and are also metasomatized. The skarn alteration surrounding the Da Lien granite contact is overprinted by albite fluorite

62

K. T. Hung et al.

greisenization accompanied by biotite and pyrrhotite, and quartz veins have appeared too. The polymetallic (Sn-W-Mo) mineralization is consists of fluorite, scheelite, native gold, native bismuth, chalcopyrite, and they are mainly developed in the greisenized rocks. Other minor minerals are allanite, cassiterite, and rare molybdenite and Pb-Zn sulphides. 3.2 Quality of Tungsten Orebody Tungsten orebodies mainly present at contact zones between two mica granitoid and the pyroxene-garnet skarns. They are mainly located in Proterozoic Phu Ngu formation and greisenized granite and are classified as outer and inter contact zones, respectively. Most of the ores belong to veinlet-disseminated type, making an account for around 90% of all ore types which are recorded in the studied area. This type of ore is similar to quartz-scheelite ore, which is commonly distributed in feldspar metamorphic skarn rocks and overlapping greisen metamorphic rocks, partly less than the distribution in the greisenized granite. Ore minerals are mainly scheelite, wolframite, chalcopyrite, molybdenite, pyrite, magnetite, ranging from 10 - 40% in the composition. Whereas, the gangue minerals are often taken a larger proportion, comprising quartz, feldspar, biotite, and clay minerals group (Fig. 3). The magnetite is anhedral to sub-euhedral in shape, with the size ranges from 0.20– 0.80 mm, rarely up to around 2 mm. Magnetite appears in ore bunch, band-shaped, disseminated and scattered in skarn rocks (Fig. 4a, b), and are often associated with pyrite I and chalcopyrite I (Fig. 4c, d). The scheelite in the veinlet-disseminated tungsten ore type is 0.05–0.50 mm in size, sometimes ≈1 mm, and majority produce in anhedral and/or granular crystals. Based on characteristics of distribution, morphology, size, ore minerals relation, and mineral association, they can be distinguished two scheelite genesis. The scheelite I have a paragenetic relationship with pyrite I and chalcopyrite I that forming a mineral association (Fig. 5a, b). Scheelite II exists in the form of semi- idiomorphic and allotriomorphic granular with sizes ranging from 0.2 to 1.5 mm, sometimes > 2 mm. Scheelite II is distributed closely with quartz I and fluorite to form fissures, veins, disseminate in greisenised (outer-greisen) skarn rocks, and even in greisenised granite (internal greisen) of Da Lien massive. Scheelite II with quartz I and fluorite that was forming a mineral association (Fig. 5c, d). Wolframite is generally anhedral granular, sometimes hypidiomorphic tabular or columnar crystals in the veinlet-disseminated tungsten ore type between 0.01–.074 mm in size. They are frequently scattered between quartz and mica or other silicate minerals; however, some wolframite grains associated with pyrites and chalcopyrites are also observed (Fig. 6). Copper minerals appear mainly in sulphides including chalcopyrite, bornite, tetrahedrite. In which, chalcopyrites are often anhedral granular textures between 0.2–0.5 mm in size. Molybdenites are subhedral to anhedral in a quartz vein with foliated or scaly aggregation texture and are associated with magnetite, pyrrhotite (Fig. 7). The grain size of the mineral varies 0.05–0.20 mm.

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

63

Fig. 3. The gangue minerals of Nui Phao tungsten deposit; Pyroxene (hedenbergite) - vesuvian mineral association (a). Garnet minerals are replacement by hastingsite (b); Hastingsite is replacement by biotite, danburite (c) and is corroded by the scheelite, fluorite, and ore minerals (d); Biotite mineral occurs with minerals of the greisenized process (e), and ore minerals (f).

4 Background of the Methods Used 4.1 One-Dimensional Statistical Models Based on exploration data, the method is applied to highlight descriptive parameters of the orebodies, including chemical compositions, thickness, technical and physical properties. Probability theory and its models: The sampling distribution of analytical results is considered as a random variable when derived from a random sample of size n; they are displayed as frequency nomogram or cumulative frequency. If we increase the number of measurement points and decrease the size of the range, then its nomogram will become a continuous curve represented by the distribution of the probability of the occurrence of random variables. This curve form is distributed as a reflection of the geological processes, properties or phenomena to be studied.

64

K. T. Hung et al.

Fig. 4. Magnetite (Mt) occurred closely with pyrrhotine I (Pyr) and chalcopyrite I (Chp), they form a mineral association (a, b); Magnetite of xenomorphic texture, band-shaped in the skarn rocks (c- is captured under reflection contrast microscopy, d- is captured under scanning electron microscope-SEM).

Fig. 5. Scheelite I is captured under reflection contrast microscopy (a) and SEM with checking points of mineral components (b). Scheelite II is distributed in the outer and inter greisenized zone (c, d).

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

65

Fig. 6. Wolframite (W) is disseminated in granite rocks, which is corroded by pyrrhotine (Pyr).

Fig. 7. Molybdenite (Mo) are lamella, foliated in greisenized rocks (a), and quartz vein cutting skarn rocks (b).

Fig. 8. Diagram of the normal standard distribution function.

66

K. T. Hung et al.

The purpose of distributional function testing is to select mathematical equations to determine the mean and appropriate variance. Based on the distribution function, it is possible to determine the probability of random values occurring in any specified interval. The law of statistical distribution is divided into two groups, namely the group of discrete distribution rules: uniform distribution, binomial, polynomial, Poisson, and the group of continuous distribution rules, including Fiser, Student, normal distribution, lognormal, gamma, power standard distributions, and so on. The authors introduce some distribution models commonly used in exploratory geological research. * Normal standard distribution: If the research parameters belong to the standard statistical distribution model, the statistical characteristic quantities are determined by the following formulas. The average is a value representing the middle of a set of data values. y=

 nj 1 1 yi or y = nj yj = fj yj (fj = ) n n n n

n

n

i=1

j=1

j=1

(1)

In case of a large number of samples, we are divided data within class intervals (Z n  = 1 + 3.322lgN), the average content is determined by the formula: y = fi yi . i=1

where yi - the mean of data within i class intervals, fi – the frequency of the respective i class interval: fi = nni , ni – total numbers of samples within i class; n - total numbers of studied samples. The formula determines sample variance. σ2 =

 1  1  (yi − y)2 or σ 2 = (yj − y)2 = fj (yj − y)2 n−1 n−1 n

n

k

i=1

j=1

j=1

(2)

where σ2 - variance; yi – the value of the studied parameter at i sample (i work); yj the mean of data within j class intervals; n - total numbers of studied samples; nj - total numbers of samples within j class. The coefficient of variation is calculated by the formulas. V =

σ 100% y

The formula determines standard deviation. √ σ = σ2

(3)

(4)

where σ- standard deviation. The equation exactly defines normal distribution. f(y) =

(y−y)2 1 √ e 2σ 2 σ 2π

(5)

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

67

The equation defines the normal distribution function. F(y)

1 = √ σ 2π

y e

(y−y)2 2σ 2

dy

(6)

−∞

* Lognormal Standard Distribution In case of data are not belong to normal standard distribution, values are corrected logarithmically to apply the lognormal distribution. The formulas determine the statistical values of the lognormal distribution: The mean of data sets is determined by the formula. 1

m = eln y+ 2 σ

2

ln

(7)

The formula determines sample variance. 2

D = e2m+σln

(8)

The formula determines the coefficient of variation.  2 V = eσln − 1 × 100%

(9)

The equation precisely defines the lognormal distribution. f(y) =

1 √

σln y 2π

e

(ln y−ln y)2 2.σ 2 ln y

(10)

The equation defines the lognormal distribution function. F(y) =

y

1 √

σln y 2π

−∞

2

y) 1 (ln2.σy−ln e 2 ln y dy y

(11)

where ln y - the mean values of ln yi ; σlny - standard deviation of ln yi . * Statistical distribution test: To test the statistical distribution, we are using the method of kurtosis and skewness improver theory. The method follows these steps. The skewness (A) is determined by the formulas. n 

A=

(y − y)3

i=1

n.σ 3

−3

(12)

The kurtosis (E) is determined by the formulas. n 

E=

(y − y)4

i=1

n.σ 4

(13)

68

K. T. Hung et al.

The formulas calculate standard deviations (σA , σE ).   6 24 σA = And σE = n n The standard for skewness is determined by tA = determined by tE =

A σA ,

(14) and standard for kurtosis is

A (with A − skewness, E − kurtosis) σE

(15)

If |tA | ≥ 3 or |tE | ≥ 3, the distribution does not conform to the normal standard distribution being studied, then testing the lognormal distribution is using to the problem. Still, the values in equations of 12 ÷ 15 as yi replace with ln yi , σ replace with σlny , respectively. 4.2 Morphological and Internal Structural Orebodies The coefficient of ore-bearing (OBp ) is determined based on thickness, distribution, and length of the orebody as follows. – Based on thickness. N 

OBpm =

i=1 N 

ti (16) Ti

i=1

where t i - thickness of the portions of ore value in the i-th work, m; T i - thickness of rock layer containing tungsten ore, m; N - number of explored works. – Based on the distribution of the ore. N 

DOpS where

N  i=1

=

i=1

TAp

AR

(17)

TAp - total area of the orebodies in the explorated region, m2 ; N - number

of orebodies; AR - the area of the explored region, m2 .

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

69

– Based on the length of the ore. N 

LOPL =

TLP

i=1 N 

(18) TLc

i=1

where m.

N 

TLP - total length of orebodies, m;

i=1

N 

TLc - total length of exploration lines,

i=1

Coefficients of Discontinuity of the Orebodies (DOnp ) DOnp =

i OBpm

(19)

where i - the number of discontinuity determined on exploration lines cross-section; OBpm - coefficient of ore bearing. Cofficient of Anisotropy in Morphological Orebody (η) X (20) Y where X, Y-thickness and width of the ore body determined based on the geological map, m; The ore dressing coefficient (ξ ) is calculated as follows. η=

ξ=

MCtb NCCN

(21)

where MC tb - mean contents of payable tungsten bodies, (WO3 , %); NC CN – cut-off grade of the ore, %. Boundary modules (BM K ). Based on comparing the real circumference and perimeter of the orebody, the complexity of the orebody boundary is calculated as follows. eϕ (22) BMK = √ Lϕ 4.7a + 1.5 a − 1.77 Lϕ where a - half of the maximum boundary value, m; Lϕ - circumference of the orebody converted to ellipse shape; eϕ - the real perimeter of the orebody, m. Orebody Shaped Index (θ ) θ=

V .BMK OScc

(23)

where V - coefficient of payable orebody thickness, %; OS cc - coefficient of orebody structure complexity, %. OScc = 1 −

mtk nk mtq nq

(24)

where mt k - average thickness of dirt rock layers in the orebody, m; nk - number of dirt rock layers; mt q - average thickness of the ore beds, m; nq - number of ore beds.

70

K. T. Hung et al.

5 Proposed Methodology In a studied area, statistical estimation, descriptive patterns, and characteristic orebodies (i.e., shape, morphology, and structure) are used to create a mining deposit group and an exploration grid. The estimated results can provide useful statistical information on the characterization of morphology and geological structural orebodies. The mineral parameters of ore deposit can be estimated to be extensively utilized for quantitating mineral resource and determining geological exploration works in a studied area by applying statistical and random function methods [18]. However, these methods have not been widely utilized to identify a mining deposit group and a pattern grid exploration [15]. Based on previous materials, the geo-mathematic methods were applied to estimate the effective characterization of tungsten mineralization in the Nui Phao area. After that, a statistical model and random function theory were applied to estimate the exploration data of the major tungsten orebody. This methodology has been proposed and shown in Fig. 9.

Fig. 9. Three steps for estimating the characterization of tungsten mineralization in the studied area.

Step 1- Exploration data collection: the obtained data on the major tungsten orebody was collected and analyzed, then the data was processed. Each calculated parameter was associated with the tungsten orebody. The goal of this step was to establish the mining deposit groups. Step 2- Statistical identification of grid pattern: As an initial step, the processing requirements were performed using statistical methods. Next, the selected orebody features were applied to the given dataset to identify the pattern grid exploration. Lastly, a distance of geological exploration works was calculated to evaluate the grid distances following the strike line (line to line), and dip direction (point to point). Step 3- Identification of grid pattern by stable random functions: the grid pattern obtained from Step 2 was combined the results of this step by application of the theory of stable random functions. At this step, the auto-correlation radius (R(h)) based on the strike and dip formats of the orebody were estimated and plotted. The goal was to establish the pattern grid exploration. The proposed methods could be used by any scientists for improving methodology and extended applying domain. The methodology is introduced below.

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

71

5.1 Establishing a Mining Deposit Group In this study, variations in geological parameters (i.e., average values, variance, and coefficient) are determined to guarantee the truthfulness, efficiency, and non-error in data processing for ensuring reliability. According to the probability distribution function, the probability of random numbers prevailing in the arbitrary selection range is identified following Wellmer (1998) [18]. 5.2 Establishing the Pattern Grid Exploration The methods are consist of the statistical and theory of random function methods. The statistical methods allow measuring the errors of estimated reserves and the density estimation of the grid exploration. The methods of random functions are determined the correlation function of the norm based on the construction of correlation plots and an anisotropy coefficient. They are presented in more detail in Khang et al. (2020) [10].

6 Results and Discussion The distribution, structural, and morphological features, as well as their relationships, are clarified based on previous synthetic [4, 5, 6, 9], and additional research materials. In this studied area, they exist at a depth of the tungsten bodies. 6.1 Elements for Making a Mining Deposit Group According to Pogrebiski (1973) [14], he argues that elements for making a mining deposit group and a pattern grid exploration are the following factors, they are the first factor of geological structure and orebody morphology; the second one is mineral deposit scale or orebody size and structural orebody, and the third factor is the stable degree of orebody thickness and evenly of orebody contents. To determine the factors, they have distinguished based on the coefficient of variation as following Tables 1 and 2. Table 1. Establishing a stable degree of orebody thickness and evenly of orebody contents Coefficient of variation (V, %)

Thickness (m)

Content (%)

150%)

Complexity Very simple degree of orebody (Mk = 1 ÷ boundary module 1.2)

Simple (Mk = 1.2 Average (Mk = 1.4 Complicated to ÷ 1.4) ÷ 1.6) very complicated (Mk > 1.6)

Evently degree of Evenly (Vc < 40%) content

Unevenly (Vc = 40 ÷ 100%)

Very unevenly (Vc = 100 ÷ 150%)

Especially unevenly (Vc > 150%)

6.2 Determination of Exploration Tungsten Deposit Group 6.2.1 Statistical Features of the Major Tungsten Orebody The results on the statistical analysis of content and thickness of the major tungsten orebody are presented in Table 3. Table 3. Statistical features of the tungsten oxide content of the major orebody Tungsten oxide (WO3 ) contents

Major tungsten body

Mean content (%)

Variance (σ2 )

Coefficient of variation (V c , %)

tA

tE

0.41

0.14

91.3

1.82

1.63

Distribution pattern

Lognormal standard

Table 4. Statistical features of tungsten thickness of the major orebody Parameters of true thickness

Distribution pattern

Average (m)

Variance (σ2 )

Coefficient of variation (V m , %)

tA

tE

56.7

124.1

61.2

2.23

2.61

Normal standard

Table 3 shows that in the major orebody, the mean tungsten oxide content is 0.41%, its coefficient of variation (Vc ) is 91.3% (unevenly to very unevenly). On the whole, the

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

73

tungsten oxide contents of the major orebody have complied with the standard normal distribution. The major tungsten body shows an average thickness of 56.7, its coefficient of variation (Vm ) of 61.2%, and stable to very unstable distributions, indicating that the major orebody thickness has complied with the standard normal distribution. 6.2.2 Characterization of Continuous Mineralization The degree of ease of available exploration geology is mainly influenced by characteristics of continuous mineralization. This suggests that a systematic investigation of the continuity of the tungsten mineralization can follow as the applying Eqs. (16), (17), and (18) which are showed below. Table 5. Measured results of coefficients of tungsten ore bearing in the Nui Phao area OBpm DOpS

LOpL

Major orebody 0.35 0.041 0.39

Applying Eqs. (19), (20), and (21), the degree of discontinuous ore, morphological anisotropy, and coefficients of extractive metallurgy, mineral processing of the major tungsten orebody can be evaluated. Table 6. Calculated results of discontinuous tungsten in orebodies

Major orebody

Coefficients of discontinuous ore

Coefficients of morphological anisotropy

Coefficients of ore dressing

11.43

0.2

2.05

The results presented in Tables 5 and 6 points out that the major tungsten body can be an interruption and uninterrupted types; its coefficient of interruption ore is complex (DOnp = 11.43). Major tungsten body is commonly anisotropy shape. Tungsten contents belong to the medium type with the coefficient of ore dressing is 2.05. 6.2.3 Complexity Degree in the Tungsten Body Boundary Module and Its Shaped Index The morphological characterizations (i.e., shapes, strike and dip formats), and the complexity degree of the internal structure of major tungsten orebody are measured by applying Eqs. (22), (23), and (24). The results in the tungsten body boundary module and its shaped index are displayed below (Table 7). The results of the complexity of the major tungsten orebody are simple to complex and its shaped index is medium (Table 7).

74

K. T. Hung et al. Table 7. Complexity degree in the tungsten body boundary module and its shaped index

Major orebody

Area (m2 )

Boundary module

Complexity degree

Shaped index

809,100

5,927

1.42

0.32

Generally, the systematic investigation of the tungsten body indicates that the thickness varies from medium to small sizes, and its shape is quite medium. Variations in coefficients of thickness are invariable to variable types and interruption ones. The tungsten oxide contents are even to uneven distribution, implying that it has belonged to the average contents, buried by the burden, and light dips. The characteristics of the Nui Phao tungsten orebody and the materials from the VMNRE [17], the Nui Phao tungsten deposit can be categorized into deposit group III. 6.2.4 Determination of Pattern Grid Exploration of the Nui Phao Tungsten Deposit Based on the exploration of geological parameters, the determination of a triangle grid exploration or optimization of the grid exploration can be identified. They are dependent on mining geological structure characteristics and consider mainly explorer objects. The key of tungsten body parameters can mostly be as point reserves. If variation in thickness or tungsten oxide contents of the orebody is the greatest, the characteristics of the largest orebody can be utilized for selecting the exploration grid. a. Exploration System of the Effectiveness Relative errors of the major tungsten orebody are calculated by using the method proposed by Khang (2020) [10], and their results are shown in Table 8. The tungsten reserve of the major orebody shows an error of less than 50%, and it is estimated following category reserves 122 (Table 6). This suggests that the pattern grid exploration has been used for the tungsten body of the Nui Phao deposit which must be suitable for calculating category reserves 122 and it has to be normalized by the VMNRE [17]. Table 8. Relative errors of the tungsten reserves of the major orebody Relative errors of the tungsten reserves (%), t = 2 Major orebody

Area

Thickness

Content

Tungsten reserve

1.28

12.6

26.5

28.5

b. Density Estimation for Grid Exploration Based on the method proposed by Khang (2020) [10], the density of the grid exploration can be identified, and its results are shown below (Table 9). The results display that a linear should be used for the grid exploration of the tungsten deposit. The spacing of the explored line is utilized to be 60 m with 35 m spacing from the explored point to the nearest one. The number of exploration work is 476 works/km2 .

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

75

Table 9. The density of the grid exploration is calculated by using the statistical method The distance along orebody (m) Major orebody

sf – strike format

df - dip format

60

35

Density (m2 )

Number of explored works/km2

2100

476

Table 10. The grid exploration density is calculated by using a stable random function Anisotropy index

The distance along orebody (m)

Density (m2 )

Number of explored works/km2

1500–2100

476–667

sf – strike format df - dip format Major orebody

0.54

50–60

30–35

There is a certain relationship between the geological parameter of the orebody and the distance between explored works. This implies that the spacing density of explored works can play a significant issue for a particular grid exploration. It is caused by the exploration conditions which are not uneven distribution over the geometric grid. As a result, the proposed method of Khang (2020) [10] should be used to transform the collected value to the explored point of the base grid cells for the study area. The spacing of the explored line is used to be 50–60 m with 30–35 m spacing from the explored point to the nearest one. To guarantee the accuracy of the method, the biggest ones are chosen to perform measuring the autocorrelation radius along with strike and dip formats for content parameters of the major tungsten body. As soon as the authors establish an experimental auto-correlation radius R(h) deals with major tungsten body, the method proposed by Khang (2020) [10] is applied to conduct modelling. And then, we convert experimental induction line (R(h)) to theoretical line (R* (h)), which led to plot the figures, and calculation size of the impact zone (H) that is estimated following strike and dip formats (Fig. 10). The study shows that the line spacing 60 m, and 35 m spacing from the explored point to the nearest one is used to display better results than the line spacing 70 m and 30 m spacing from the explored point to the nearest one. The number of explored works is calculated to be between 476–667 works/km2 . The methods of statistical analysis and stable random function are combined to help to identify the grid exploration of reserve 122. The line spacing is used to be from 50 to 60 m, and spacing from the explored point to the nearest one is between 30 ÷ 35 m (Table 11).

76

K. T. Hung et al.

Fig. 10. Auto-correlation plots of the major tungsten body: R* - the theoretical tungsten line of correlation plots; 2σ – the experimental lines.

Table 11. Grid exploration calculated for tungsten reserves code 122 The distance along orebody (m) Major orebody

sf – strike format

df - dip format

Number of explored works/km2

50 ÷ 60

30 ÷ 35

430 ÷ 606

7 Concluding Remarks Quantification of exploration data and collection data obtained tungsten orebody are using for assessment in the mining deposit group and pattern grid explorations. The study introduced a method by combining statistical analysis and quantitative data of the orebody to determine in the distance of grid along strike and dip of orebody for tungsten metallic exploration object in the Nui Phao area, Northeaster Vietnam. The statistical method and random function overcome and obtain the tungsten mining groups and a pattern grid exploration. The results show that the major tungsten orebody in the Nui Phao area is mainly lens-shaped, fully distributed in granitoid rocks of the Phia Oac complex. The tungsten oxide contents are between 0.20% ÷ 1.11% and variations in the coefficient of (Vc ) are 91.2%, implying that they can be arranged in the standard lognormal rule. The tungsten bodies are characterized by 56.7 m thickness and 61.2% of variations in coefficient (Vm ). Based on the Circular of the VMNRE and the quantitative calculation results, this study can categorize Nui Phao tungsten deposit into deposit group III. The linear grid pattern should be used for exploration. In this area, the pattern grid exploration for

Establishing a Tungsten Deposit Group and a Pattern Grid Exploration

77

mineral reserve code 122 is chosen to be (50 ÷ 60) m × (30 ÷ 35) m. The results of this study are significant materials for recommending a mining deposit group and a pattern grid exploration not only for tungsten ore in the Nui Phao area but also for other tungsten ores which display similarly geological conditions. Acknowledgments. First of all, we would like to send our sincerest thanks to the leaders and staff of HUMG for providing us with excellent facilities in research and unconditional help to carry on the study. Furthermore, we thank our reviewers and editors at the ISRM2020 for their time and inspiring discussions.

References 1. Anh, P.L.: The petrologic genesis of high aluminum granitoid Pia Oac massive based on petrologic and geochemical isotope analysis. J. Geosci. 25(2): 134–141 (2003). (in Vietnamese) 2. Battalgazy, N., Madani, N.: Categorization of Mineral Resources Based on Different Geostatistical Simulation Algorithms: A Case Study from an Iron Ore Deposit. Natural Resources Research (2019) 3. Cheng, Y.B., Mao, J.W., Liu, P.: Geodynamic Setting of Late Cretaceous Sn-W Mineralization in Southeastern Yunnan and Northeastern Vietnam. Solid Earth Sci. 1(3), 79–88 (2016) 4. Dudka, S. (ed.): Report on the exploration results of W, Sn, Bi ores in Nui Phao mine. General Department of Geology and Minerals of Vietnam (2003) 5. Dung, V.T.: Characteristics of polymetallic W-Pb-Zn ores in Nui Phao deposit, Dai Tu, Thai Nguyen province. Doctor thesis. Hanoi University of Mining and Geology (in Vietnamese) (2017) 6. Hai, T.T., Farmar, M., Stemler, J.: The structural controls on mineralization at the Nui Phao polymetallic (W-F-Cu-Au-Bi) deposit, Northern Vietnam. Geological Association of Canada/Mineralogical Association of Canada Annual Meeting, Abstracts 28, 64 (2003) 7. Hung, K.T., Golonka, J.: Major plates and events shaping the complex tectonic of Northwest Vietnam. In: Németh, Z., Plašienka, D. (eds.) SlovTec 08, 6th Meeting of the Central European ˇ Tectonic studies Group (CETeG) & 13 Meeting of the Czech Tectonic Studies Group (CTS), 23–26 April 2008, Upohlav, Pieniny Klippen Belt, Slovakia. Proceedings and Excursion Guide. Dionyz Štúr Institute, Bratislava, pp. 60–61 (2008) 8. Kazdan, A.B.: Prospecting and exploration of mineral deposits. Nedra Publishers, Moscow (in Russian) (1997) 9. Ke, N.D. (ed.): Report on the prospective evaluation of W, Sn, Bi ores, and accompanying minerals in Dai Tu-Bac Thai province. General Department of Geology and Minerals of Vietnam (1991). (in Vietnamese) 10. Khang, L. Q., Hung. K. T., Hung, N. P., Ton, D. M.: Research identify the group for lead – zinc ore type to prepare exploration in Ban Lim area, Caobang province, Vietnam, E3S Web of Conferences 168, 00011 (2020) 11. Leloup, P.H., Tapponnier, P., Lacassin, R., Searle, M.P., Dailai, Z., Xiaoshan, L., Langshang, Z., Shaocheng, J., Trinh, P.T.: The Ailao Shan-Red River shear zone (Yunnan, China). Tertiary Transf. Bound. Indochina. Tectonophys. 251, 3–84 (1995) 12. Lui, Y., Xia, Q., Carranza, E.J.M.: Integrating sequential indicator simulation and singularity analysis to analyze uncertainty of geochemical anomaly for exploration targeting of tungsten polymetallic mineralization, Nanling belt. South China, Gexplo (2018)

78

K. T. Hung et al.

13. Mao, J.W., Cheng, Y.B., Chen, M.H., et al.: Major types and time-space distribution of mesozoic ore deposits in South China and their geodynamic settings. Miner. Deposita 48(3), 267–294 (2013) 14. Pogrebiski, E.O.: Prospecting and exploration of mineral deposits. Nedra Publishers, Moscow. (in Russian) (1973) 15. Saikia, K., Sarkar, B.C.: Exploration drilling optimization using geostatistics: A case in Jharia Coalfield. India. Appl. Earth Sci. 115(1), 13–25 (2006) 16. Tapponnier, P., Peltzer, G., Le Dain, A.Y., Armijo, R., Cobbold, P.: Propagating extrusion tectonics in Asia: New insights from simple experiments with plasticine. Geology 10, 611–616 (1982) 17. Vietnam Ministry of Natural Resources and Environment (VMNRE): The Circular of Promulgating the Regulation on the classification of solid-mineral deposits and resources (Number: 06/2017/QÐ-BTNMT), Hanoi, December 08, 2017 (2017). (in Vietnamese) 18. Wellmer, F.W.: Statistical evaluations in exploration for mineral deposits. Springer, Heidelberg, Printed in Germany (1998) 19. Yousefi, M.: Recognition of an enhanced multi-element geochemical signature of porphyry copper deposits for vectoring into mineralized zones and delimiting exploration targets in Jiroft area. SE Iran. Ore Geol. Rev. 83, 200–214 (2017) 20. Yuan, F., Li, X., Zhou, T., Deng, Y., Zhang, D., Xu, C., Jowitt, S.M.: Multifractal modellingbased mapping and identification of geochemical anomalies associated with Cu and Au mineralisation in the NW Junggar area of northern Xinjiang Province. China. J. Geochem. Explor. 154, 252–264 (2015) 21. Zhao, Z.Y., Hou, L., Ding, J., et al.: A Genetic Link between Late Cretaceous Granitic Magmatism and Sn Mineralization in the Southwestern South China Block: A Case Study of the Dulong Sn-Dominant Polymetallic Deposit. Ore Geol. Rev. 93, 268–289 (2018)

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors in Wastewater/Stormwater Network Using GIS-Based Fuzzy Analytical Hierarchy Process: A Case of Study in Ålesund, Norway Lam Van Nguyen1,2(B) , Dieu Tien Bui3 , and Razak Seidu1 1 Department of Ocean Operations and Civil Engineering, Norwegian University

of Science and Technology, NO-6025 Ålesund, Trondheim, Norway {Lam.V.Nguyen,rase}@ntnu.no 2 Department of Geodesy, Faculty of Geomatics and Land Administration, Hanoi University of Mining and Geology, No. 18 Pho Vien, Duc Thang, Bac Tu Liem, Hanoi 10000, Vietnam 3 Geographic Information Science Group, Department of Business and IT, University of South-Eastern Norway, N-3800 Bø, Telemark, Norway [email protected]

Abstract. Identification of optimal sensor placement in wastewater/stormwater networks plays a crucial role in monitoring the network’s status. This study proposes and verifies a new approach based on the fuzzy Analytical Hierarchy Process (AHP) and Geographic Information System for delineating potential areas for placing sensors of the wastewater/stormwater networks. The coastal city of Ålesund (Norway) was selected as a case study. In this regard, a GIS database was constructed, which consists of eight criteria, altitude, rainfall, geology, manholes, population density, critical infrastructures, road network, and traffic load. Using the fuzzy AHP, weights for the eight criteria were computed, and then, suitability maps for placement of the sensor position were generated in a GIS environment. The results showed that manholes, altitude, and rainfall are sensitive factors for placing sensors in wastewater and stormwater pipe network. The suitability maps in this study may provide initial information for the placement of flood monitoring sensors in the wastewater/stormwater network. Keywords: Optimal location · Geographic information system · Fuzzy analytical hierarchy process · Climate projection · Ålesund · Norway

1 Introduction Underground pipeline network infrastructure, including wastewater and stormwater pipes, play a crucial role in cities across the world. Wastewater and stormwater pipelines are designed to transport wastewater and stormwater from the city-sphere to protect © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 79–97, 2021. https://doi.org/10.1007/978-3-030-60269-7_5

80

L. Van Nguyen et al.

public health, environment, and properties. The robustness of wastewater and stormwater pipelines to perform this role can be significantly reduced overtime during their operational period due to intrinsic factors (size, age, material, and type) and extrinsic factors (earthquake, high pressure, and temperature environment) [1–3]. The adoption of the right maintenance strategies is, therefore, critical to ensure that the wastewater and stormwater pipes are performing their role effectively. This can be achieved through the development of a good predictive maintenance framework to support the maintenance activities of the pipes. Predictive maintenance is essential to ensure that the system can work efficiently, enhance equipment’s longevity, and minimize maintenance costs as well [4]. Predicting future conditions of wastewater pipe networks can be significantly improved through sensor networks. Installation of a large number of sensors in a network system for monitoring and predicting can provide valuable information on the health as well as the operational status of the wastewater pipe network. However, these sensors are expensive [5]; and therefore, choosing optimal locations for sensor placement will help water utilities to reduce the cost of using sensor technologies for operation and maintenance of the pipes. Recently, Sahoo, Yin and Liu [6] applied a clustering approach to determine optimal sensor placement and a minimum number of sensors for agro-hydrological systems. However, the authors showed that this method required a direct measurement of the state variable, which involves a complex computational process. In recent years, Geographic Information Systems (GIS) have proven as a useful tool for spatial analysis and prediction and have been employed for determining optimal sensor locations [7, 8]. Boulos and Schade [9] developed a GIS-based and hydraulic model to identify feasible nodes in a large-scale water distribution system for sensor placement and compared the results with PipelineNET. Although this method used a GIS-based approach and hydraulic model for analyzing data, the process of weight assessment was applied only to the nodes based on their locations compared with critical facilities without considering any other factors. Several models have been combined with GIS for optimal sensor placements in water and wastewater pipe network. An applied genetic algorithm combined with GIS has been successfully implemented for optimal placement of pressure sensors in a water distribution network [10]. The greedy-based algorithm was also successfully employed to identify the optimal placement of sensors in a water distribution network with beneficial computation [11]. The greedy algorithm was also applied for optimizing the location of wastewater quality monitoring sensors in Italy with good performance compared with the genetic algorithm [12]. Furthermore, optimal placement of water quality sensors has been successfully undertaken in the urban drainage network using Bayesian algorithm [13]. Multi-Criteria Decision Analysis (MCDA) methods have been developed, such as the analytical hierarchy process (AHP), analytic network process (ANP), the technique for order performance by similarity to ideal solution (TOPSIS), etc., [14]. The methods, integrated with a GIS-based approach, have been employed to identify optimal placements in many studies [7, 15–23]. In these methods, the AHP was used for calculating weights of different factors [20]. However, the normal AHP was remarkably influenced by the subjective judgment of experts. Therefore, the fuzzy algorithm was proposed

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

81

to deal with this problem [18, 24]. The objective of this study is to apply the Fuzzy Analytical Hierarchy Process (FAHP) method combined with GIS for the identification of suitable locations for sensor placements in a wastewater and stormwater network in Ålesund, Norway.

2 Methodology 2.1 Study Area The study was undertaken in Ålesund city, which is located in the Møre and Romsdal region of Norway [25]. The city is surrounded by ocean and is located between longitudes 6°05 E and 6°42 E and latitudes 62°25 N and 62°32 E (Fig. 1). Ålesund has an area of about 633.6 km2 and a population of about 66.148 in 2019 [26].

Fig. 1. Location of the Ålesund city, Norway

Ålesund has a heavily moderated oceanic climate with mild winters. The average annual rainfall and temperature of Ålesund are about 120 mm and 6°C, respectively [27]. The topography of Ålesund comprises a high altitude that gradually decreases from the western and eastern parts to the central part. There are two mountainous regions in the West and the East that have corresponding altitudes of 300 m and 500 m, respectively, and the central part with altitudes of 0–100 m. Ålesund due to its peculiar location is significantly affected by climate change phenomenon, especially sea-level rise and heavy rainfall [28].

82

L. Van Nguyen et al.

2.2 Sensor Placement Analysis Generally, selecting criteria for optimal sensor placement in a pipe-network can be performed based on hydraulic/hydrological simulations, literature review, and experts’ opinion. Some questionnaires can be prepared and sent to the experts to respond. In this research, eight criteria, including Digital Elevation Model (DEM), rainfall, geology, manhole, population density, critical infrastructure, road network, and traffic load, were used (Fig. 2).

Fig. 2. Assessment criteria for determining the location of sensors in the wastewater/stormwater network

Fig. 3. Process for establishing optimal location map

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

83

Traffic load data for the main road was used due to limited traffic data in the study area. The above assessment criteria in Fig. 2 can be changed (increase or decrease) depending on the circumstances, or special requirement of decision markers. The selected criteria were calculated using separate weights and entered as input layers into GIS software, and corresponding optimal location maps were built.

Fig. 4. A map of criteria: (a) Altitude; (b) Rainfall; (c) Geology; (d) Manholes; (e) Population Density; (f) Critical Infrastructures; (g) Road Network, and (h) Traffic Load

84

L. Van Nguyen et al.

The process of building an optimal location map for sensor placement in the wastewater and stormwater network is shown in Fig. 3. Eight criteria were converted to pixels of 1m × 1m and shown in Fig. 4. 2.3 Fuzzy Analytical Hierarchy Process Method The AHP method was proposed and developed by Saaty in 1971–1975 [29] and was used as a useful method to support decision-makers to construct utility functions [14]. Its fundamental theory is to reduce complex decisions through the use of a series of pairwise comparisons [15]. The comparison matrix (C), presented in Eq. (1), was built using a 1–9 scale that is shown in Table 1. Table 1. The Saaty’s AHP scale Important level

AHP Scale FAHP Scale

Equal importance

1

(1, 1, 1)

Moderate importance

3

(2, 3, 4)

Strong importance

5

(4, 5, 6)

Very strong importance

7

(6, 7, 8)

Extremely strong importance 9

(9, 9, 9)

Intermediate values

2

(1, 2, 3)

4

(3, 4, 5)

6

(5, 6, 7)

8

(7, 8, 9)

⎡

1

C12 C13 ⎢ 1 1 C 23 ⎢ C12 ⎢ 1 1 ⎢C C 1 ⎢ 13 23 C=⎢ ⎢... ... ... ⎢ 1 1 1 ⎢ C1k C2k C3k ⎢ ⎣... ... ... 1 1 1 C1n C2n C3n

. . . C1k . . . C2k . . . C3k ... ... ... 1 ... ... . . . C1kn

⎤ . . . C1n . . . C2n ⎥ ⎥ ⎥ . . . C3n ⎥ ⎥ ... ... ⎥ ⎥ . . . C3n ⎥ ⎥ ⎥ ... ... ⎦ ... 1

(1)

n×n

where n is the total number of criteria, and Ckn (1 ≤ k ≤ n), (1 ≤ Ckn ≤ 9) is the point according to the Saaty’s AHP scale. As mentioned above, the assessment criteria using the AHP method does not take into account the uncertainty associated with the process [21] and is significantly affected by uncertainty, biases, and the vagueness of the experts’ opinions. In order to solve this problem, the FAHP method, which involves a combination of the fuzzy set theory proposed by Zadeh [24], and the basic AHP method was employed. The FAHP methodology

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

85

was used in this study due to its accuracy, flexibility, and robustness [18]. There was a total of six steps to determine the weight for criteria using the FAHP method: Step 1. Determine the fuzzy comparison (FC) matrix from a pairwise comparison (C) matrix using the FAHP scale in Table 1 as Eq. (2): ⎡

. . . (C1k − 1, C1k , C1k + 1) ⎢ ... ... ⎢ ⎢ 1 , 1 , 1 ... (1, 1, 1) FC = ⎢ C +1 C C −1 ⎢ 1k 1k 1k ⎢ ... ... ⎣ ...  1 , 1 , 1 1 , 1 , 1 . . . C1n +1 C1n C1n −1 Ckn +1 Ckn Ckn −1

 = (FCijk−1 , FCijk , FCijk=1 ) (1, 1, 1) ...

⎤ . . . (C1n − 1, C1n , C1n + 1) ⎥ ... ... ⎥ ⎥ . . . (Ckn − 1, Ckn , Ckn + 1) ⎥ ⎥ ⎥ ... ... ⎦ ... (1, 1, 1)

n×n

(2)

In Eq. (2), i, j, k = 1, 2, . . . , n, and FC matrix was defined as follows: ⎫ ⎧ (1, 1, 1); i = jorCi = 1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎬ ⎨ ⎪  = 9 9, 9); C (9, i k−1 k+1 k   = FC ij , FC ij , FC ij Cij − 1, Cij , Cij + 1 ; 1 ≤ i < j ≤ n ⎪, k = 1, 2, . . . , n ⎪ ⎪ ⎪ ⎪ ⎪ 1 1 1 ⎭ ⎩ ; n ≥ i > j ≥ 1 , , Cij +1 Cij Cij −1 (3) Step 2. Calculate fuzzy geometric mean (FGM) values as explained by Buckley, Feuring and Hayashi [30] by using Eq. (4): ⎞⎤ ⎡⎛  n 1  n 1  n 1 n n n    ⎟⎥ ⎢⎜ FGM = ⎣⎝ FC k−1 , FC kij , FC k+1 ⎠⎦ ij ij m=1

m=1

m=1

= [(FGT m , FGM m , FGP m )]n×1 n×1

(4) Step 3. Compute Consistency Ratio (RC) from Consistency Index (CI), Random Index (RI), and Principal Eigenvalue (λ) as Eqs. (5–7) [31]: CI RI

(5)

λ−n n−1

(6)

RC = CI =

1 FGM i C × n n j=1 FGM j n

=

(7)

i=1

where n is the total number of criteria. RC index was below 10% indicated that the comparison matrix provided by the expert is consistent [31]. RI values obtained from a randomly generated pairwise comparison matrix [32] as Table 2. Step 4. Compute intermediate values (M) as Eq. (8) and arrange them ascending order, and the organized results were presented in a vector, called AV as Eq. (9): ! 1 1 1 (8) , n , n M = n i=1 FGT i i=1 FGM i i=1 FGP i

86

L. Van Nguyen et al. Table 2. Satty’s RI values [32] n

1 2 3

4

5

6

7

8

9

10

RI 0 0 0.52 0.89 1.11 1.25 1.35 1.40 1.45 1.49

AV = (AV min , AV mean , AV max )

(9)

where AV min = min(M ), AV max = max(M ) and AV mean is the remaining value. Step 5. Compute parameters l, m and u (l ≤ m ≤ u) that indicate the smallest, the promising and the largest value of the triangular fuzzy number [18, 33] in Fig. 5.   FGM ik−1 FGM ki FGM ik+1 , , (10) (li , mi , ui )n×1 = AV min AV mean AV max n×1

Fig. 5. Triangular fuzzy number

Step 6. Compute non-fuzzy weights (A) and the normalized weight (w) of each criterion as Eq. (11): " Ai = li +m3i +ui , i = 1, 2, . . . , n (11) wi = nAi A k=1

k

2.4 Suitability Maps for Placement of Flood Monitoring Sensor in Wastewater/Stormwater Networks The assessment criteria (C1 to C8) were entered as separate layers and were reclassified using ArcGIS Pro software. The purpose of the reclassification was to assign a point rank from 1 to 5 (1 for the least important and 5 for the most essential) for elements

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

87

in each criterion based on spatial relationships between them and the main criteria. For example, in criterion C1, the element’s points were assigned depending on their heights. With criterion C3 (geology), because there was only one geological type in the study area, this layer was assigned as point rank 5. For criterion C5 (population density), uninhabited areas were assigned point rank 1. Based on this rule, important elements were assigned higher points. In the classification stage, the Euclidean Distance tool in ArcGIS Pro was used to calculate spatial distance from two points in each criterion layer; after that, the point of each element layer was assigned by using the Reclassify tool. In the stage of calculating the Euclidean distance, we had to determine the distance to assign point rank. The Raster Calculator tool in ArcGIS Pro was used in this stage. The sensor location suitability maps were created based on five different situations (including one for equal weight and four for each experts’ opinions). They were scaled from 1 to 4 corresponding to four categories: 1 - “No/Very Low Suitability”, 2 - “Low Suitability”, 3 - “Moderate Suitability”, and 4 - “High Suitability”.

3 Results and Discussion 3.1 Collecting AHP Questionnaires and Calculating Weights for Each Criterion The results of the Satty’s scale based on the questionnaire survey are presented in Appendix A. Table 3 shows the Consistency Ratio (CR) values calculated from the survey and reveals a consistency in experts’ opinions (CR < 10%). The weights assigned to the different criteria are presented in Table 4, with details provided in Appendix B. Table 3. Statistics of the CR index for each expert Expert

λ

Expert 1

8.6187 0.08839 1.4 6.3

Expert 2

8.5761 0.08230

5.9

Expert 3

8.5276 0.07537

5.4

Expert 4

8.6011 0.08587

6.1

CI

RI CR (%)

In this study, one more case is created by assigning the equal weight of all factors. The equal weight case was used in work to identify optimal locations. The weights of an equal case and experts’ opinions are summarized in Table 4. 3.2 Map of Optimal Sensor Placements In this study, the average distance of pipelines in the wastewater/stormwater network is approximately 15 m (Fig. 6). Therefore, the distance of 8 m is used as threshold for assigning point of each criterion. The results of the distance calculation and rank point assignment are shown in Fig. 7

88

L. Van Nguyen et al. Table 4. Statistic of criteria’s weights (%) Criteria Equal Weight Expert 1 Expert 2 Expert 3 Expert 4 C1

12.5

23.3

32.9

25.6

26.2

C2

12.5

32.4

21.2

19.5

27.1

C3

12.5

9.7

21.6

23.6

15.4

C4

12.5

14.7

7.2

7.3

14.7

C5

12.5

5.9

4.9

8.1

4.3

C6

12.5

7.1

3.9

10.0

5.2

C7

12.5

3.8

4.4

3.5

4.5

C8

12.5

3.1

3.9

2.4

2.6

Fig. 6. Pipeline network in the study area

The Weighted Overlay tool in ArcGIS Pro software was used to overlay criterion layers based on their importance. The results are shown in Fig. 8. The results in Fig. 8 showed that criteria C1 (elevation) and C2 (rainfall) were sensitive factors for the placement of sensors in wastewater and stormwater pipe network. The high and very high suitability areas (in Figs. 8a–8d) had low altitude (the areas with the green color in Fig. 4a). This was reasonable because almost all the drainage pipes in the study area are gravity-based, and the water in the pipes tend to flow from high to low altitudes. Areas of low elevation also tend to accumulate more stormwater than areas with high elevations during flooding events. In terms of rainfall, the figures from 8a to 8d showed the southwest region of the study area was significantly influenced than others. It could be explained that the rainfall in the southwestern part is higher, and this area could be easily flooded than the other

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

89

Fig. 7. Reclassified maps for criteria: (a) DEM; (b) Rainfall; (c) Geology; (d) Manhole; (e) Population Density (PD); (f) Critical Infrastructure (CIC); (g) Road Network (RN); (h) Traffic Load (TL)

areas (Fig. 4b). The next important criteria were C3 (geology) and C4 (manholes) (Table 4). The result in Fig. 8 showed manholes that were in the southwestern part and along with the roads were important than others. The next important criteria were C7 (road network) and C8 (traffic load), Fig. 8 showed that the areas around the main roads, were

90

L. Van Nguyen et al.

Fig. 8. Suitability map for placement of flood monitoring sensor: (a) Expert 1; (b) Expert 2; (c) Expert 3; (d) Expert 4; and (e) Equal weight

highly suitable for sensor replacement. The pipelines near the roads could be affected more than the pipes far away from the roads, and the installation of sensors in these pipes can provide an early warning failure. The criterion traffic load had the least weight because there was only one main road (with traffic load data) was used in this study. Fig. 8 gave the overview of sensor location distribution; we provided the percentage of the total land area corresponding to each scale in Ålesund, Norway, in Fig. 9. From Fig. 8 and Fig. 9, it could be seen that areas for “Moderate Suitability” and “High Suitability” locations calculated from four expert’s opinion was more significantly high than using equal weightage for each criterion. The equal weightage method was less reliable than others because it did not concern the variables’ importance.

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

91

Fig. 9. The percentage of total land area corresponding to each scale

4 Conclusion This paper proposed a new approach based on an integration of the fuzzy AHP method with GIS to identify potential areas for the placement of sensors for pipe networks for wastewater and stormwater monitoring, with a case study of in Ålesund, Norway. In this regard, eight factors (altitude, rainfall, geology, manholes, population density, critical infrastructures, road network, and traffic load) were used. Based on the results, some conclusions are below: • The integration of the fuzzy AHP method with GIS could help to determine initial areas for the placement of flood monitoring sensors in wastewater and stormwater networks. • The manhole, road-related criteria, altitude, and rainfall are important factors for identifying optimal sensor placements in the wastewater and stormwater pipe. • A major limitation of this study is the use of only spatial criteria for identifying the suitable areas for sensor placements. Critical criteria such as physical-related features of the pipelines (diameter, status, flow, etc.,), hydraulics/hydrology, or type of used sensors have not been considered. Besides, the weight of each criterion was derived by experts’ opinions, therefore, may be somewhat subjective. • Future studies should consider integrating the fuzzy AHP with hydrological and hydraulic models in order to build more detailed and reliable maps for optimal sensor placement.

Acknowledgments. I would like to thank the Norwegian Climate Service Center, the Norwegian Water Resources and Energy Directorate, the Weather Atlas, and the Mapping Authority for providing data for this research. This research was funded by the Smart Water Project, Project number 90392200. The data analysis and write-up thesis were operated as a part of the first author’s Ph.D. studies at the

92

L. Van Nguyen et al.

Department of Ocean Operations and Civil Engineering, Norwegian University of Science and Technology, Norway.

Appendix Appendix a. Result of AHP Questionnaire of Experts Expert 1. Criteria

C1

C2

C3

C4

C5

C6

C7

C8

C1

(1,1,1)

(1,1,1)

(3,4,5)

(1,2,3)

(4,5,6)

(1,2,3)

(3,4,5)

(6,7,8)

C2

(1,1,1)

(1,1,1)

(5,6,7)

(1,2,3)

(9,9,9)

(4,5,6)

(7,8,9)

(7,8,9)

C3

(1/5,1/4,1/3)

(1/7,1/6,1/5)

(1,1,1)

(1,1,1)

(2,3,4)

(1,1,1)

(3,4,5)

(3,4,5)

C4

(1/3,1/2,1)

(1/3,1/2,1)

(1,1,1)

(1,1,1)

(2,3,4)

(1,2,3)

(3,4,5)

(5,6,7)

C5

(1/6,1/5,1/4)

(1/9,1/9,1/9)

(1/4,1/3,1/2)

(1/4,1/3,1/2)

(1,1,1)

(2,3,4)

(1,2,3)

(1,2,3)

C6

(1/3,1/2,1)

(1/6,1/5,1/4)

(1,1,1)

(1/3,1/2,1)

(1/4,1/3,1/2)

(1,1,1)

(2,3,4)

(1,2,3)

C7

(1/5,1/4,1/3)

(1/9,1/8,1/7)

(1/5,1/4,1/3)

(1/5,1/4,1/3)

(1/3,1/2,1)

(1/4,1/3,1/2)

(1,1,1)

(1,2,3)

C8

(1/8,1/7,1/6)

(1/9,1/8,1/7)

(1/5,1/4,1/3)

(1/7,1/6,1/5)

(1/3,1/2,1)

(1/3,1/2,1)

(1/3,1/2,1)

(1,1,1)

Expert 2. Criteria

C1

C2

C3

C4

C5

C6

C7

C8

C1

(1,1,1)

(2,3,4)

(1,2,3)

(3,4,5)

(3,4,5)

(6,7,8)

(7,8,9)

(7,8,9)

C2

(1/4,1/3,1/2)

(1,1,1)

(1,1,1)

(3,4,5)

(2,3,4)

(5,6,7)

(6,7,8)

(6,7,8)

C3

(1/3,1/2,1)

(1,1,1)

(1,1,1)

(2,3,4)

(7,8,9)

(4,5,6)

(6,7,8)

(2,3,4)

C4

(1/5,1/4,1/3)

(1/5,1/4,1/3)

(1/4,1/3,1/2)

(1,1,1)

(1,1,1)

(1,2,3)

(1,1,1)

(4,5,6)

C5

(1/5,1/4,1/3)

(1/4,1/3,1/2)

(1/9,1/8,1/7)

(1,1,1)

(1,1,1)

(1,1,1)

(1,1,1)

(1,1,1)

C6

(1/8,1/7,1/6)

(1/7,1/6,1/5)

(1/6,1/5,1/4)

(1/3,1/2,1)

(1,1,1)

(1,1,1)

(1,1,1)

(1/3,1/2,1)

C7

(1/9,1/8,1/7)

(1/8,1/7,1/6)

(1/8,1/7,1/6)

(1,1,1)

(1,1,1)

(1,1,1)

(1,1,1)

(1,2,3)

C8

(1/9,1/8,1/7)

(1/8,1/7,1/6)

(1/4,1/3,1/2)

(1/6,1/5,1/4)

(1,1,1)

(1,2,3)

(1/3,1/2,1)

(1,1,1)

Expert 3. Criteria

C1

C2

C3

C4

C5

C6

C7

C8

C1

(1,1,1)

(1,1,1)

(1,2,3)

(2,3,4)

(4,5,6)

(3,4,5)

(4,5,6)

(6,7,8)

C2

(1,1,1)

(1,1,1)

(1,1,1)

(2,3,4)

(1,2,3)

(1,2,3)

(5,6,7)

(6,7,8)

C3

(1/3,1/2,1)

(1,1,1)

(1,1,1)

(3,4,5)

(4,5,6)

(2,3,4)

(6,7,8)

(9,9,9)

C4

(1/4,1/3,1/2)

(1/4,1/3,1/2)

(1/5,1/4,1/3)

(1,1,1)

(1/4,1/3,1/2)

(1,1,1)

(4,5,6)

(2,3,4)

C5

(1/6,1/5,1/4)

(1/3,1/2,1)

(1/6,1/5,1/4)

(2,3,4)

(1,1,1)

(1/3,1/2,1)

(1,2,3)

(3,4,5)

C6

(1/5,1/4,1/3)

(1/3,1/2,1)

(1/4,1/3,1/2)

(1,1,1)

(1,2,3)

(1,1,1)

(2,3,4)

(5,6,7)

C7

(1/6,1/5,1/4)

(1/7,1/6,1/5)

(1/8,1/7,1/6)

(1/6,1/5,1/4)

(1/3,1/2,1)

(1/4,1/3,1/2)

(1,1,1)

(1,2,3)

C8

(1/8,1/7,1/6)

(1/8,1/7,1/6)

(1/9,1/9,1/9)

(1/4,1/3,1/2)

(1/5,1/4,1/3)

(1/7,1/6,1/5)

(1/3,1/2,1)

(1,1,1)

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

93

Expert 4. Criteria

C1

C2

C3

C4

C5

C6

C7

C8

C1

(1,1,1)

(1,2,3)

(1,2,3)

(1,1,1)

(4,5,6)

(6,7,8)

(6,7,8)

(6,7,8)

C2

(1/3,1/2,1)

(1,1,1)

(1,2,3)

(2,3,4)

(5,6,7)

(7,8,9)

(5,6,7)

(7,8,9)

C3

(1/3,1/2,1)

(1/3,1/2,1)

(1,1,1)

(1,1,1)

(4,5,6)

(4,5,6)

(1,2,3)

(4,5,6)

C4

(1,1,1)

(1/4,1/3,1/2)

(1,1,1)

(1,1,1)

(3,4,5)

(2,3,4)

(1,2,3)

(7,8,9)

C5

(1/6,1/5,1/4)

(1/7,1/6,1/5)

(1/6,1/5,1/4)

(1/5,1/4,1/3)

(1,1,1)

(1,1,1)

(1,1,1)

(1,2,3)

C6

(1/8,1/7,1/6)

(1/9,1/8,1/7)

(1/6,1/5,1/4)

(1/4,1/3,1/2)

(1,1,1)

(1,1,1)

(3,4,5)

(2,3,4)

C7

(1/8,1/7,1/6)

(1/7,1/6,1/5)

(1/3,1/2,1)

(1/3,1/2,1)

(1,1,1)

(1/5,1/4,1/3)

(1,1,1)

(1,2,3)

C8

(1/8,1/7,1/6)

(1/9,1/8,1/7)

(1/6,1/5,1/4)

(1/9,1/8,1/7)

(1/3,1/2,1)

(1/4,1/3,1/2)

(1/3,1/2,1)

(1,1,1)

Appendix B. The Intermediate Values and Weights Calculated from Experts Expert 1. Criteria l

m

u

A

w (%)

C1

0.14335 0.23676 0.36589 0.24867 23.3

C2

0.22791 0.33332 0.47815 0.34646 32.4

C3

0.06737 0.09844 0.14527 0.10369

C4

0.0851

C5

0.03384 0.05814 0.09779 0.06326

5.9

C6

0.04078 0.06769 0.12055 0.07634

7.1

C7

0.0223

3.8

C8

0.01822 0.02829 0.0512

9.7

0.14128 0.24383 0.15674 14.7

0.03608 0.06261 0.04033 0.03257

3.1

Expert 2. Criteria l

m

u

A

w (%)

C1

0.21212 0.33225 0.49516 0.34651 32.9

C2

0.14624 0.21187 0.30908 0.22239 21.2

C3

0.14286 0.2137

C4

0.04857 0.07196 0.10806 0.0762

7.2

C5

0.03795 0.04949 0.06773 0.05172

4.9

C6

0.02667 0.03773 0.0594

0.04127

3.9

C7

0.03282 0.04526 0.0621

0.04673

4.4

C8

0.02493 0.03773 0.05991 0.04086

3.9

0.32629 0.22762 21.6

94

L. Van Nguyen et al.

Expert 3. Criteria l

m

u

A

w (%)

C1

0.16446 0.26107 0.39617 0.2739

25.6

C2

0.12396 0.20029 0.30287 0.20904 19.5

C3

0.15865 0.23627 0.36329 0.25274 23.6

C4

0.04685 0.07189 0.117

0.07858

7.3

C5

0.04513 0.07698 0.13802 0.08671

8.1

C6

0.05939 0.0968

C7

0.02104 0.03362 0.05689 0.03718

3.5

C8

0.01557 0.02309 0.03768 0.02545

2.4

0.16273 0.10631 10.0

Expert 4. Criteria l

m

u

A

w (%)

C1

0.16623 0.26849 0.40967 0.28146 26.2

C2

0.16506 0.26875 0.43848 0.29076 27.1

C3

0.09123 0.14924 0.25625 0.16557 15.4

C4

0.09578 0.14968 0.22969 0.15838 14.7

C5

0.02925 0.04363 0.06597 0.04628

4.3

C6

0.03517 0.05233 0.08018 0.05589

5.2

C7

0.02682 0.04301 0.07457 0.04814

4.5

C8

0.01623 0.02487 0.04323 0.02811

2.6

Appendix C. AHP Questionnaire Template Circle one number per row below using the scale: 1 = Equal 3 = Moderate 5 = Strong 7 = Very strong 9 = Extremely strong 1

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Rainfall (C2)

2

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Geology (C3)

3

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Slope (C4)

4

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Population Density (C5)

5

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Critical Infrastructure (C6)

6

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network (C7)

8

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

9

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Geology (C3)

10

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Slope (C4)

11

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Population Density (C5) (continued)

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors

95

(continued) Circle one number per row below using the scale: 1 = Equal 3 = Moderate 5 = Strong 7 = Very strong 9 = Extremely strong 1

DEM (C1)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Rainfall (C2)

12

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Critical Infrastructure (C6)

13

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network (C7)

15

Rainfall (C2)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

16

Geology (C3)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Slope (C4)

17

Geology (C3)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Population Density (C5)

18

Geology (C3)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Critical Infrastructure (C6)

19

Geology (C3)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network (C7)

21

Geology (C3)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

22

Slope (C4)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Population Density (C5)

23

Slope (C4)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Critical Infrastructure (C6)

24

Slope (C4)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network (C7)

26

Slope (C4)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

27

Population Density (C5)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Critical Infrastructure (C6)

28

Population Density (C5)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network (C7)

30

Population Density (C5)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

31

Critical 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Road Network Infrastructure (C6) (C7)

33

Critical 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8) Infrastructure (C6)

34

Road Network (C7)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Traffic Load (C8)

References 1. Harvey, R.: Predictive and spatial analytics for planning inspections of linear sewer infrastructure. Int. J. Environ. Prot. 4, 48–57 (2014) 2. Ugarelli, R., Venkatesh, G., Brattebø, H., Di Federico, V., Sægrov, S.: Historical analysis of blockages in wastewater pipelines in Oslo and diagnosis of causative pipeline characteristics. Urban Water J. 7, 335–343 (2010) 3. Ugarelli, R., Venkatesh, G., Brattebø, H., Federico, V.D., Sægrov, S.: Asset management for urban wastewater pipeline networks. J. Infrastruct. Syst. 16, 112–121 (2010) 4. Chen, C., Liu, Y., Sun, X., Cairano-Gilfedder, C.D., Titmus, S.: Automobile maintenance prediction using deep learning with GIS data. Procedia CIRP 81, 447–452 (2019)

96

L. Van Nguyen et al.

5. Khorshidi, M.S., Nikoo, M.R., Sadegh, M.: Optimal and objective placement of sensors in water distribution systems using information theory. Water Res. 143, 218–228 (2018) 6. Sahoo, S.R., Yin, X., Liu, J.: Optimal sensor placement for agro-hydrological systems. AIChE J. 65, e16795 (2019) 7. Sahoo, K., Mani, S., Das, L., Bettinger, P.: GIS-based assessment of sustainable crop residues for optimal siting of biogas plants. Biomass Bioenerg. 110, 63–74 (2018) 8. Ahmad, N.F.A., Kaamin, M., Razali, S., Sahat, S.: Geographical information system (GIS) application for flood prediction at Sungai Sembrong (2017) 9. Boulos, P.F., Schade, T.: A GIS-centric approach for optimal sensor placement in large-scale water distribution systems. In: World Environmental and Water Resources Congress 2007, pp. 1–13 (2007) 10. Soroush, F., Abedini, M.J.: Optimal selection of number and location of pressure sensors in water distribution systems using geostatistical tools coupled with genetic algorithm. J. Hydroinf. 21, 1030–1047 (2019) 11. Winter, C.D., Palleti, V.R., Worm, D., Kooij, R.: Optimal placement of imperfect water quality sensors in water distribution networks. Comput. Chem. Eng. 121, 200–211 (2019) 12. Banik, B.K., Alfonso, L., Di Cristo, C., Leopardi, A.: Greedy algorithms for sensor location in sewer systems. Water 9, 856 (2017) 13. Sambito, M., Di Cristo, C., Freni, G., Leopardi, A.: Optimal water quality sensor positioning in urban drainage systems for illicit intrusion identification. J. Hydroinf. 22, 46–60 (2019) 14. Ishizaka, A., Nemery, P.: Multi-Criteria Decision Analysis: Methods and Software. Wiley, Hoboken (2013) 15. Alami Merrouni, A., Elwali Elalaoui, F., Mezrhab, A., Mezrhab, A., Ghennioui, A.: Large scale PV sites selection by combining GIS and analytical hierarchy process. Case study: eastern morocco. Renew. Energy 119, 863–873 (2018) 16. Singh, L.K., Jha, M.K., Chowdary, V.M.: Multi-criteria analysis and GIS modeling for identifying prospective water harvesting and artificial recharge sites for sustainable water supply. J. Cleaner Prod. 142, 1436–1456 (2017) 17. Gigovi´c, L., Pamuˇcar, D., Božani´c, D., Ljubojevi´c, S.: Application of the GIS-DANP-MABAC multi-criteria model for selecting the location of wind farms: a case study of Vojvodina, Serbia. Renew. Energy 103, 501–521 (2017) 18. Ali, S., Lee, S.M., Jang, C.M.: Determination of the most optimal on-shore wind farm site location using a GIS-MCDM methodology: evaluating the case of South Korea. Energies 10, 2072 (2017) 19. Al Garni, H.Z., Awasthi, A.: Solar PV power plant site selection using a GIS-AHP based approach with application in Saudi Arabia. Appl. Energy 206, 1225–1240 (2017) 20. Suh, J., Brownson, J.R.S.: Solar farm suitability using geographic information system fuzzy sets and analytic hierarchy processes: case study of Ulleung island, Korea. Energies 9, 648 (2016) 21. Noorollahi, E., Fadai, D., Akbarpour Shirazi, M., Ghodsipour, S.H.: Land suitability analysis for solar farms exploitation using gis and Fuzzy Analytic Hierarchy Process (FAHP)—a case study of Iran. Energies 9, 643 (2016) 22. Adham, A., Sayl, K.N., Abed, R., Abdeladhim, M.A., Wesseling, J.G., Riksen, M., Fleskens, L., Karim, U., Ritsema, C.J.: A GIS-based approach for identifying potential sites for harvesting rainwater in the Western Desert of Iraq. Int. Soil Water Conserv. Res. 6, 297–304 (2018) 23. Singhai, A., Das, S., Kadam, A.K., Shukla, J.P., Bundela, D.S., Kalashetty, M.: GIS-based multi-criteria approach for identification of rainwater harvesting zones in upper Betwa subbasin of Madhya Pradesh, India. Environ. Dev. Sustain. 21, 777–797 (2019) 24. Zadeh, L.A.: Fuzzy sets. Inf. Control 8, 338–353 (1965)

Identification of Sensitive Factors for Placement of Flood Monitoring Sensors 25. 26. 27. 28.

29. 30. 31.

32. 33.

97

https://en.wikipedia.org/wiki/%C3%85lesund https://www.ssb.no/kommunefakta/ https://www.weather-atlas.com/en/norway/alesund-climate#rainfall Simpson, M., Øie Nilsen, J.E., Ravndal, O., Breili, K., Sande, H., Kierulf, H., Steffen, H., Jansen, E., Carson, M., Vestol, O.: Sea Level Change for Norway: Past and Present Observations and Projections to 2100 (2016) Saaty, R.W.: The analytic hierarchy process—what it is and how it is used. Math. Modell. 9, 161–176 (1987) Buckley, J.J., Feuring, T., Hayashi, Y.: Fuzzy hierarchical analysis revisited. Eur. J. Oper. Res. 129, 48–64 (2001) Nefeslioglu, H.A., Sezer, E.A., Gokceoglu, C., Ayas, Z.: A Modified Analytical Hierarchy Process (M-AHP) approach for decision support systems in natural hazard assessments. Comput. Geosci. 59, 1–8 (2013) Chen, C.F.: Applying the Analytical Hierarchy Process (AHP) approach to convention site selection. J. Travel Res. 45, 167–174 (2006) Wang, C.N., Nguyen, V.T., Thai, H.T.N., Duong, D.H.: Multi-criteria decision making (mcdm) approaches for solar power plant location selection in Viet Nam. Energies 11, 1504 (2018)

Evaluating the Service Quality of the First Bus Rapid Transit Corridor in Hanoi City and Policy Implications Minh Hieu Nguyen(B) Faculty of Transport Economics, University of Transport and Communications, Hanoi, Vietnam [email protected], [email protected]

Abstract. Bus Rapid Transit (BRT) is a concept born in the 1970s and developed successfully in many developing countries because of its advantages (e.g., lower required investment, more flexibility, faster establishment, and much larger capacity) over traditional bus services. However, the first BRT system in Hanoi (Vietnam), which runs between the Kim Ma terminal and the Yen Nghia terminal, is flawed evidence with a daily ridership at approximately 13,500 passengers. This patronage is only comparable with that of the conventional bus route. This study aimed to find underlying factors determining the service quality to propose policy implications for improving the Hanoi BRT. The data used were the responses of 243 BRT passengers gathered in a customer satisfaction survey conducted in March 2017. The method was an adapted version of the well-known SERVQUAL model with the use of Exploratory Factor Analysis. As regards the findings, the four factors found were Quickness and Timeliness (beta = 0.278), Safety and Reliability (beta = 0.249), Access (beta = 0.217), and Comfort (beta = 0.188). They explained 62.5% of the variance of passengers’ satisfaction. Among the eight policy solutions proposed to enhance the four factors mentioned above, the three most important and feasible ones were the implementations of the prioritized traffic signal for BRT, busway enforcement, and automatic pre-board ticket collection. Keywords: Bus rapid transit · Public transport · Sustainable development · Servqual method · Service quality · Hanoi

1 Introduction With the ability to provide a fast, green, safe, affordable and punctual service [1, 2], BRT systems have achieved prominence as an effective method for the urban mass rapid transit [3]. The world has witnessed a BRT wave in many developing countries and some developed countries [1] for the last two decades. According to [4], the total constructed length of the BRT was 3000 km in 2005, which was four times as large as that in 1995. The length reached 4500 km in 2010, before hitting the top of about 5500 km in 2015. Up © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 98–123, 2021. https://doi.org/10.1007/978-3-030-60269-7_6

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

99

to now, a total of 172 cities in the world have established BRT systems for the operation, and 126 other cities have planned to build or are constructing the BRT corridors [4]. The BRT concept can be attributed to the existence of some first dedicated bus lanes in Chicago in 1939 and Washington DC in 1969 [5]. The cradle of modern BRT that was Curitiba (Brazil) in 1974 has inspired other cities worldwide [6]. Subsequently, the most well-known systems introduced include Quito Metrobus-Q in 1995, Bogotá TransMilenio in 2000, and Guangzhou in 2010. Among them, the success of TransMilenio has received the attention of the world community as the state of the art version, which formed the basis for the first Asian BRT system in Jakarta, Indonesia, in 2004 [3]. Notably, the replication of the BRT concept varies across cities since there is no precise definition of what constitutes a BRT system [8]. Simply, BRT can be defined through its name. Specifically, BRT is the system emphasizing priority for and rapid movement by bus on the securing segregated busways [7] to provide a rail-like service [8]. To put it another way, BRT would be a train on rubber tires. One of the extensively cited definitions is introduced in TCRP Report 90 [9], in which BRT is defined as “a flexible, rubber-tired rapid-transit mode that combines stations, vehicles, services, running ways, and Intelligent Transportation System (ITS) elements into an integrated system with a strong positive identity that evokes a unique image.”. A recent way of explanation concentrates on the benefits and components of systems (i.e., service and infrastructure) introduced by the Institute for Transportation and Development Policy. This organization defines BRT as a high-quality bus-based transit systems delivering fast, comfortable, and cost-effective service at metro-level capacities on the dedicated lanes and iconic stations with off-board fare collection along with fast and frequent operations [10]. For obtaining the qualities of rail transit and the flexibility of buses [11], the BRT system is benefited by a range of features, namely physical infrastructure, operation, business and institutional structure, technology, marketing, and customer service [1]. Similarly, [9] notes seven components, that is, runways, stations, vehicles, fare collection, service, route structure, and intelligent transportation system. In the spectrum of tirebased public transport, the minimum standard for a BRT system (i.e., BRT-lite) consists of essential features, including some form of bus priority, some sections of full segregated busways, improved travel times, higher quality shelters, clean vehicle, and marketing identity [1]. Citizens in Hanoi are suffering from poor traveling conditions day by day, coming from the limited infrastructure system, the uncontrolled proliferation of private vehicles, predominantly motorcycles, and the low capacity of public transport [12]. The bus is the only one public transport mode, a demand-served portion of which was at a low level of approximately 10% in 2016 and has kept falling until now to reach around 8.5% in 2019 [13]. As a result of envisaging the present transport problems, improving the city’s urban transport network, and strengthen the public transport capacity, the World Bank provides financial support to design and establish the Bus Rapid Transit (BRT) network in Hanoi in May 2007 [14, 15]. After ten years, the first line, connecting between the Kim Ma terminal and the Yen Nghia terminal, was in the official operation on the first day of 2017. It was expected to drive the public’s attention and attract more commuters through an excellent public transport service, thus alleviating the severe congestion, adverse pollution as well as improving the city’s image.

100

M. H. Nguyen

Nevertheless, the Hanoi BRT has failed to meet the expectations above, with the ridership of only 13,500 passengers per day on average [15]. This result is insignificantly larger than the number of passengers transported by a conventional route. Consequently, it is considered a flawed case among a host of successful systems over the world [16]. To the best of my knowledge, there are three peer-review English publications concerning the Hanoi BRT. They focused on analyzing barriers to the establishment of the Hanoi BRT and its drawbacks in order to propose solutions to enhance the system performance [15–17]. The first one is a review of five BRT systems in the Global South, which fail to perform or expand. The Hanoi BRT was indicated as a new but unsuccessful system compared with the corridors in Lagos (Nigeria) and Lima (Peru) [17]. An in-depth analysis of barriers to the establishment of the Hanoi BRT is presented in [16]. The authors deployed a framework with seven aspects (i.e., (1) institutional and legislative context, (2) political leadership and commitment, (3) physical design and operation, (4) management of competing modes, (5) adequate funding, (6) public participation, and (7) image promotion) to explain why the Hanoi BRT could not be a successful case. Recently, the authors of [15] made an interesting contribution. Through applying the BRT standards, they found that the Hanoi BRT achieved the Bronze standard in terms of design but it only obtained the basic level in the practical operation due to the main limitations associated with a low design capacity, a low frequency, a limited speed and lack of reliability along with convenience. All of the mentioned-above studies were based on the views of planners and operators rather than passengers. This should be a big gap because there are existing gaps between the opinions of customers and the thought of providers [18]. If a customer does not appreciate the quality of the service offered, it is likely for him/her not to use this service again and spread negative word of mouth. Solutions proposed based on the opinions of operators and/or planners, therefore, would not be compatible with expectations of passengers, whose evaluations of the quality of service received are subjective [19]. More importantly, in the context of a budgetconstrained developing country like Vietnam, implementing all of the possible measures is unaffordable. Instead of this, choosing some of the adequate measures to enhance the system economically is desired. To do it, knowing the most important factors affecting how passengers perceive the BRT service is the key. For filling the gaps presented above, this study aims at proposing adequate policy implications for improving the Hanoi BRT based on passengers’ views on the service quality by carrying out an analysis of factors influencing service quality. The rest of this paper is structured as follows. Section 2 reviews the existing research on analyzing customer satisfaction together with service quality in public transport. Afterward, Sect. 3 is an introduction to the Hanoi BRT along with data collection. Subsequently, Sect. 4 describes the proposed method of this study. Next, descriptive results and factors related to service quality of the Hanoi BRT are presented and discussed rigorously in Sect. 5. Section 6 provides recommendations for improving the Hanoi BRT based on factors of service quality found. Finally, Sect. 7 concludes this paper.

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

101

2 Background of the Method Used 2.1 Service Quality and Customer Satisfaction There is no universal definition of customer satisfaction, securing a position in social science. It is defined as the feeling of pleasure or disappointment when a customer compares a product’s perceived performance with his or her prior expectations [20, 21]. In this way, it is a psychological comparison between expectations and perceptions. Determining the satisfaction levels based on personal feelings among individual groups is challenging. The definition stated in [22] is agreed by many researchers. The authors defined that “Customer satisfaction is identified by a response (cognitive or affective) that pertains to a particular focus (i.e., purchase experience and/or the associated product) and occurs at a certain time (post-purchase, post-consumption).” [22]. The above notion of “by a response” is in line with the idea of [23], who declares that the simplest way to know how customers feel and what they want is to ask them through carrying out a customer satisfaction survey (CSS). Then, customer feedback can be converted into measurable quantitative data. Customer satisfaction is close to service quality. The authors of [24] conceptualize that quality is the totality of features and characteristics of a service that bears on its ability to satisfy needs. In recent years, an array of studies on the relationship between them also have been conducted in different sectors (e.g., bank, mobile, and library). The common result is that service quality has a significant impact on customer satisfaction. It is in line with the comment that service quality plays a crucial role in determining customer satisfaction [19], or customer satisfaction is based on the level of service quality [25]. As can be seen in Fig. 1, re-drawn based on [19, 26], customer satisfaction determines the loyalty of customers, while service quality has significant impacts on the level of satisfaction. Therefore, increasing service quality will make customers keep on buying the company’s service. Service quality is defined as the customer-based difference between the expectations about the service performance and the perceptions about the service provided. To measure service quality, the SERVQUAL (SERVice QUALity) model, created in [26], has been the most preferred. In SERVQUAL, the quality is evaluated through five

Reliability Responsive -ness

Expected service

Assurance Empathy Tangibles

Perceived service

Product quality Service quality

Situational factor

Customer satisfaction

Price Personal factor

Fig. 1. Service quality and customer satisfaction

Loyalty

102

M. H. Nguyen

dimensions, that is, reliability, responsiveness, assurance, empathy, and tangibility. It is important to note that SERVQUAL is intended to evaluate both technical and functional quality [26]. While functional properties of almost all service industries are similar, technical ones of different industries are dissimilar, leading to the need for adapting SERVQUAL for a specific application. 2.2 Assessing the Quality of Public Transport Service The review of [18] emphasized that SERVQUAL has been most widely applied in the public transport industry. Bus and rail sectors have witnessed a number of SERVQUALbased applications to measure passengers’ satisfaction/quality. Generally, almost all of the research began with the basic version of SERVQUAL and then made some modifications to seek the most important factors influencing service quality. Several adapted versions of SERVQUAL are QUALBUS (QUALity of BUS), RAILQUAL (RAILway QUALity), and P-TRANSQUAL (Public TRANSport QUALity). In the bus sector, some researchers found the relationship between service quality and basic dimensions (tangibles, responsiveness, reliability, assurance, and empathy) [27, 28], while others created the modified versions to extend the original dimensions [29, 30]. In [30], for example, the extension of the generic SERVQUAL dimensions was shown with the addition of “culture”. New approaches were also available in the analysis. The authors of [31] utilized the structural equation model to explore how service quality attributes impacted bus user satisfaction. Among the three factors responsible for the satisfaction, the strongest was “service planning and reliability”, which was also related to frequency and reliability. The importance of two other variables (i.e., network design, comfort, and other factors) were similar and small. In [32], a similar analysis for Johannesburg was carried out to extract four quality dimensions of service. Interestingly, comfort and service were negative attributes compared to positive others (i.e., reliability and safety). Comfort and safety were also two of three dimensions in analysis for Malaysian bus transit, the remainder was a familiar component (accessibility) [33]. In an examination highlighting behavioral intentions, thanks to the deployment of an original measurement scale, two dimensions apart from recognized factors in former studies were explored, that is, core transit services (i.e., frequency, coverage, information) and physical environment (i.e., cleanliness, safety and stability) [34]. These findings were supported by the project in [35]. According to their results, dimensions of physical environment along with functions and convenience were confirmed as underlying factors to customer’ quality assessment [35]. Some in-depth studies have issued a description of dimensions that are close to service aspects. The authors of [36] depicted dimensions of vehicle maintenance, off-board facilitates, information provision, etc. In a more advanced way, service aspects can be looked at as objective indicators in comparison with the subjective judgment of passengers. The observed indicators associated with service performance included route characteristics, service characteristics, service reliability, comfort, cleanliness, fare, information, safety and security, personnel, customer services, and environmental protection [37]. The rail sector is the context of many studies on customer satisfaction. The author of [38] confirmed the key role of service quality to customer satisfaction in the Iran’s rail system. Attributes of rail user satisfaction were quite similar review described above.

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

103

Passenger satisfaction was the dependable variable on five observed ones (i.e., refreshments, information system, behavioral factors, basic facilities, safety, and security) [39]. In [40], the authors supported the dimensions of security, staff’s behavior, information and emphasized the importance of waiting time in case of Bangladesh. Among factors, frequency was consistently considered as the most crucial factor, while reliability, waiting time, access, and service plan served as the main contributors to both rail and bus passenger satisfaction [41]. Since the vast majority of the BRT systems have been a relatively new public transport means compared with conventional bus, the number of studies on BRT has been fewer than the figure for bus’s studies. Among research on BRT passenger’s satisfaction, [42] presented an application of SERVQUAL for the BRT Tehran. They found four dimensions – service, speed, driver behavior, and ergonomics – to be important in the users’ assessment of perceived satisfaction. With the case of Cape Town, also based on the SERVQUAL instrument, five basic dimensions were scrutinized before stated as the culprits for low commuter satisfaction [43]. In recent research conducted in New York, service quality, reflected by frequency, on-time performance, and speed, affected radically to customer satisfaction. Other visible attributes were bus-only lane, bus with three doors, bus comfort proximity of bus stops, real-time information, limited stops, and ticket system [44]. To summarize, service quality pertains closely to passenger satisfaction. Although they are not entirely interchangeable, it is widely accepted that service quality determinants are also customer satisfaction attributes. The noticed point is that service quality could be affected by technical, visible aspects such as vehicles, stations, fare collection but should be measured subjectively by an array of factors, such as reliability, comfort, and empathy. Consequently, titles of attributes discovered in different studies vary across cases of study; however, their meanings would be homogeneous to some extent. To look at factors, SERVQUAL is one of the most useful models but it is necessary to make some modifications based on the context of service considered. While knowledge on factors of service quality in bus and rail sectors are relatively rich, understandings of attributes of BRT service quality are much more limited.

3 Study Area and Data 3.1 The Hanoi Bus Rapid Transit Hanoi, the capital of Vietnam, was home to 7.59 million in the area of 3,328.9 Km2 [45], leading it to be the largest city. This city has an unbalanced mode share with the domination of motorcycles, which deteriorates traffic problems such as congestion, accidents, and pollution [46]. In contrast to the private vehicle proliferation, public transport, including bus only, is much less preferable with the decreasing share of under 9% [47, 48]. The BRT route, connecting between the Kim Ma terminal and the Yen Nghia terminal (Fig. 2), commenced at the beginning of 2017 under the financial aid of the World Bank. It is one of the pragmatic solutions to promote travel conditions in Hanoi. The corridor stretches 14.7 km cross five districts (Ba Dinh, Dong Da, Thanh Xuan, Nam Tu Liem, and Ha Dong). There is one 3.5-m-exclusive-lane per direction which is converted from

104

M. H. Nguyen

Fig. 2. Description of the BRT route (Source: created by the author)

a mixed lane and located at the median of the road along most of the corridor (Table 1). The physical barriers between the BRT lane and neighboring ones are established at several short segments; hence other vehicles can enter the BRT lanes illegally (Fig. 3). No priority signal control is provided at intersections for the BRT buses. On the route, there are median 21 stations and two terminals with Kim Ma being the control center while Yen Nghia taking as the place of the depot and the office of the BRT company. Ten standardized stations are equipped with overpasses without escalators or elevators. Among the remainder located near intersections and connected by passages, two offer the priority signal for the pedestrian. The Hanoi BRT utilizes the non-articulated vehicles with a length of 12 m and with capacity being similar to that of large conventional buses. All vehicles are decorated impressively with the green color and symbol of the Hanoi BRT, making them prominent on the road. Manual off-board fare collection is available at the stations while the real-time system and electronic signs are absent. As regards service features, the time span is 17 h per day, from 5h00 till 22h00. The peak interval is 5 min, while headways during normal and off-peak periods are 10 and 15 min, respectively. The system has an average ridership of roughly 13,500 passengers per day. The operating speed is about 21.5 km/h. As a result of governmental subsidy, the fare is flat of about 30 cents per one-way trip. An inter-route monthly ticket, which is valid for both conventional bus and BRT system, is available at the level of 8 euros for non-prioritized people. The senior and the students benefit from a 50% discount. 3.2 Data Primary data were gathered through a CSS. It, done from March 01 to March 10, 2017, to collect responses of the BRT passengers, employed a questionnaire consisting of three parts: (1) administrative, (2) classification, and (3) target questions.

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

105

Fig. 3. Description of the Hanoi BRT’ infrastructure and vehicle (Source: Taken by the author at the Thanh Cong station)

• The first part required a surveyor to fill information on the spatial and temporal background of an interview. • Subsequently, data on the interviewee’s demographics and BRT usage are collected in the classification part. Among the ten questions, the number of ones related to using BRT was five. • The target group, which was formed based on an adjusted SERVQUAL model, was the most important in the questionnaire. The basic SERVQUAL scale was chosen as the initial instrument. Based on the author’s understanding of the BRT conceptualization and the consultation of some researchers at University of Transport and Communications (Vietnam), some modifications were implemented. These changes were mainly involved in enhancing the reliability and familiarity of questions, allowing surveyors to avoid further explanation, thus reducing survey duration. For five traditional dimensions (i.e., Reliability, Assurance, Empathy, Responsiveness, Tangibles), the numbers of statements were five, three, five, three, four, respectively. Two question groups added were access (5 items) and ticket (3 items), which were documented as dimensions of satisfaction and quality on a fairly regular basis in previous studies [33, 41, 49–53]. The last part (Satisfaction) encompassed three items. Passengers’ points of view on each positive statement was presented through five ordered response levels ranging from 1 to 5 and corresponding to: strongly disagree, disagree, neither agree nor disagree, agree, strongly agree, respectively. This rating scale has been well known as a five-level Likert [54].

106

M. H. Nguyen Table 1. Operating parameters of the Hanoi BRT No Indicator (unit) 1 Corridor length (km)

Value 14.7

2 Segregated length (km)

14.0

3 Number of pedestrian overpasses

10

4 Number of stations

21

5 Vehicle capacity (places)

90

6 Number of total fleets/ operating ones (vehicles)

26/22

7 Average ridership per vehicle trip (passengers/trip) 40 8 Average daily ridership (passengers/day)

13,500

9 Average operating (commercial) speed (Km/hour)

21.5

10 Interval at peak time (minutes)

5

11 Interval at normal time (minutes)

10

12 Interval at off-peak time (minutes)

15

13 Service span (hours/day)

18

14 Opening time

5h00

15 Closing time

22h00

The duration of an interview has direct implications on the quality of data collected. The short BRT interval at from 5 to 10 min and a passenger’s unknown onboard time may impose the risk of an incomplete questionnaire. For limiting this issue, some principles and guidelines were applied as follows: • Applying convenience sampling and two collection options. The first was the face to face technique where a surveyor spoke out loud and ticked the passenger’s answers, or a participant filled the questionnaire by herself or himself. • Recommending every interviewer to begin the survey at a station and board with his or her interviewee to complete the survey. • Printing the questionnaire used as an answer sheet in a two-side paper to minimize the risk of losing a particular page in the survey process. Introducing the two-page instrument to participants was mandatory for all of the surveyors. • Distributing surveyors on the spatial and temporal balance. There was no concentration on some specific stations and particular periods. • Requesting a 15-min-attendance in advance of an interview. Passengers could calculate the time at a station and onboard to decide whether participating or not. An interview was undertaken only after a respondent’s consent. When asked, the surveyors should clarify queries immediately and carefully. • Offering one BRT ticket and one colorful bus network map in Hanoi to each participant at the beginning and the end of the interview, respectively.

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

107

For gaining a comprehensive analysis, strict requirements for primary data collected were imposed. Only samples with all of the questions completed were accepted for further analyses. After the process of data collection, among a total of 275 questionnaires spread, 32 ones were rejected as passengers failed to respond at least one question or chose more than one option for a statement. The size of the dataset was 243, and the successful response rate was 88.4%. The average duration of one survey, where a passenger ticked answers, was about 10 min while an oral interview took a longer time at about 14 min. Out of 243 candidates fitting the standards, 172 (58.4%) came from the interviews beginning at a station and ending on board. Approximately 40% (98 passengers) recorded their answers by themselves.

4 Proposed Methodology for This Research Data, which had filtered to remove the samples with incomplete responses to all of the questions, were considered. The analysis process encompassed three main steps. • First, the reliability of the original factors was tested by the Cronbach’s alpha coefficients. • Subsequently, exploratory factor analysis was implemented for all of the attitudinal statements that had passed the test of the Cronbach’s alpha. The results of this step were new and underlying factors determining of the service quality. • Next, for estimating the importance of the found factors to service satisfaction, a regression model was developed. The author will explain such steps in more detail when showing and discussing their results in Sect. 5.

5 Results and Discussions 5.1 Descriptive Analysis Table 2 provides the characteristics of the samples. • Gender: The smaller proportion of males (41.8%) compared with females (58.2%) questioned used BRT. • Occupation: Students were dominant at 43.5%, followed by the comparable figures for the retired and the officers at 20.6% and 18.3%, respectively. A small minority of respondents were those seeking work or doing housework. • Educational level: A significant proportion of the sample (63.4%) was under-graduate and graduate students. Compared to the student cohort, the high school group and the post-graduate group accounted for the considerably smaller percentages at less than 15% for each. • Income level: There was an inverse trend between income groups and their percentages. Passengers earning less than 250 euros per month was roughly two-thirds whereas the highest income group (above 350 euros/month) made up only 12.6%, the smallest percentage.

108

M. H. Nguyen

• Age: BRT was preferred by the youngest and the oldest pools. 40.4% of those questioned were less than 24 yr old, followed by the above 55 age group with the figure of just over 25%. The characteristics showed that the BRT was chosen dominantly by the young, the majority of whom were students with low income. Table 2. Breakdown of sample Variable

Statistics

1. Gender

Male: 41.8%; Female: 58.2%

2. Occupation

Student: 43.5%; Officer: 18.3%; Business: 3.8%; Worker: 9.2%; Homemaker: 2.3%; Seeking work: 1.5%; Retired: 20.6%; Others: 0.8%

3. Education

High school: 14.5%; Student and graduate: 63.4%; Post-graduate: 10.7%; Others: 11.5%

4. Income (Euros/month)

Under 150: 35.8%; 150-under 250: 28.4%; 250–350: 24.2%; Above 350: 12.6%

5. Age group

Under 24: 40.4%; 24-under 35: 11.7%; 35–55: 12.0%; Above 55: 25.9%

6. Alternative mode of trip

Bicycle: 1.6%; Motorcycle: 31.2%; Car: 0.8%; Bus: 64.0%; Others: 2.4%

7. Trip purpose

Work: 22.4%; Education: 25.2%; Visit: 20.0%; Shopping: 1.6%; Back home: 22.0%; Others: 8.8%

8. Ticket types

One-way ticket: 34.4%; Monthly ticket: 64.8%; Others: 0.8%

9. Frequency of BRT use

Very frequently (above 5 days/week): 31.2%; Frequently (3 – 5 days/week): 16.8%; Not often (1 – 2 days/week): 26.4%; Very not often: 25.6%

10. Multimodal

BRT + Walking: 55.6%; BRT + Bus + Walking: 30.6%; BRT + Motorbike + Walking: 4.8%; Others: 9.9%

• Trip purpose: Regardless of trips back home, the three most common trip purposes were education (25.2%), work (22.4%), and visit (20%). The highest rate for the educational purpose was in line with the largest number of BRT passengers being students. The high percentage for visit meant that BRT would be a favorable choice for tourism. A tiny percentage of respondents (1.6%) chose BRT for shopping. • Ticket: Most of the respondents possessed a monthly ticket, which was available for all bus routes and BRT. In the remainder, over a quarter of people purchased one-way tickets. • Frequency of BRT use: Frequent passengers using at least three days per week accounted for nearly 50% of the sample. Among them, the figure for heavily BRT

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

109

utilizers (more than 5 days during a week) was the largest at 31.2%. The figure at slightly more than half was for infrequent BRT users. • Modal shift: The main source of BRT ridership was bus passengers, while its attraction to motorists was marginal. 64% of those surveyed chose BRT as a replacement for the bus. This figure was double the percentage of the motorcycle (31.2%). The low rate for modal shift from a bicycle was predicted because of the bicycle’s extremely limited usage in Hanoi [55]. There was an interesting comparison between Jakarta and Hanoi about the modal shift when the first BRT systems were in operation. In a survey conducted by the Japan International Cooperation Agency (JICA) in the first month of TransJakarta BRT system, similar to Hanoi, the largest shifting in Jakarta came from the bus. The following mode was private cars (14%) compared to motorcycle (31.2%) in Hanoi. This difference may be explained that Hanoi is the motorcycle dependent city, in contrast to Jakarta where car is the most popular private mode. • Multimodal: BRT as a standalone mode was used by the largest proportion (55.6%), while 30% of respondents transferred between bus and BRT for their trips. The minority of sample (just under 5%) employed the combination of BRT with a private mode like motorbikes. So, the attraction of BRT was not comparable to that of the motorcycle. And the BRT system’s performance may depend largely on the existing bus network’s operation. 5.2 Factors Associated with Quality of the Hanoi BRT Service As noted above, a three-step approach was implemented to find and analyze factors affecting service quality. First, all items and all priori dimensions underwent the Cronbach’s alpha test, which is a numerical coefficient concerning the precision of the measurement. Cronbach’s alpha values of priori dimensions and their items ranged from 0.712 to 0.875 (see Table 3 for values of dimensions), exceeding the level of acceptable reliability [56]. Thus, they supported internal consistency among items and dimensions. Table 3. Cronbach’s alpha values of dimensions Original dimensions Values Reliability

0.823

Empathy

0.782

Assurance

0.734

Responsiveness

0.819

Tangibles

0.805

Access

0.775

Ticket

0.748

Second, the Exploratory Factor Analysis (EFA), which is a method of data reduction, was performed in order to identify the underlying unobservable constructs, which

110

M. H. Nguyen

are potential attributes of service quality. The EFA utilized the Principal Component Analysis with Varimax rotation to recognize how an identified construct correlated with each of others. Eigenvalue was used for determining the number of extracted factors. It represented the amount of variance explained by each factor. A factor with an eigenvalue of just under 1 was not better than a single variable, thus only constructs with eigenvalues greater than 1 were accepted. The Kaiser Meyer Olkin measure of sampling adequacy (KMO) and Bartlett’s test of sphericity were run. The former result varied between 0 and 1. Its values closer to 1 were better. The minimum value acceptable was suggested at 0.6. The latter tested the null hypothesis that the correlation matrix was an identity one. It was based on the Chi-Square test, whose value was proportional to the possibility of eliminating the null hypothesis. The outcome of this step was four extracted factors, which explained 70.3% of the variance in the data. The KMO test had the value of a meritorious level (see Table 4). Four explored constructs were labelled: Quickness – Timeliness (% of variance = 31.8%), Safety – Reliability (% of variance = 20.6%), Comfort (% of variance = 13.1%), and Access (% of variance = 4.8%). The final step is the regression analysis in order to explore the relationship between customer satisfaction and extracted factors. Applying the EFA technique for responses of the three statements on passenger satisfaction. One construct was found and named Satisfaction (SA). In the regression analysis, this variable (SA) was the dependent one while predictors were four aforesaid factors. A beta coefficient of a factor represented its magnitude to the dependent variable. Therefore, it was the comparable index on the importance of attributes (predictors) to the dependent variable (SA). Table 5 reveals that Quickness and Timeliness was the most important attribute while comfort had the weakest impact. Safety and Reliability, Access secured the second and third position, respectively. The model explained 62.5% of the variance of the satisfaction construct. The variance inflation factor (VIF) ranging from 1.083 to 1.541 was at the acceptable level recommended by [57]. This is compatible with the fact that tolerance used as an indicator of multicollinearity was higher than the recommended minimum value of 0.1 in [58]. 5.3 Discussions on Factors of the BRT Service Quality Among four constructs, there were two double ones, including Quickness – Timeliness and Safety – Reliability while Comfort and Access were single dimensions. 5.3.1 Quickness – Timeliness of Service This is a double dimension whose name has been hardly appeared in the previous articles. The quickness implies how fast the BRT service runs. The quickness is calculated by time or speed. In general, speed is the continuously varied parameter during the trip. Thus it is challenging for commuters to measure it. On the contrary, time can be reckoned straightforwardly by the difference between the ending time and the starting time. Hence, the quickness dimension should be collected through time comparison. On the other hand, the BRT system is established newly in the traffic of urban areas. Therefore, BRT users are riders of other means. Asking passengers about the visible comparison

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

111

Table 4. Results of exploratory factor analysis No

Statement on BRT service

Original dimension

Factor loadings

I

The first attribute (factor): Quickness and timeliness

I.1

BRT’s speed is significantly higher than that of the conventional bus

Reliability

0.864

0.209

I.2

Waiting time for BRT bus is small

Empathy

0.832

0.211

I.3

Frequent BRT service is offered at off-peak time

Empathy

0.805

I.4

Off-board ticket can be bought quickly

Ticket

0.732

I.5

Incidents are introduced to passengers immediately

Reliability

0.703

I.6

Service frequency during a day is arranged appropriately

Empathy

0.695

0.228

I.7

The number of BRT trips per day is large

Empathy

0.646

0.338

II

The second attribute (factor): Safety and reliability

II.1

BRT service operation fits timetable

Reliability

0.230

0.855

II.2

BRT service fulfills Reliability introduced standards

0.311

0.814

II.3

BRT passenger does Assurance not worry about their belongings at stations and onboard

II.4

BRT service is stable Reliability

0.721

II.5

BRT drivers obey the Responsiveness road rule strictly

0.689

II.6

Passengers do not feel tired after a BRT-used trip

Assurance

0.662

II.7

BRT bus stops at right place to match stations

Responsiveness

0.621

F1

F2

F3

F4

0.274

0.796

0.388

0.304

0.324

(continued)

112

M. H. Nguyen Table 4. (continued)

No

Statement on BRT service

Original dimension

Factor loadings F1

F2

II.8

Amenities (e.g., overpasses, ramps…) are sufficient quality

Tangibles

0.356

0.604

III

The third attribute (factor): Access

III.1

Access to the station at rush hours is possible

Access

III.2

Platform station fits with the floorboard of BRT bus

Access

III.3

To access the station, Access BRT passenger does not walk for a long distance

III.4

Information on BRT service is found easily

III.5

Priority and Empathy assistance are offered for special passengers (e.g., pregnant women, the senior, the disabled)

0.301

0.702

III.6

Ticket integration between BRT and conventional bus is convenience

Ticket

0.438

0.687

III.7

BRT passenger does not wait for a long time at passage (to cross a road)

Access

III.8

BRT passenger does not miss a BRT trip at the peak time

Access

0.612

III.9

The price of ticket is affordable

Ticket

0.578

IV

The fourth attribute (factor): Comfort

IV.1

BRT buses and stations are clean

Tangibles

IV.2

BRT bus is driven smoothly

Responsiveness

0.271

0.293

F3

0.832

0.799

0.391

Assurance

0.756

0.734

0.293

F4

0.389

0.633

0.305

0.339

0.782 0.210

0.721

(continued)

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

113

Table 4. (continued) No

Statement on BRT service

Original dimension

IV.3

Equipment on BRT vehicles and stations are modern

Tangibles

IV.4

Equipment both in station and on board operates well

Tangibles

Factor loadings F1

F2

F3

F4 0.634

0.293

0.534

Extraction method: PCA; Rotation method: Varimax with Keiser Normalization KMO: 0.849; Chi-Square of Bartlett’s: 200.878; Sig.: 0.000

Table 5. Results regression analysis Factor

Standardized coefficients

t

Sig

Beta

Collinearity Statistics Tolerance

VIF

Constant

-

Quickness and timeliness

0.278

7.636

0.000

0.923

1.083

Safety and reliability

0.249

5.702

0.000

0.649

1.541

Access

0.217

5.492

0.000

0.801

1.248

Comfort

0.188

4.527

0.000

0.736

1.359

Adjusted R Square: 62.5%

between BRT and previous modes is a good way to assess the quickness. The first candidate group suitable for comparison may be bus because BRT is the advancement of the conventional bus with higher speed (item I.11 ). The following group comprises of riders by private vehicles such as cars and motorbikes. BRT can take considerable advantages if the comparison was implemented at peak hours. Although not indicated directly, the quickness dimension is presented as the speed factor [44]. Notably, speed is also one of four determinants of BRT passenger satisfaction in Tehran [42]. Timeliness collects statements which are connected with the availability of service to fulfill travel demand. Long waiting time or no available service deteriorates customer satisfaction, thus service quality. Timeliness covers such issues as frequency, service span (item I.2, item I.3, item I.6, and item I.7). Higher frequency is equivalent to a shorter waiting time of the passenger at the station. Frequency is a sensitive subject

1 For the content of item, please see Table 4. For example, item I.1 of Quickness-Timeliness

should have No = “I.1” and Statement on BRT service = “BRT’s speed is significantly higher than that of the conventional bus”.

114

M. H. Nguyen

during the off-peak period. Due to the economic effectiveness, BRT operators employ a longer interval, which results in a higher standing time at the station. Besides, the service span has direct implications on timeliness. It determines the potential markets that transit serves [41]. Operators often pay attention to the number of operating hours per day while passengers concern more about the beginning and the closing time. Passengers tend to be more subject to BRT service closed soon because other mode choices in the evening would be costly, even unavailable. Timeliness here is in line with the results produced by the authors of [37], who show that service frequency and daily service time are indicators of service characteristics and affect quality. The timeliness is also witnessed through the time taken to buy tickets (item I.4) and notifying unexpected changes to passengers (item I.5), which escapes passengers from confusion and anxiety unnecessarily. Theoretically, giving information on incidents should be an item of reliability [26]. However, timeliness, in this study, concentrates on the moment of introduction. The sooner the notification, the better the service. In a short way, the construct Quickness-Timeliness connects with how much time a passenger spends for a BRT trip. Quickness relates to the onboard process while timeliness associates procedures before catching a BRT bus. 5.3.2 Safety – Reliability of Service This factor comprises two sub-attributes, which are widely considered as the core factors to customer satisfaction. Reliability is defined as the ability to eliminate or minimize the differences between promised parameters and practical performance. The level of reliability is inversely proportional to the differences. Reliability can be reflected apparently in the match between the planned timetable and the BRT operation in the spatial and temporal contexts (item II.1). Specifically, not arriving at a station on time or running cross the station causes adverse effects on the reliability of service. In view of the customer, the compatibility should be manifested in the broader background, where all announcement from BRT operators has to be respected and implemented (item II.2). In earlier research, reliability has been appeared as either an independent dimension [27, 28, 30, 32, 39, 51] or a sub-attribute [31, 53]. Reliability is also indicated as the dimension of punctuality [59] or predictability by means of using the schedule. Concerning the safety, it involves the possibility of passengers’ health and belongings being risked at stations and/or on board (item II.3, item II.6). Safety probably relates to the security of others, which describes the passengers’ the public responsibility. In this way, although being safe, passengers possibly blame BRT service for the shortage of safety due to damage in accidents between a BRT bus and other modes. This agrees with the calculation of indicators of Safety attribute undertaken in [37]. Competence and carefulness of driver (item II.5, item II.7) along with the total of road accidents are in close-knit connection with Safety and Security dimension [37]. The quality of infrastructure is also considered as one observation of safety (item II.8). Compared to the reliability, safety is indicated less frequently [37, 51, 53]. Interestingly, [60] emphasized the role of safety to satisfaction and disregarded reliability attribute.

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

115

5.3.3 Access to the Service Access of service pertains to the conditions allowing customers to catch the BRT service. They are underpasses and overpasses, which provides paths to the median stations without waiting time at the pavement. If these facilities are not available, the prioritized signal for pedestrian decides the accessibility to stations (item III.7). Along with waiting time, the walking distance plays a vital role in customer’s judgment about the access (item III.3). Besides, facilities supporting the movement of the special utilizers (e.g., the disabled) are also important to the measurement of access (item III.5). This is highlighted in [60] and in TCRP Report 165 [41]. The size of the station and the capacity of the vehicle affect the accessibility. If a passenger fails to get on station or vehicle, especially at rush hours, this means the service is inaccessible fro him/her (item III.1, item III.8). What’s more, the level of compatibility between the station platform and the BRT bus floor is representative of access (item III.2). Another concept, lying in Access attribute, is the ticket. Undeniably, without a ticket, boarding is unacceptable. In this way, the pricing, the method of purchasing, ticket integration between modes are indicators of access (item III.6, item III.9). This is quite different from the outcomes of scholars who explore that fare (or cost) is a specific dimension [59–61]. Last but not least, information provision is considered as a quality attribute [36], which is in view of how easily information can be obtained (item III.4) in this study. 5.3.4 Comfort of Service The comfort of service is dependable on responses of passengers to the operation status, the cleanliness of facilities, vehicles, shelters, and even new features offered (item IV.1, item IV.3, and item IV.4). For BRT vehicle’s elements, the number and the size of seats, the availability of air conditioning, the standing space, the smoothness of movement (item IV.2) are regarded as indicators of the comfort. For stations, amenities included benches, shelters, lighting, vending facilities, air conditioning, informational signing and etc. are responsible for passenger’s attitude on the comfort. Simply, comfort, from the customer’s view, is the evaluation of experiencing improved standards for both vehicles and stations. In research on bus user’s satisfaction, the comfort is frequently extracted as an important determinant [32, 37, 62], sometimes with the title of cleanliness [59] or a sub-attribute (comfort and convenience of service [53]). Because of the discrepancy in defining attributes and background of the survey, a comparison should be made with caution. For example, [63] underlines that fare price is the most important determinant with a sample dominated by low-income students, who represent a market segment instead of the population. One of the appropriate candidates for comparing the attributes’ important is [51] analyzing the Hanoi bus system based on an adjusted SERVQUAL version. Similar to this study, in [51], the majority of utilizers are students using service frequently with below-average income. Notably, despite the similarity of descriptive characteristics and context, the most crucial dimension is the cost, which is an item of access, the third most important factor in this study. Speed, an indicator of the strongest attribute (Quickness and timeliness), is insignificantly statistical meaning in [51]. The conversion of customer satisfaction attitude on public transport is believed to originate from the changes in economic and traffic conditions. In [51],

116

M. H. Nguyen Table 6. Recapitulation of factors extracted and their components

Ranking Attribute (Determinant)

Components (Items) of attributes

1

Quickness and timeliness – – – –

Speed; Frequency; Service span; Waiting time for the vehicle, for the ticket, announcing incidents

2

Safety and reliability

– – – – –

Match between actual operation and schedule; Respect to introduced parameters; Security at stations and on the vehicle; Driver’s responsibility; Quality of infrastructure

3

Access

– – – – –

Waiting time at the passage; Distance walking; Priority and support for special passengers; Ticket pricing and integration; Information collection

4

Comfort

– Cleanliness on vehicle and at the station; – Availability of equipment onboard and at the station

Note: The list of components in the third column do not imply the importance of items to attributes

as a result of the continuously growing fuel price, modal shifts are chosen by many motorcyclists. Bus with the local subsidy offering such an affordable price at only about 20 cents are an effective alternative. In this study, a considerable decrease in the bus speed due to regular severe traffic congestion is blamed for the steadily falling bus ridership. This is supported by the result of a recent survey [13]. In the broader view, the aforesaid change of passenger attitude on the role of speed and price ticket is confirmed by [64], maintaining that passengers tend to be more sensitive to changes in time cost rather than ones in fares.

6 Recommendations The goal of proposing policy implications is to improve service quality and customer satisfaction, thus the Hanoi BRT’s performance and ridership. The satisfaction has four attributes explored above; however, categorizing suggestions based on such attributes is inappropriate because a proposal may have multi-effects on different determinants simultaneously. For that reason, the author suggests solutions according to the components of the BRT system highlighted in the literature review and discussing their potential impacts on attribute components listed in Table 6. 6.1 The Segregated Median Busway As mentioned in Sect. 2.1, the segregated median busway is the origin of the state of the art BRT concept. This is the main feature to enhance the speed, frequency thus reducing

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

117

waiting time and increasing reliability. Thanks to priority bus lanes, which boost route punctuality, a 40% increase in patronage over the first two years is seen in the town of Tyne and Wear, UK [65]. The position of the bus lane is a crucial aspect. In Seoul, transfer from curbside bus lane to median ones results in a decrease in the interval of 3–5 min at peak hours [66]. Safety is another great benefit of the busway. The central median busway set up for Macrobús in Guadalajara is reported as the cause for a 35% fall in the number of injury crashes [67]. However, the positive effects of the busway are potentially confined by the illegal entry of other vehicles. In the original design, the Kim Ma – Yen Nghia BRT corridor was protected from other mixed-lanes by lines of the physical barrier. The removal of the barrier system was approved because of the government’s anxiety about the heavy public outcry. At present, running on the lane of BRT is illegal; yet, punishment and prevention are bypassed. A serious collision between a BRT bus and a private car happened on the busway. Motorbikes run ahead BRT buses quite commonly (see Fig. 4). To obtain the effectiveness of the busway, establishing physical segregation is necessary. Also, capturing the photos of vehicles penetrating the BRT lanes through the camera onboard and at intersections are feasible and urgent solutions applied worldwide [66].

Fig. 4. Motorcycles occupy the BRT lane illegally (Source: Taken by the author at the Hoang Dao Thuy station)

118

M. H. Nguyen

6.2 Vehicle and Station Vehicle and station are involved closely with attributes of access and comfort. Service aesthetics (e.g., the availability of seats as well as space, air conditioning, and other equipment such as benches, lightings, fans) influence significantly of comfort. Regular maintenance is a fundamental way to ensure a clean environment and well-functioned devices. The capacity of both station and vehicle play essential roles in accessing BRT service at peak hours and the comfort level. Changes related to infrastructure are problematic; therefore, the infrastructure should be paid attention to in the design process. The typical example is Guangzhou, where the size of the station was chosen based on the forecast passenger volume. Hence the lengths of stations range from 55m to 250m [68]. Vehicle adjustment is feasible, but not a simple task. Regardless of other traffic conditions, the vehicle has to be compatible with the positions of doors, the platform height, and the size of the station. According to [67], safety on the station is taken into consideration because some passengers tend to refuse using amenities (e.g., over or underpass) and/or waiting for the signal for a pedestrian when arriving at stations. For the Hanoi BRT, extensive modifications to the types of vehicles and stations are impossible. The availability of more seats in the station, the provision of air conditioning, and maintaining the good status of equipment like fans, lighting in all stations and terminals are the critical solutions. This is reflected by complaints of the press about the unstable operations of these devices. Along with the sound system introducing the following stop point, displays showing the same information on board is necessary, especially for the deaf. Along with more overhead walks constructed, fences along the ramp to guide the passengers utilize bridges are necessary. Since BRT vehicles are different from the conventional buses, despite the similar length, drivers have to learn and train how to drive the vehicle smoothly, particularly practicing how to stop at the right place, enabling the station platforms to fit well the bus floors. 6.3 Ticket Collection and Service Off-board collection system at station, allowing to reduce the stopping time at stations for boarding, is another distinct feature of BRT. Pre-board collection will be perfect if the integration ticket among public transportation modes is offered through smartcards. In this sense, the modern fare collection can enhance both access and timeliness. For the service pattern, the variety of services on the corridor is appreciated. Yet, it is subject to the restriction of the busway. With only one lane in a direction, dispatching the express service whose BRT buses ignore some station and overcome some other BRT buses is impossible. For the Hanoi corridor, the off-board ticket system is the advantage; however, the manual sales should be eliminated and be superseded by automatic machines at all stations. Free transportation mode change may a need. 6.4 Intelligent Technology System (ITS) Application of ITS technologies includes an automatic vehicle location system, passenger information system, and transit preferential treatment systems at signalized intersections [9]. One of the advantages of ITS is to provide the priority for buses at junctions,

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

119

thus improving both the speed and the comfort because of the limitation of acceleration and deceleration. As a side of ITS, the development of websites providing operation parameters like the timetable, traffic conditions and even the reason for the failure to arrive at the destination on time is widely applied. The presence of displays with updated information at stations is the basis for judgment on reliability and enriching accessibility in terms of information. Besides, creating mobile apps for searching schedules and proposing an approachable set of routes based on inputs like the starting point and the destination helps passengers to make their travel planning well. Table 7. Recommendations for customer satisfaction improvement No

Recommendation

Points for attributes Quickness and timeliness √

√

√

√

1

Prioritized traffic signal for BRT

2

Busway enforcement (Camera, police patrol, physical barrier line)

3

Integrated and automatic pre-board ticket system

4

Features for pedestrian (traffic signal, overhead bridges)

-

5

Development of website and apps

-

6

Driver training

-

7

Displays on board

-

8

Guided-fences at stations

-

√

Safety and reliability

-

Access -

√

√

√

√

√

√ √

Comfort √

-

-

-

√

√

√

-

-

One of the complicated problems of Hanoi is the high density of intersections, prohibiting an improvement of the BRT speed. In this case, the role of ITS becomes more important. Traffic light systems dedicated to BRT at intersections adjacent to the existing lights, which are popular in Seoul, should be concerned. The official website of the Hanoi BRT needs to be re-designed to become more attractive with more updated information and a forum for passengers’ discussion as well as answering queries. A mobile version with the integration between BRT and all conventional bus routes in Hanoi should be provided.

120

M. H. Nguyen

Among the above recommendations, cognizance of the magnitude of service quality’s factors assists in ranking them. Because one has multi impacts, it can receive values from more than one attribute. Establishment of the signal dedicated for BRT to give priority at intersections is the most pressing based on the satisfaction attribute importance (see Table 7). Other vital solutions are to enhance the protection for busway from other vehicle entry and the ticket system.

7 Conclusion This paper presents a passenger-based analysis rigorously to find the main factors of service quality and then based on these factors to make some policy implications for improving the quality of the Hanoi BRT, enabling its better performance with higher ridership. There are four factors responsible for BRT’s passenger satisfaction. The strongest attribute is Reliability and Timeliness, which is related to passenger’s waiting and boarding time. The second is Safety and Reliability, which is involved in the security of passenger’s health and belongings as well as the BRT service providers’ respect to introduced operating parameters. The third determinant is access, which is judged by how passengers can reach the station, ticket, and service. The weakest is comfort, which depends mainly on the cleanliness and availability of equipment and devices in both stations and vehicles. Factors found are discussed and compared with the results of previous studies carefully to highlight both differences and similarities. Based on the magnitude of identified satisfaction attributes, among policy recommendations, prioritized traffic signal and busway protection are the most important. In general, this study may be a good reference for authorities, BRT operators in Hanoi. Moreover, it can be served as a supplement to the treasure of works on service quality and customer satisfaction in the public transport industry. Nevertheless, the interpretation of attribute meanings would be confusing to some extent. The grouping of determinants that are not indicated in observed questions is controversial. This is the limitation of representative, which generally is inherent and common in numerous research on service quality and customer satisfaction. Acknowledgements. The authors would like to give many thanks to Prof. Tien Dieu Bui, the Editor of ISRM2020 for his valuable comments to the first version of this chapter.

References 1. Wright, L., Hook, W.: Bus Rapid Transit Planning Guide. Institute for Transportation and Development Policy, New York (2007) 2. Zhang, X., Liu, Z., Wang, H.: Lessons of bus rapid transit from nine cities in China. Transp. Res. Rec. J. Transp. Res. Board 2394, 45–54 (2013) 3. Ernst, J.: Initiating bus rapid transit in Jakarta, Indonesia. Transp. Res. Rec. J. Transp. Res. Board. 1903, 20–26 (2005) 4. brtdata.org: Evolution of the number of cities per year. https://brtdata.org/panorama/year 5. WSP: Bus Rapid Transit and modal integration. WSP Canada Inc., Canada (2016)

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

121

6. Matsumoto, N.: Analysis of Policy Processes to Introduce Bus Rapid Transit Systems in Asian Cities from the Perspective of Lesson-Drawing: Cases of Jakarta, Seoul, and Beijing. Chapter IV.7, pp. 351–376 (2007) 7. Wright, L., Fulton, L.: Climate change mitigation and transport in developing nations. Transp. Rev. 25, 691–717 (2005) 8. Hsu, L.R.: Cost estimating model for mode choice between light rail and bus rapid transit systems. J. Transp. Eng. 139, 20–29 (2013) 9. Levinson, H., Zimmerman, S., Clinger, J., Rutherford, S., Smith, R., Cracknell, J., Soberman, R.: TCRP report 90: Bus Rapid Transit, Volume 1: Case Studies in Bus Rapid Transit. Transportation Research Board, Washington, D.C. (2003) 10. Institute for Transportation and Development Policy (ITDP): What is BRT?. https://www. itdp.org/library/standards-and-guides/the-bus-rapid-transit-standard/what-is-brt/ 11. Thomas, E.: Bus Rapid Transit. Annual meeting - Institute of Transportation Engineers, Chicago (2001) 12. Huynh, D.T., Gomez-Ibañez, J.: Vietnam. In: Pojani, D., Stead, D. (eds.) The Urban Transport Crisis in Emerging Economies, pp. 267–282. Springer, Cham (2017) 13. Nguyen, T.C.: Quality improvement and bus system development in Hanoi vision to 2025. Improvement of quality and the development of the public transport system by bus in Hanoi capital vision to 2025. Nikko Hotel, Hanoi, Vietnam (2016) 14. Japan International Cooperation Agency [JICA]: Socialist Republic of Vietnam: Data collection survey on BRT in Hanoi (2016) 15. Nguyen, H.N., Tu, S.S., Nguyen, M.H.: Evaluating the maiden BRT corridors in Vietnam. Transp. Commun. Sci. J. 71(4), 336–346 (2020) 16. Nguyen, M.H., Ha, T.T., Tu, S.S., Nguyen, T.C.: Impediments to the bus rapid transit implementation in developing countries – a typical evidence from Hanoi. Int. J. Urban Sci. 4, 464–483 (2019) 17. Nguyen, M.H., Pojani, D.: Chapter Two - Why do some BRT systems in the global south fail to perform or expand? In: Shiftan, Y., Kamargianni, M. (eds.) Preparing for the New Era of Transport Policies: Learning from Experience, pp. 35–61. Academic Press, Cambridge (2018) 18. Ojo, T.K.: Quality of public transport service: an integrative review and research agenda. Transp. Lett. 11, 104–116 (2019) 19. Wilson, A., Zeithaml, V.A., Bitner, M.J., Gremler, D.D.: Services Marketing: Integrating Customer Focus Across the Firm. McGraw Hill, New York (2012) 20. Oliver, R.L.: Measurement and evaluation of satisfaction processes in retail settings. J. Retail. 57, 25–48 (1981) 21. Tse, D.K., Wilton, P.C.: Models of consumer satisfaction formation: an extension. J. Mark. Res. 25, 204–212 (1988) 22. Giese, J.L., Cote, J.A.: Defining consumer satisfaction. Acad. Mark. Sci. 2000, 1–24 (2002) 23. Levy, P.: Growing numbers, more opportunity to spend (2009) 24. Kotler, P., Armstrong, G., Sounders, J., Wong, V.: Principles of Marketing. Prentice Hall, Harlow, England , New York (2002) 25. Lee, H., Lee, Y., Yoo, D.: The determinants of perceived service quality and its relationship with satisfaction. J. Serv. Mark. 14, 217–231 (2000) 26. Parasuraman, A., Zeithaml, V.A., Berry, L.L.: SERVQUAL: a multiple-item scale for measuring consumer perceptions of service quality. J. Retail. 64, 12–37 (1988) 27. Suleman, D., Ojo, T.K., Mireku, D.O., Butsogbodo, R.Y.: Service quality and customer satisfaction of public transport on Cape Coast-Accra Route, Ghana. Dev. Country stud. 4(18), 142–149 (2014)

122

M. H. Nguyen

28. Verma, M., Verma, A., Ajith, P., Sindhe, S.: Urban bus transport service quality and sustainable development: understanding the gaps. Presented at the 13th WCTR, Rio de Janeiro, Brazil, July (2013) 29. Barabino, B., Deiana, E., Tilocca, P.: Measuring service quality in urban bus transport: a modified SERVQUAL approach. Int. J. Qual. Serv. Sci. 4, 238–252 (2012) 30. Randheer, K., AL-Motawa, A.A., Prince Vijay, J.: Measuring commuters’ perception on service quality using SERVQUAL in public transportation. Int. J. Mark. Stud. 3, 21 (2011) 31. Eboli, L., Mazzulla, G.: Service quality attributes affecting customer satisfaction for bus transit. J. Public Transp. 10, 21–34 (2007) 32. Vilakazi, A.M., Govender, K.K.: Exploring public bus service quality in South Africa: a structural equation modelling approach. J. Transp. Supply Chain Manag. 8 (2014). 10 pages 33. Noor, H.M., Nasrudin, N., Foo, J.: Determinants of customer satisfaction of service quality: city bus service in Kota Kinabalu, Malaysia. Procedia Soc. Behav. Sci. 153, 595–605 (2014) 34. Lai, W.T., Chen, C.F.: Behavioral intentions of public transit passengers—the roles of service quality, perceived value, satisfaction and involvement. Transp. Policy 18, 318–325 (2011) 35. Yaya, L.H.P., Fortià, M.F., Canals, C.S., Marimon, F.: Service quality assessment of public transport and the implication role of demographic characteristics. Public Transp. 7, 409–428 (2015) 36. Carreira, R., Patrício, L., Natal Jorge, R., Magee, C.: Understanding the travel experience and its impact on attitudes, emotions and loyalty towards the transportation provider–a quantitative study with mid-distance bus trips. Transp. Policy 31, 35–46 (2014) 37. Eboli, L., Mazzulla, G.: A methodology for evaluating transit service quality based on subjective and objective measures from the passenger’s point of view. Transp. Policy 18, 172–181 (2011) 38. Esmaeili, A.A., Manesh, B.A., Golsham, E.: Service quality, customer satisfaction and customer loyalty in RAJA rail transportation company. Int. Res. J. Appl. Basic Sci. 5, 347–352 (2013) 39. Geetika, Nandan, S.: Determinants of customer satisfaction on service quality: a study of railway platforms in India. J. Public Transp. 13, 6 (2010) 40. Rahaman, K.R., Rahaman, M.A.: Service quality attributes affecting the satisfaction of railway passengers of selective route in southwestern part of Bangladesh. Theor. Empir. Res. Urban Manag. 4, 115–125 (2009) 41. Arup, Texas A&M Transportation Institute, KFH Group, Inc., Parsons Brinckerhoff, Inc., Kittelson & Assoc, Inc.: TCRP Report 165 - Transit Capacity and Quality of Service Manual, 3rd edn., Washington, D.C. (2013) 42. Mahmoudi, S.M., Verdinejad, F., Gholamreza, J., Mughari, A.M.: Analysis and establishment of bus rapid transit (BRT) on customer satisfaction in Tehran. AJBM 4, 2514–2519 (2010) 43. Ugo, P.D.: The bus rapid transit system: a service quality dimension of commuter uptake in Cape Town, South Africa. J. Transp. Supply Chain Manag. 8 (2014). 10 pages 44. Wan, D., Kamga, C., Hao, W., Sugiura, A., Beaton, E.B.: Customer satisfaction with bus rapid transit: a study of New York City select bus service applying structural equation modeling. Public Transp. 8, 497–520 (2016) 45. General Statistics Office: Statistical Yearbook of Vietnam. Statistical Publishing House, Vietnam (2017) 46. Nguyen, M.H., Armoogum, J.: Hierarchical process of travel mode imputation from GPS data in a motorcycle-dependent area. Travel Behav. Soc. 21, 109–120 (2020) 47. Vu, H.T., Ha, T.T.: Developing public transport systems - experience from practical activities of hanoi bus system. J. Transp. Minist. Transp. 11(2016), 80–83 (2016) 48. Nguyen, M.H., Ha, T.T., Le, T.L., Nguyen, T.C.: Challenges to development of bus system evidence from a comparative analysis of surveys in Hanoi. In: Transportation for a Better Life: Mobility and Road Safety Managements, Bangkok, Thailand, pp. 1–10 (2017)

Evaluating the Service Quality of the First Bus Rapid Transit Corridor

123

49. Guirao, B., García-Pastor, A., López-Lambas, M.E.: The importance of service quality attributes in public transportation: narrowing the gap between scientific research and practitioners’ needs. Transp. Policy 49, 68–77 (2016) 50. Morton, C., Caulfield, B., Anable, J.: Customer perceptions of quality of service in public transport: evidence for bus transit in Scotland. Case Stud. Transp. Policy 4, 199–207 (2016) 51. Nguyen, M.H., Nguyen, T.T.L.: Measuring urban bus service quality using Servqual model in Hanoi. In: Proceeding of 8th Atrans Symposium - Transportation for a Better Life: Harnessing Finance for Safety and Equity in Asean Economic Community, Bangkok, Thailand (2015) 52. Redman, L., Friman, M., Gärling, T., Hartig, T.: Quality attributes of public transport that attract car users: a research review. Transp. Policy 25, 119–127 (2013) 53. Vuchic, V.R.: Urban Transit: Operations, Planning, and Economics (2005) 54. Dawes, J.: Do data characteristics change according to the number of scale points used? An experiment using 5 point, 7 point and 10 point scales. Int. J. Market Res. 50, 61–77 (2012) 55. People’s Committee of Hanoi: The transportation plan in Hanoi up to 2030 with a vision towards 2050 (2016) 56. Byrne, B.M.: Structural Equation Modeling With AMOS: Basic Concepts, Applications, and Programming. Psychology Press, Mahwah (2000) 57. Rogerson, P.A.: Statistical Methods for Geography. SAGE Publications Ltd., London, Thousand Oaks (2001) 58. Tabachnick, B.G., Fidell, L.S.: Using Multivariate Statistics. Allyn & Bacon, Boston (2000) 59. de Oña, R., Eboli, L., Mazzulla, G.: Monitoring changes in transit service quality over time. Procedia Soc. Behav. Sci. 111, 974–983 (2014) 60. Stradling, S., Carreno, M., Rye, T., Noble, A.: Passenger perceptions and the ideal urban bus journey experience. Transp. Policy 14, 283–292 (2007) 61. Suman, H.K., Bolia, N.B., Tiwari, G.: Comparing public bus transport service attributes in Delhi and Mumbai: policy implications for improving bus services in Delhi. Transp. Policy 56, 63–74 (2017) 62. de Oña, J., de Oña, R., Eboli, L., Mazzulla, G.: Perceived service quality in bus transit service: a structural equation approach. Transp. Policy 29, 219–226 (2013) 63. Eboli, L., Mazzulla, G.: A stated preference experiment for measuring service quality in public transport. Transp. Plan. Technol. 31, 509–523 (2008) 64. O’Sullivan, A.: Urban Economics, 7th edn. McGraw-Hill Europe, Boston (2009) 65. Hensher, D.A., Mulley, C., Yahya, N.: Passenger experience with quality-enhanced bus service: the tyne and wear ‘superoute’ services. Transportation 37, 239–256 (2010) 66. Agrawal, A., Goldman, T., Hannaford, N.: Shared-use bus priority lanes on city streets: approaches to access and enforcement. J. Public Transp. 16, 2 (2013) 67. EMBARQ: Traffic Safety on Bus Priority Systems | World Resources Institute (2014) 68. Hughes, C., Zhu, X.: Guangzhou, China Bus Rapid Transit: Emission Impact Analysis. ITDP (2011)

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam Using Multi-criteria Analysis Tuan Anh Pham1(B)

and My Van Nguyen2

1 Landscape Architecture Department, Faculty of Architecture and Planning,

National University of Civil Engineering, Hanoi, Vietnam [email protected] 2 Vingarden Joint Stock Company, Hanoi, Vietnam

Abstract. The Vung Tau master plan vision to 2035, which was approved in 2019, aims to build the city become a tourist, financial and commercial service center with green, clean, beautiful, civilized and friendly orientations. In this regards, the urban tree system, especially roadside trees, has a critical role in the above strategies. However, due to the natural and urban conditions, the tree system has been suffering many challenges for sustainable development, such as expanding concrete pavements, narrow spaces for living, lack of water and sea salt wind in the dry season… These challenges become more servere because of climate change phenomena. At the same time, tropical hurricanes becomes stronger and more popular in this area. Moreover, there is not any standard guideline which has been proposed to guide the local government of the city to develop the urban trees in a sustainable manner. The paper firstly reviews historical development and existing conditions of roadside trees system in order to understand the larger context. Secondly, it proposes and verifies an approach based on multi-criteria analysis and hierarchical cluster analysis to find the most adaptable shade plant species planting on the roadside system for the sustainable development strategies in Vung Tau. These species were selected by assessment of the main selection parameters of natural and urban conditions, aesthetics, growth characteristics, and specific features. Finally, the article suggests some specific species to create typical landscape values for Vung Tau city. Keywords: Analytical hierarchy process · Hierarchical cluster analysis · Multi-criteria analysis · Species selection · Roadside tree · Sustainable development · Vung tau

1 Introduction The roadside trees in Vung Tau city began to be planted in the late 19th and early 20th centuries, mainly for the French and the local people living in this area [2]. Due to spontaneous planting, some plant species grow unevenly, even not suitable for local © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 124–148, 2021. https://doi.org/10.1007/978-3-030-60269-7_7

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

125

climate and soil conditions. Street trees are concentrated mainly to grow in inner urban roads; especially in 12 ward and Long Son commune, the roadside trees have only planted along the main roads. In 1999, the number of shade trees in Vung Tau was only about 5,000 trees [2]. In the last 10 years, with the attention of the authorities, the city’s tree system has been mainly invested; therefore, up to now, the number of shade trees on the streets has increased rapidly in Vung Tau. Today, the whole city has about 29,000 roadside trees of 93 species [14]. According to the city’s development strategy, by 2020, Vung Tau will strive for an urban greenery area from 12 to 12.5 m2 per capita [2]. However, this is currently an impossible task for the city when, in fact, the average one is only about 1.45 m2 /person in the inner city; it is about 8.05 m2 /person if including the natural forest areas in Lon and Nho Mountains [14]. On the other hand, street trees are affected heavily by the regional climate in two distinct seasons: 1) Dry season: prolonged hot and sunny that make trees developt hardly and leaves, branches and trunks of trees are stunted and dried. 2) Rainy season: this is a revival period of trees. Vung Tau trees are heavily invested but lack of control leading to the fact that there are too diversity of species and sizes on each road; such as Truong Cong Dinh (21 species), Vo Truong Toan (15 species) Tran Phu (14 species), Le Loi (12 species), Nguyen Trung Truc (10 species), Thong Nhat, Ly Thuong Kiet, Luu Chi Hieu, Nguyen An Ninh and Nguyen Van Cu (9 species) [14]. Even many species that should not be planted on the street also appear quite popular, such as Eucalyptus spp., Ceiba pentandra (L.) Gaertn., Acacia auriculiformis A. Cunn., Muntingia calabura L…. Some species have large trunk size and many buttress roots, such as Ficus benjamina L., Ficus microcarpa L.F., Ficus pilosa L., Ficus religiosa L., Khaya senegalensis (Desr.) A. Juss…., invasive alien trees such as Spathodea campanulata P. Beauv…. The main reason is due to local people planting spontaneous species; many old and unsuitable trees have not been replaced due to the influence of public opinion or planting additional trees has not kept up with the actual needs. Meanwhile, Vung Tau city has no standard guideline for the selection of roadside plants [14]. Roadside trees are one of the essential components of urban green space; they also play an essential role in streetscape [8]. People’s first impression of the city comes for its streetscape [6]. Street trees have a strong relationship with the natural and socioeconomic conditions as well as the cultural characteristics of the city. They are impacted extremely by urban infrastructure and lack of space for growth. Therefore, aiming to select suitable roadside tree species needs to deal with all above-mentioned problems that become a very significant task for sustainable urban development of Vung Tau city. A standard method to quantify expert knowledge for suitability modeling is Analytic Hierarchy Process (AHP), which was proposed by Saaty [17]. There are also several multi-criteria analysis methods which are quantitative analysis methods. They are commonly used to compare and select the optimal option based on the analysis of comparison criteria. [16] These methods can assist landscape architects, planting designers, and urban horticulturists in the plant selection process effectively [1, 15]. In this paper, firstly, the authors assess the current conditions, both natural and urban aspects, of the roadside trees system in Vung Tau city to reveal the challenges as well as potentials for sustainable devlepment of the roadside plant species. Secondarly, literature is reviewed to provide salient features for the methods which can be used in this paper.

126

T. A. Pham and M. V. Nguyen

Thirdly, applied multi-criteria ananlysis methods and process to analyze, evaluate and select adaptive roadside plant species for Vung Tau city are inctroduced. Finally, the paper show the results and discussion as well as give some suggestions for selecting roadside plant species for their sustainable development in Vung Tau city.

2 Assessment the Current Conditions of the Roadside Tree System in Vung Tau City To analyze, evaluate and select adaptive roadside plant species for Vung Tau city, the assessment of the existing conditions of the roadside tree system is critical and necessary. There are 93 shade plant species, which belong to 35 families planting on the roadside system in Vung Tau. However, there is not any method which applied for selecting roadside tree system in Vung Tau. These species are chosen base much on the experience of urban managers and popular urban plant species in Vietnam. Currently, there are about 29,000 trees planted on a total of 209 roads and 52 alleys in Vung Tau. In order to clarify the roles and functions of this tree system in the overall streetscape, firstly, it is necessary to analyze the spatial, natural, and microclimate characteristics of the roads. Since then, they are fundaments for proposing several solutions to improve the greenery system in a close relationship with the architectural landscape space on the roads system as well as meeting to sustainable development strategies of Vung Tau city. Some shade trees grow well, such as Terminalia ctappa L., Arecaceae, Dipterocarpus alatus Roxb. (Fig. 1), Ficus microcarpa L.F., Peltophorum pterocarpum (DC.) K.Heyne., Pithecoloblum saman (Jacq.) Merr., Cassia fistula L., Casuarina equisetifolia L., Hopea odorata Roxb., Terminalia mantaly H.Perrier and Coccoloba uvifera L..

Fig. 1. The Hopea odorata Roxb. Is growing well and evenly on Nam Ky Khoi Nghia Street (Pham 2018).

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

127

In particular, the Hopea odorata Roxb. Is a species accounting for a large proportion in the shade trees system planting on the streets. (Fig. 1) Every year, this species continues to be grown on new and renovated streets, and even old ones [14]. Many streets planted 1 or 2 shade tree species are promoting the values of landscape architecture and the urban ecosystem. (Fig. 2) Typically, the Vo Nguyen Giap a gateway route is: The Hopea odorata Roxb. is intercropped with Cassia fistula L.; they are protected from climate invasion by a thick outside line of Ficus microcarpa L.F. However, presently, there is a huge dispute between these two species due to the planting distance (4m). Moreover, in the future, it will be a beautiful landscape when the Hopea odorata Roxb. Reaches maturity, while the Cassia fistula L. enters a barren stage (after about 20 yr). It becomes a very important corridor in the urban ecosystem. Additionally, shading ability to improve microclimate on the median strips is still very weak due to the lack of big trees. However, as a gateway, it can be a chance to develop a complete solution of growing flower plants (Bougainvillea spectabilis or Plumeria) to emphasize an impressive and featured scene for Vung Tau [14].

Fig. 2. Street has 1 or 2 shade plant species promotes effectively the urban landscape and ecological values on 3–2 Street (Pham 2017).

Vung Tau has mainly sandy soil, which is the main reason leading to the rapid dehydration after irrigation and the poor ability to stabilize the trees during whirlwinds and storms (Fig 3). On November 25th , 2018, the city suffered from typhoon namely No. 9 with high intensity and rarely occurred with the city. According to Vung Tau Urban Management Department, after the storm, there were about 1,991 trees broken branches and collapsed in the whole city (including 334 fallen trees and 1,657 trees with inclined and broken branches). The most collapsed trees were Hopea odorata Roxb. (89), Peltophorum pterocarpum (DC.) K.Heyne. (36), Cassia fistula L. (23), and Khaya senegalensis (22)… This information shows that the quality of shade trees in the city is relatively good, limiting the collapse when exposed to natural disaster risks. Besides, the statistics show that several streets where has the most number of plants needed to be clearance are Hoang Hoa Tham (23), 3/2 (17), Le Hong Phong (15), and Truong Cong Dinh (10). These streets are basically in the direction of Southeast - Northwest

128

T. A. Pham and M. V. Nguyen

and Northeast - Southwest. They are the leading wind routes from the East Sea to the city.

Fig. 3. Sandy soil affects the health of urban trees significantly due to the possibility of rapid dehydration (Trinh 2014).

Otherwise, the extreme climate and sandy soil characteristics have restrained the growth of the system of shrubs and groundcover plants. There are only a few welldeveloped species such as Cordia latifolia, Tecoma stans (L.) Juss. ex Kunt., Bougainvillea spectabilis, Lantana camara L., Plumeria, Melampodium paludosum… These species are useful for the narrow medians strips where the shade trees can not plant and grow (Fig. 4).

Fig. 4. Diversity of the plant species and landscape quality of the median strips but extreme influence by climate and sandy soil on Le Hong Phong Street (Pham 2017).

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

129

As the same as other coastal cities in Vietnam, Casuarina equisetifolia L. has really good resistance to sea salt wind – an extreme natural phenomenon in the dry season in Vung Tau that cruelly impacts the urban tree system (Fig. 5a). This species is widespread and density on the main roads. However, they are not attractive to the streetscape, especially for the Bai Sau beach where is the main and popular swimming place and has many resorts in the city. (Fig. 5b) Pruned casuarina is also popular in this area, but this type is not suitable on the street and does not create any shade for microclimate improvement. They should be removed into the parks. Otherwise, casuarina trees are only beautiful when they are planted into forests and promote the effectiveness in coastal protection forests.

Fig. 5. (a). Leaves of coconut trees are drying extremely due to the sea salt wind in the dry season on the Bai Sau park. (Pham 2018). (b). Casuarina trees can withstand well sea salt wind, but do not promote the landscape role on Le Hong Phong Street (Pham 2017).

Shade trees planting on the road system in the industrial zone have invested and developed well both in species diversity and form. There are some plant species growing well, such as Peltophorum pterocarpum (DC.) K.Heyne. And Pithecoloblum saman (Jacq.) Merr…. (Fig. 6) However, there are some problem remained, such as using Roystonea regia is not suitable and less effects microclimate for region; the large median strips are monotonic in terms of form and diversity of species composition. Planting techniques and distances are not rational, such as too close distance between trees, planting trees far from flush curbs, and growing big trees on narrow sidewalks. (Fig. 7) The distance between planting holes is not consistent. Due to spontaneous planting by the local people, there are several species that do not belong to the list of shade trees (Spondias lakonensis Pierre, Artocarpus heterophyllus,…) are quite popular in new urban areas (results of the ineffective management and interdisciplinary activities). As the overhead wires and underground constructions strongly encroach the same as other big cities in Vietnam, such as Hanoi, Da Nang, Ho Chi Minh City, roadside trees. Trees are growing quite popularly under the power line corridors; therefore, they must be cut their tops to control the height resulting in deviated canopy development, reducing the quality of landscape architecture space, impacting the urban trees’ health and decreasing

130

T. A. Pham and M. V. Nguyen

Fig. 6. Selection of suitable tree species creates a beautiful streetscape in Dong Xuyen Industrial Zone (Pham 2017).

Fig. 7. The Hopea odorata Roxb., big tree species, are planted on narrow sidewalks (2 m in width) on Ben Nom Street that not only is unsuitable and reduces the quality of urban landscape but also has a very strong impact on buildings and infrastructure system as well as seriously affects the health of urban trees due to lack of their living space (Pham 2016).

the role of microclimate improvement and getting high risks of collapsing. (Fig. 8a and Fig. 8b) Especially, this phenomenon affects extremely to the city’s gateway routes such as 2/9, 3/2… The invasion of living space is quite a common impacted on roadside trees. The phenomenon of compressing trees happens both in the air and on the ground along the roads, where has both overhead wires system and underground infrastructure. It becomes

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

131

Fig. 8. (a). Clipping the top of Dipterocarpus alatus Roxb. (big tree species) not only affects the health of trees but also reduces the quality of landscape architecture of the gateway – Vo Nguyen Giap Boulevard seriously. (Pham 2016) (b). Tree has grown with a deviated canopy due to cutting its top and falling to the road for more living space in Dong Xuyen Industrial Zone (Pham 2017).

Fig. 9. The infrastructure system squeezes the growing space of the trees both in the air and on the ground at Nguyen Huu Canh Street (Pham 2019 adpated from Vung Tau Urban Management Department 2017).

more seriously on the roads where has an underground tunnel system when they do not have an appropriated design. The tunnel system reduces the living space great for the trees’ roots, which affects seriously and directly to the health of trees. (Fig. 9) Moreover, due to inadequacies design, construction and operation after constructing, nowadays

132

T. A. Pham and M. V. Nguyen

these tunnels system has not promoted its effects in Vung Tau. It is not only wasting urban properties but also difficult to improve the quality of the streetscape. More seriously, due to the natural conditions, some plant species face to pests and get big problems of peeling and drying after a period of growth, such as Mimusops elengi L., Spathodea campanulate… (Fig. 10a) It reveals that the process of species choosing has not concerned much to the existing conditions as well as plants’ ecological characteristics. On the roads directing to the East Sea, especial at the end of the road near the coast, the trees are much less developed. The main manifestations for this phenomenon are stunted trees, deviated canopies, dried branches and trunks. It is a clear evidence of influences from dry and sea salt wind. However, there are some shade trees species growing well, Terminalia ctappa L., Barringtonia asiatica L. and Coccoloba uvifera L.,… (Fig. 10b) Especially for decorating plant species, Plumeria rubra is unique species which has the ability to grow well and produce beautiful flowers in the annual saline monsoon season. At the same time, the other trees are affected extremely. These are the most species that need replication for Vung Tau city.

Fig. 10. (a). Top of trees were died and their canopies were deviated due to directly impact of sea salt wind on Nguyen An Ninh Street. (Pham 2016). (b). The Coccoloba uvifera L. is growing very well in the dry season when it is extremely impacted by sea salt wind in the corner of Le Hong Phong Street and Thuy Van Street (Pham 2018).

3 Background of the Methods Used 3.1 Analytical Hierarchy Process Arcording to the literature, Analytical hierarchy process (AHP) was developed by Saaty [16, 17]. It is a general theory of measurement [21]. The analytic hierarchy process is a mulitcriteria decision making approach in which factors are arranged in a hierachic structure. It is rigorously concerned with the scaling problem and what sort of number to use, and how to correctly combine the priorities resulting from them. A scale of measurement consists of three elements: A set of objects, a set of numbers, and a mapping of the objects to the numbers. In a standard scale a unit is used to construct the rest of the

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

133

numbers of the scale [18]. There are three main phases in the AHP model methodology, which consist of: structuring the problem; comparative judgments, and priority analysis [9]. The AHP methods is one of the multi criteria decision analysis methods (MCDA) with the aim of evaluating tangible and intangible criteria based both on the judgment of knowledgeable and expert people and on existing measurements and statistics needed to make a decision [20]. The AHP is a method that can be used to establish measures in both the physical and social domains[10, 21]. The AHP as a compensatory method is used most frequently (especially for preference aggregation) [3]. In term of plant species selection, Asgarzadeh (2014) applied AHP method to suggest 60 plant species for Tehran urban landscape using [1]. It proves that the AHP is the a suitable method which can be applied to assess plant species for the roadside concerned all aspects of existing natural and urban conditions, aesthetics, growth characteristics, and specific features. The standard process requires firstly the identification of a set of alternatives and a hierarchy of evaluation criteria (value tree), followed by pairwise comparisons to evaluate alternatives’ performance on criteria (scoring) and criteria among themselves (weighting) [4]. All the weights/alternatives are compared in respect to the criteria by using a priority on a scale from 1 to 9, with 1 indicating equal importance and 9 extreme importance [18, 19]. Intermediate values are used to express increasing preference/performance for one weight/alternative [4]. The resulting output of this procedure is a matrix of comparisons expressed as ratios, and the next step is the reduction of the pairwise comparison matrix to a set of scores representing the relative importance of each weight and performance of alternatives (priority vectors) [4]. Once the criteria weights and alternatives scores have been derived with the described process, overall performance of the alternative can be calculated by means of a linear additive model [4]. The final result is a value between 0 and 1, where the weights indicate the trade-offs between the criteria [4]. The total value of each cluster always is 1 [18]. 3.2 Hierarchical Cluster Analysis The hierarchical cluster analysis (HCA), it is also called hierarchical cluster, is the most popular and widely used method to analyze social network data [5]. The goal of hierarchical cluster analysis is to build a tree diagram where the creteria that were viewed as most similar by the participants in the study are placed on branches that are close together. They are sub-criteria which support for a same criterion on the tree diagram. The key to interpreting a hierarchical cluster analysis is to look at the point at which any given criteria “join together” in the tree diagram [22]. In the HCA process, according to Saaty (1990) comparisons of elements requires that they be homogeneous or close with respect to the common attribute. The number of elements being compared must be small (not more than 9) to improve consistency and corresponding accuracy of measurement. Clustering is process of grouping elements with respect to a common property. One can then decompose the set of ordered elements with respect to an attribute into clusters of from largest to smallest. The relative weights of all elements in the second cluster are divided by the weight of the comment element and then multiplied by its weight in the first cluster in this manner both clusters become commensurate and are pooled together. The process is then repeated to the remaining clusters [18].

134

T. A. Pham and M. V. Nguyen

4 Proposed Methodology for the Assessment of Plant Species for the Roadside at Vung Tau City Step1. Data collection and processing. To analyze, evaluate and select adaptive roadside plant species which promote the landscape values, create typical values for Vung Tau and in accordance with the development strategy orientation of the city, the authors have used a combination of several research methods, namely: Inheritance method: To collect and to explore relevant documents, previous research topics on urban greenery in general and Vung Tau’s roadside trees in particular to gain more information and arguments to apply in the process of analysis, evaluation about existing conditions as well as formulating new proposals. Fieldwork method: To investigate and recorgnize potentials as well as current problems of existing plant species that have been planted in Vung Tau. This activity helps to decide more precisely to choose criteria and subordinate parameter which will be used to evaluate plant species through using the AHP method. This method is also helps to decide the final list of plant species which will be evaluated. Expert method: To interview tree management officials, workers who are directly planting, tending and maintaining shade trees, local people and experts related to urban forestry about the existing problems and requirements of the roadside trees system in Vung Tau as well as suitable criteria applying for the AHP method. Step2. Creation of criteria for seclecting the roadside plant species. To establish criteria for selecting roadside trees species for Vung Tau city, the authors have to follow some principles as below: Principles of selecting the roadside trees species. Based on the combining of interviews with the urban managers in Vung Tau and experts working in the field of urban forestry, doing the fieldworks, collecting related documents and inheritance of previous research results regarding the urban forestry, the authors proposed some key principles in selecting roadside tree species for sustainable developepment strategies in Vung Tau, details are below: Priority for native plant species. Commonly, the native trees have been planted and proven by humans. They are capable of growing and developing in the local area. Therefore, it is necessary to take advantage of these native plant species. They not only perform well the role of roadside trees but also ensure the landscape effect, economic efficiency and bring about landscape stability for the whole streets. Indigenous species with the whole regional characteristics, therefore, using these ones as the dominant species will be a chance to create and ensure the identity of the city. [13]. Principle of suitability between soil and tree. In the process of selecting the plant species, it is necessary to combine the assessment of site conditions with the characteristics of the trees, the adaptability of the trees to conditions of climate, soil, hydrology… It helps these species are capable of growing and developing stably, high preventing pests and diseases.

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

135

Principle of used functions. Transportation are very important for urban development. Based on the types and functions of roads, the selection of plants is variety to meet the appropriate functions. For the purposes of improving the environment and the quality of life for citizens, the shade tree species need to meet functional requirements such as creating ecological environment and shade, cleaning the air environment, noise restrictions…. [12] They also meet the requirements for the streetscape such as types of human activities along streets, aesthetics and visual ballance between charateristics of trees (landscape and functional values, biological properties) with urban facilities, sizes of roads and height limitation of the buildings on both sides of the roads. Principle of meeting urban landscape requirements. According to the city’s master planning and development goals, choosing appropriate plant species will help to increase the biodiversity, combining with the regional environment and historical culture to enrich the natural landscape elements and maximized promote the landscape values. The most importance is the necessary to find specific plant species which are capable of characterizing the Vung Tau’s landscape. Principle of economic savings. It needs to be considered in the selection of roadside plant species. Based on the actual situation of each region and working area, selection the species will be proposed accordingly. Beside of urban landscape values, selecting species have to assess the economic benefits. Criteria for selecting the roadside plant species in Vung Tau city. According to the Ministry of construction (2005), all shade trees are only planted on the sidewalks and median strips where are not less than 3m width [11]. It is the minimum dimension for plant growing normally. After deciding on specific goals, the authors will classify a system of criteria according to each target group and establish a multi-layer structure system for evaluating and selecting roadside tree species for Vung Tau city. Step 3. To determine the weight of the criteria. After establishing the hierarchical model structure, the authors will implement to compare the importance of each pairwise of criteria with the same hierarchy according to Cij value. Assuming that the Ak weight is normal, the criteria lower than C1 , C2 … C15 have branch relationships. Through expert evaluations of Ak -weighting on the importance of criteria Ci and Cj , the authors will operate quantitative comparisons and constituted the process of comparison and evaluation in the following hierarchies: A B; B1 - C; B2 - C; B3 - C and B4 - C. The scale to use in making pairwise comparison is given in Table 1. The process of calculating the single classes by the pairwise comparison mutually compares pairwise until it achieves the desired value, and can be described as follows: Given an input matrix (aij values, with i in row and j in column) measuring the priority for the criteria. + Step 1: Compute the value Di for the i-th row of the input matrix (aij ) as: Di =

n  j=1

(aij )1/n where n is the order of the assessment process.

(1)

136

T. A. Pham and M. V. Nguyen Table 1. The fundamental scale [18].

Intensity of Definition importance on an absolute scale

Note

1

Equal importance

Two activities contribute equally to the objective

3

Moderate importance of one over another

Experience and judgment strongly favor one activity over another

5

Essential or strong importance

Experience and judgment strongly favor one activity over another

7

Very strong importance

An activity is strongly favored and its dominance demonstrated in practice

9

Extreme importance

The evidence favoring one activity over another is of the highest possible order of affirmation

2, 4, 6, 8

Intermediate values between the tow adjacent judgments

When compromise is needed

Reciprocals

If activity i has one of the above numbers assigned to it when compared with activity j, then j has the reciprocal value when compared with i

Rationals

Ratios arising from the scale

If consistency were to be forced by obtaining n numerial values to span the matrix

Table 2. Average RI table of criteria Number 1 2 3 of process steps RI

4

5

6

7

8

9

10

11

12

13

14

15

0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.54 1.56 1.57 1.59

+ Step 2: Compute by induction to find the weights of criteria by the following formula: Wi =

Di , where n is the order of the assessment process. n  Di i=1

(2)

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

137

+ Step 3: Compute the largest eigenvalue λmax of matrix (a_ij) of the assessment process as: 1  1 ( aij × wi ) × n wi n

λmax =

(3)

i=1

+ Step 4: Compute the test criterion (also called consistence index) CI by: CI = (λmax − 1)/(n − 1)

(4)

The smaller CI is the more accurate the test is. + Step 5: Compute the consistency ratio CR as: CR = CI /RI

(5)

RI (random index) is the average consistence index depending on the order n of matrix (aij ). If CR < 0.1, the result is consistent and the assessment process is high confident. If the CR > 0.1, there is evidence of discrepancy in the expert’s evaluation and it needs to re-evaluate and recalculate. Summarizing and evaluating criteria for selecting roadside plant species for Vung Tau by the formula: E=

n 

Qi Pi

(6)

i=1

In which: E is result of evaluation and selection of plant species; Qi is the evaluation weight of criterion i for plant species; Pi is the evaluation score of criterion i; n is the number of evaluation criteria. Step 4. Establish a list of alternative shade plant species in capable of growing in Vung Tau city. After doing the fieldwork, collecting and investigating the relevant documents as well as previous researches about the urban forestry which have done for Vung Tau and its region, combining with interviewing the experts related to urban forestry and relative works, the authors will establish a list of alternative shade plant species which can grow well in Vung Tau city. Step 5. General assessment the shade plant species according to the proposed criteria set. After having the list of selected shade plant species and proposed criteria for roadside trees in Vung Tau city, the authors will assess and caculate the evarage value of each plant species according to the proposed criteria set. Finally, the list of plant species is sorted by evarage values from largest to smallest. Step 6. Finally, propose a list of alternative shade plant species for Vung Tau city. After completing the sorted list of shade plant species, according to the authors’ opinions and interviewing experts, the authors will suggest a short list of alternative shade plant species which can grow well and can create an attractive and characteristic streetscape for Vung Tau city.

138

T. A. Pham and M. V. Nguyen

5 Results and Discussion From methods mentioned above, the criteria system for selecting roadside plant species for Vung Tau are clasified in 3 levels: 1) Class A, level 0 - the main goal for sustainable development for roadside tree system, it is a highest class; 2) Class B, level 1 - the main content class constitutes the principal criteria; 3) Class C, level 2 - The lowest class is the subordinate parameters according class B. The authors classified class B into 4 main criteria: Adaptability (B1 ); Landscape value (B2 ); Functional value (B3 ) and Biological characteristics (B4 ) which contains 15 sub-criteria of class C (from C1 to C15 ), they are shown in Table 3. Based on the specific contents in each criterion, the authors have given the criteria score according to the 9 grades (from high to low grade) that are shown in Table 1. Additionally, the criteria system is evaluated with a national standard of space for roadside tree growing: the sidewalk and and median strips are 3m width as minimum. Table 3. Table of creteria structure for selecting roadside plant species. Criteria frame for selecting roadside plant species (Class A)

Adaptabilities (B1 )

Diseases and pests resistance (C1 ) Sandy soil resistance (C2 ) Pollution resistance (C3 ) Wind and storm resistance (C4 ) Sea salt wind resistance (Cfv5 )

Landscape values (B2 )

Canopy morphology (C6 ) Form (trunk and branch) morphology (C7 ) Leaf (C8 ) Flower (C9 ) Root (C10 )

Functional Values (B3 )

Canopy width (C11 ) Shading effect (C12 )

Biological properties (B4 )

Longevity (C13 ) Secretion (C14 ) Toughness of the trunk and branch (C15 )

Based on the creteria structure above, the authors implemented the quantitative comparisons and constituted the process of comparison and evaluation in the following hierarchies: A - B; B1 - C; B2 - C; B3 - C and B4 - C. There are 5 pairwise comparison matrices in all: one matrice for the criteria with respect to the goal (class A), which are shown in Table 4, four matrices for the criteria (class B), the first of which for the subordinate parameters belong to adaptabilities, shown in Table 5, are: Diseases and pests resistance (C1 ), Sandy soil resistance (C2 ), Pollution resistance (C3 ), Wind and storm resistance (C4 ), Sea salt wind resistance (C5 ); and ten subordinate parameters belong to Landscape values (B2 ), Functional Values (B3 ), Biological properties (B4 ) are shown in

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

139

Table 6, Tables 7 and 8 respectively. They were implemented in the same way to decide their scores and priorities. Finally, the weights of all criteria for selecting roadside tree species in Vung Tau were synthesised and shown all together in Table 9. Table 4. Table of input matrix of criteria for selecting roadside plant species (Class A). Adaptabilities

Landscape values

Functional Values

Biological properties

Prioritiesa

Adaptabilities

1

2

3

2

0.409

Landscape values

1/2

1

1

1/3

0.143

Functional Values

1/3

1

1

1/3

0.125

Biological properties

1/2

3

3

1

0.323

a The priorities are calculated by dividing each criterion (value) by the sum of all criteria CR =

0.053

In Table 4, from the target class to the criteria class (from A to B), the criterion of adaptabilities of the tree (B1 ) has the highest priority with 40,9% of the influence, indicating that the adaptability is the highest requirement for selecting the roadside tree species. When the trees adapt well to the environment, they will have chances to grow perfectly and create other values such as landscape, ecological efficiency… In this case Functional values has the lowest priority with 12,5% of the influence. In Table 5, the criterion of sea salt wind resistance (C5 ) has the highest priority with 44,2% of the influence, indicating that the most important issue for selecting the roadside tree species, in the adaptabilites aspect, is ability of sea salt wind resistance. It also reveals that the sea salt wind phenomenon affect mostly to the shade tree system in Vung Tau. In this table, diseease and pets resistance has the lowest priority with 3,7% of the influence, showing that they are not a big problem in Vung Tau. In Table 6, the criterion of canopy morphology (C6 ) has the highest priority with 53,4% of the influence, indicating that the most important issue for selecting the roadside tree species, in the landscape values aspect, is canopy morphology. It is the most important factor that impact significantly to the aesthetics of streetscape. In this table, root has the lowest priority with 3,7% of the influence, showing that they do not contribute much the landscape values for road system in Vung Tau. In Table 7, the criterion of Shading effect (C12 ) has the highest priority with 66,7% of the influence, indicating that the most important issue for selecting the roadside tree species, in the functional values aspect, is shading effect. It means that the role of environmental improvement is more important than canopy width. The shading effect value has trong related to other ecological characters, such as density, typology and morphology of leave and Form (trunk and branch) morphology…

140

T. A. Pham and M. V. Nguyen

Table 5. Table of input matrix of subordinate parameters according to criterion of adaptabilities (B1 ). Diseases and pests resistance

Sandy soil resistance

Pollution resistance

Wind and storm resistance

Sea salt wind resistance

Prioritiesa

Diseases and pests resistance

1

1/7

1/3

1/5

1/9

0.037

Sandy soil resistance

7

1

3

1/2

1/2

0.211

Pollution resistance

3

1/3

1

1/2

1/5

0.092

Wind and storm resistance

5

2

2

1

1/3

0.218

Sea salt wind resistance

9

2

5

3

1

0.442

a The priorities are calculated by dividing each sub-criterion (value) by the sum of all sub-criteria

CR = 0.046 Table 6. Table of input matrix of subordinate parameters according to criterion of landscape values (B2 ). Canopy morphology

Form (trunk and branch) morphology

Leaf

Flower

Root

Prioritiesa

Canopy morphology

1

3

7

5

9

0.534

Form (trunk and branch) morphology

1/3

1

2

1

5

0.171

Leaf

1/7

1/2

1

1/2

3

0.087

Flower

1/5

1

2

1

7

0.172

Root

1/9

1/5

1/3

1/7

1

0.037

a The priorities are calculated by dividing each sub-criterion (value) by the sum of all sub-criteria

CR = 0.051

In Table 8, the criterion of Longevity (C13 ) has the highest priority with 55,6% of the influence, indicating that the most important issue for selecting the roadside tree species, in the biological properties aspect, is longevity. It is the most important factor that impact significantly to the sustainable development of the roadside trees system as

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

141

Table 7. Table of input matrix of subordinate parameters according to criterion of functional values (B3 ). Canopy width Shading effect Prioritiesa Canopy width 1

1/2

0.333

Shading effect 2

1

0.667

a The priorities are calculated by dividing each sub-criterion

(value) by the sum of all sub-criteria CR = 0.000 Table 8. Table of input matrix of subordinate parameters according to criterion of biological properties (B4 ). Longevity Secretion Toughness of the trunk and Prioritiesa branch Longevity

1

5

2

0.556

Secretion

1/5

1

1/2

0.090

Toughness of the trunk and 1/2 branch

5

1

0.354

a The priorities are calculated by dividing each sub-criterion (value) by the sum of all sub-criteria

CR = 0.048

well as landscape values for long-term. In this table, secretion has the lowest priority with 9,0% of the influence, showing that this criterion does not effect much the biological properites values in Vung Tau. In Table 9, there are the 5 highest weighted subordinate parameters: Sea salt wind resistance (C5 ), Longevity (C13 ), Toughness of the trunk and branch (C15 ), wind and storm resistance (C4), and Sandy soil resistance (C2) with corresponding weight of the criteria respectively: 0.181, 0.179, 0.114, 0.089, and 0,086. These prove that, presently, the roadside trees are attaching great importance to effectiveness of shading, wind and storm withstanding, and trees’ longevity. These are present some basic problems nowadays happening not only in Vung Tau but also in the other cities in Vietnam. They are irrefutable evidences of breaking, collapsing and uprooting whenever whirlwinds or tropical storms appear. Additionally, the weights of criteria for leaf and root are 0.012 and 0.005 respectively. It shows that the current selection of roadside plant species to solve mainly the problems of sustainable development in Vung Tau city; it still has not focused much on urban landscape architecture aesthetics yet. 5.1 General Assessment the Shade Plant Species According to the Proposed Criteria Set for Vung Tau According to the interview results from local managers and experts working in the field of urban forestry, combining with investigation of the common urban shade tree species as well as reality of planted and tested roadside tree species in Vung Tau’s region, the

142

T. A. Pham and M. V. Nguyen Table 9. Table of weighted criteria for selecting roadside tree species in Vung Tau. Target class

Weight of class 0 (A)

Criteria class

Weight of class 1 (B)

Subordinate parameters class

Internal weighta

Weight of class 2 (C)b

Criteria frame for selecting roadside plant species (Class A)

1

Adaptabilities (B1 )

0.409

Diseases and pests resistance (C1 )

0.037

0.015

Sandy soil resistance (C2 )

0.211

0.086

3

Pollution resistance (C3 )

0.092

0.037

4

Wind and storm resistance (C4 )

0.218

0.089

5

Sea salt wind resistance (C5 )

0.442

0.181

Canopy morphology (C6 )

0.534

0.076

7

Form (trunk and branch) morphology (C7 )

0.171

0.024

8

Leaf (C8 )

0.087

0.012

9

Flower (C9 )

0.172

0.025

10

Root (C10 )

0.037

0.005

Canopy width (C11 )

0.333

0.042

Shading effect (C12 )

0.667

0.084

Longevity (C13 )

0.556

0.179

14

Secretion (C14 )

0.090

0.029

15

Toughness of the trunk and branch (C15 )

0.354

0.114

1

2

6

11

Landscape values (B2 )

Functional Values (B3 )

0.143

0.125

12 13

Biological properties (B4 )

0.323

a Weight of each sub-criterion (C) is calculated in each criterion (B) b Weight of each sub-criterion (C) is calculated in the whole criteria frame (A) according to the

weight of each criterion (B)

authors decided to create a short list of 52 species which can be able to plant on the roads system in Vung Tau. The authors assessed and caculated the evarage value of each plant species according to the proposed criteria set. Finally, the list of alterantive plant species is sorted by evarage values from largest to smallest. Evaluation results are shown below:

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

143

Table 10. Table of syntherized assessment about roadside tree species according to criteria. Species

Criteria for roadside trees in Vung Tau

C1

C2

C3

C4

C5

C6

Synthetic evaluating value C7

C8

C9

C10

C11

C12

C13

C14

C15

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

1

13

15

14

15

15

15

13

13

15

14

14

15

13

11

14

14.222

2

14

15

13

14

15

14

13

14

14

12

13

14

14

13

14

14.124

3

14

15

13

15

15

13

12

15

13

12

13

15

13

12

15

14.089

4

14

14

15

14

14

14

13

13

11

13

12

14

14

12

13

13.665

5

12

13

12

15

15

15

13

12

15

15

14

13

13

10

12

13.527

6

13

9

13

13

15

13

14

10

12

13

14

13

15

10

15

13.520

7

12

14

13

14

14

15

14

12

12

13

15

12

13

12

13

13.452

8

13

9

13

13

15

12

14

10

12

13

14

13

15

10

15

13.443

9

13

14

14

13

14

13

14

12

12

14

13

13

14

11

12

13.304

10

14

13

13

14

13

12

13

12

13

13

13

14

14

12

12

13.135

11

14

14

13

14

12

14

13

13

13

12

13

13

13

10

14

13.108

12

13

13

13

12

14

15

13

14

14

13

14

14

13

10

11

13.091

13

13

13

13

12

13

14

12

13

14

14

13

13

14

10

13

13.085

14

13

14

13

14

13

13

13

13

13

12

13

13

12

12

14

13.076

15

7

15

13

12

15

13

13

13

10

12

13

13

13

10

12

13.073

16

13

13

14

14

13

13

12

12

11

13

13

14

13

11

13

13.066

17

13

12

12

13

10

14

13

13

11

14

14

15

15

11

13

12.876

18

13

14

12

12

11

14

14

13

15

8

14

15

13

15

12

12.876

19

13

14

12

12

11

14

12

13

14

8

14

15

13

15

12

12.802

20

11

13

14

13

13

13

14

14

15

15

12

13

12

10

12

12.681

21

10

14

12

13

13

14

13

12

13

13

14

14

12

13

10

12.671

22

13

13

14

14

12

12

10

10

10

13

12

10

14

10

14

12.598

23

14

13

11

12

12

13

13

12

10

13

13

14

13

10

13

12.580

24

14

15

10

15

15

8

10

10

13

15

10

14

10

10

14

12.579

25

10

12

13

13

13

14

13

12

13

13

14

14

12

13

10

12.536

26

13

9

14

15

15

12

13

10

11

13

14

12

12

10

10

12.417

27

12

14

13

12

13

13

13

12

10

13

13

14

12

10

10

12.370

28

10

12

13

13

13

14

12

10

13

13

12

13

12

13

10

12.319

29

13

11

10

12

11

13

12

12

11

9

12

14

14

11

12

12.206

30

10

14

14

13

13

10

13

12

15

12

9

13

10

10

13

12.088

31

10

13

13

11

14

12

11

12

11

12

12

9

12

10

11

11.893

32

13

11

13

13

11

12

13

12

14

12

13

14

12

10

10

11.871

33

13

10

12

13

10

12

12

12

10

13

13

14

13

10

12

11.856

34

12

13

14

12

10

13

13

12

14

11

12

14

12

10

10

11.826

35

12

12

10

10

10

12

14

12

10

13

13

14

14

10

10

11.672

36

12

13

12

13

14

12

13

14

10

12

5

10

10

10

12

11.660

(continued)

144

T. A. Pham and M. V. Nguyen Table 10. (continued)

Species

Criteria for roadside trees in Vung Tau

Synthetic evaluating value

C1

C2

C3

C4

C5

C6

C7

C8

C9

C10

C11

C12

C13

C14

C15

37

12

13

12

13

14

12

13

14

10

12

5

10

10

10

12

11.660

38

12

13

12

13

14

12

13

14

10

12

5

10

10

10

12

11.660

39

10

10

13

12

10

12

10

11

14

12

12

13

12

10

12

11.486

40

12

12

13

10

10

12

5

10

10

10

13

12

13

14

10

11.293

41

9

15

15

14

15

10

12

10

10

12

5

5

10

7

10

11.208

42

13

14

12

14

12

12

10

12

9

12

3

5

10

12

14

11.151

43

13

14

12

14

12

12

10

12

9

12

3

5

10

12

14

11.151

44

13

14

12

14

12

10

12

12

9

11

3

5

10

12

14

11.042

45

10

12

13

10

13

10

10

10

13

10

10

13

11

10

7

10.989

46

7

12

13

10

13

12

13

10

14

13

11

10

10

10

7

10.822

47

7

12

13

10

13

12

13

10

14

13

11

10

10

10

7

10.822

48

9

13

13

9

9

12

13

10

14

10

12

12

11

10

9

10.728

49

14

10

13

7

9

12

13

12

12

13

11

13

12

12

7

10.409

50

14

7

13

13

7

12

13

13

15

14

10

14

10

10

10

10.382

51

5

13

13

14

7

13

13

10

5

14

13

13

10

10

7

10.344

52

3

13

12

14

6

13

13

10

5

10

12

13

10

10

7

10.033

1: Mangifera indica L.; 2: Calophyllum inophyllum L.; 3: Coccoloba uvifera L.; 4: Dipterocarpus alatus Roxb.; 5: Shorea siamensis Miq.; 6: Dalbergia bariensis Pierre.; 7: Pithecoloblum saman (Jacq.) Merr.; 8: Dalbergia cochinchinensis Pierre.; 9: Hopea odorata Roxb.; 10: Dipterocarpus costatus Gaertn. F.; 11: Xylia xylocarpa (Roxb.) Taub.; 12: Chysophyllum cainito L.; 13: Dalbergia tonkinensis Prain.; 14: Cassia siamea (Lam.) H.S.Irwin & Barneby. 15: Terminalia ctappa L.; 16: Swietenia mahogani (L.) Jacq.; 17: Euphoria logan Lour.;; 18: Michelia alba (DC.) Figlar; 19: Michelia champaca L.; 20: Tabebuia aurea (Silva Manso) Benth. & Hook.f. ex S.Moore; 21: Syzygium cumini (L.) Skeels.; 22: Tamarindus indica (L).; 23: Sindora siamensis Teysm. ex Miq.; 24: Barringtonia asiatica L.; 25: Sizygim samarangense (Blume) Merr. & L.M.Perry; 26: Dalbergia cochinchinensis Pierre.; 27: Tectona graudis Linn.; 28: Syzygium nervosum A.Cunn. ex DC.; 29: Shorea roxburghii C. Don.; 30: Barringtoria racemosa (L) Spreng.; 31: Pterocarpus pedatus (Pierre) Gagnep.; 32: Saraca dives Pierre.; 33: Afzelia xylocarpa (Kurz.) Craib.; 34: Peltophorum pterocarpum (DC.) K.Heyne.; 35: Anisoptera costata L.; 36: Normanbya normanbyi (W.Hill) L.H.Bailey.; 37: Bismarckia nobilis Hildebr. & H.Wendl.; 38: Borassus flabelliformis L.; 39: Banhinia sp.; 40: Cinnamomum camphora (L.) Sieb.; 41: Cocos nucifera L.; 42: Elaeis guineensis Jacq.; 43: Pritchardia filifera; 44: Phoenix loureiri Kunth. Var. Loureiri Kunth.; 45: Erythrina fusca Lour.; 46: Cassia javanica L.; 47: Cassia fistula L.; 48: Delonix regia (Boj. ex Hook.) Raf.; 49: Alstonia scholaris R.Br.; 50: Lagerstroemia flosreginae (L.) Pers; 51: Terminalia tomentosa Roxb.; 52: Terminalia mantaly H.Perrier

In Table 10, all of 52 assessed plant species (including indigenous, typical, common urban trees and current well growing roadside plants in Vung Tau) meet the standards of trees planting on the roads and get scores relatively high (>10 points). It means that all species can be planted on the roads system in Vung Tau. Among them, 16 species

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

145

have achieved scores above 13 points. They are the most suitable species for sustainable development in Vung Tau. With 16 suggested species, they are enough for bio-diversity on the road system and they will become the featured species for the city in the future. Additionally, they are all big trees which have large canopy or quite height. Therefore, they are selected to suggest as the most suitable species to plant on the main road system in Vung Tau where have wide sidewalks and median strips (Table 11). Especially, they are proposed as the main species to plant on the gateway roads of the city. Otherwise, these results also shows that the scores of all species are not too different, indicating that the species are similar values of landscape, environment and adaptation to Vung Tau’s conditions. For the other species, including some species having small and medium sizes, they are proposed to plan on the other roads of the city where have less important roles to create the urban feature landscape or have narrower sidewalks. Table 11. List of 16 alternative plant species to plant on the main roads in Vung Tau. Species

Synthetic evaluating value

1

Mangifera indica L

14.222

Notes

2

Calophyllum inophyllum L

14.124

3

Coccoloba uvifera L

14.089

4

Dipterocarpus alatus Roxb

13.665

Featured species

5

Shorea siamensis Miq

13.527

Featured species

6

Dalbergia bariensis Pierre

13.520

Native species

7

Pithecoloblum saman (Jacq.) Merr

13.452

8

Dalbergia cochinchinensis Pierre

13.443

Featured species

9

Hopea odorata Roxb

13.304

Featured species

10

Dipterocarpus costatus Gaertn. F

13.135

Featured species

11

Xylia xylocarpa (Roxb.) Taub

13.108

Native species

12

Chysophyllum cainito L

13.091

13

Dalbergia tonkinensis Prain

13.085

14

Cassia siamea (Lam.) H.S.Irwin & Barneby

13.076

15

Terminalia ctappa L

13.073

16

Swietenia mahogani (L.) Jacq

13.066

Featured species

In the Table 11, there are several new tree species (do not exist in Vung Tau at the moment) that could fit well into the city environment and could be used extensively in Vung Tau’s urban landscape in general as well as its streetscape, such as Calophyllum inophyllum L., Shorea siamensis Miq., Dalbergia bariensis Pierre., Dalbergia cochinchinensis Pierre., Xylia xylocarpa (Roxb.) Taub., and Cassia siamea (Lam.) H.S.Irwin & Barneby.. Therefore, it need to have a long - term policies and investments. Introducing new tree species to plant in a city is a very difficult mission. It needs to have strategies for

146

T. A. Pham and M. V. Nguyen

supporting nurseries to propagate a new tree species as well as distribution information about the importance of the urban biodiversity. Otherwise, we should test, select and identify untraditional species [7]. It is also important to consider the negative selection that occurs in many natural populations by harvesting the best specimens or collecting seeds from native vegetation [7]. Moreover, selection of untraditional species should be encouraged, in order to enhance the biodiversity of cities.

6 Conclusions This paper has investigated the potential and challengs of roadside plant species in Vung Tau and applied multi-critera analysis methods (AHP and HCA) and caculated the average values of 52 plant species based on 15 criteria’s weights to select the most suitable plant species planting on the roadside system in Vung Tau. Based on the obtain results, the following conclusions are derived: Selection of roadside plant species base on the current natural and urban conditions as well as available experience of experts working in the field of landscape architecture urban forestry is necessary. It will be more efficient when it could be incorporated with the scientific methods to find the most adaptable plant species. The application of the multi-criteria analysis methods is a high feasibiligy in analyzing, evaluating priorities of criteria in selecting plant species that are adaptable to not only the natural conditions but also the urban characteristics in Vung Tau. The evaluation process is based on a combination of specific criteria in Vung Tau. Therefore, it is not only giving optimal results to select the urban plant species but also playing roles to direct sustainable development of urban trees system and to create typical landscape values for the city, which can help Vung Tau reach to the goals of the Vung Tau master plan vision to 2035. Otherwise, based on the criteria to select the indigenous plant species and the biological characteristics of each species, it will contribute to emphasize the Vung Tau’s identity values as well as the spirit of the city. Moreover, combining with others tree characteristics related to physiology and morphology or even predicted climate change scenarios, it will be useful to develop various and flexible plant options for selection of the planting and landscape design purposes, a fruitful tool for urban designers and planners. The weighting process mentioned in this paper is also helpful for policy makers, urban managers, landscape architects, designers, and services working in the field of urban forestry as well as citizen and gives them all a belief to build a future sustainable city. Even, when the commercial and ecological interests of selecting roadside tree species are in conflict, these methods can provide a clear answer which species are responding to the affordability of seedlings, nursering, growing, maintaining and the ecological situations of the region. Moreover, special creteria and subordinate parameters were selected alternatively, which becomes an apparent guarantee for suggestion of the roadside tree species for Vung Tau city. The used methods to select roadside plant species for Vung Tau city can also be applied for other cities in Vietnam with the adjustions of creteria frame as well as their weights and factors based much on the different characteristics of the cities.

Assessment of Plant Species for the Roadside at Vung Tau City of Vietnam

147

Acknowledgement. The authors gratefully acknowledgement to every supports from Vung Tau Urban Management Office and financial support from Vung Tau People’s Committee, Vung Tau city, Vietnam, for this work under a project named “Project for development of urban tree system in Vung Tau for a vision to 2050”. This project has been implemented for three years (2017–2019).

Referencess 1. Asgarzadeh, M.: Plant selection method for urban landscapes of semi-arid cities (a case study of Tehran). Urban For. Urban Greening 13(2014), 450–458 (2014) 2. BaRia-VungTau online, Green area of Vung Tau city increased rapidly https://www.baobar iavungtau.com.vn/kinhte/201504/dientichcayxanhtpvungtautangnhanh598036/. Accessed 4 Dec 2017 3. Bauman, M., et al.: A review of multi-criteria decision making approaches for evaluating energy T storage systems for grid applications. Renew. Sustain. Energy Rev. 107, 516–534 (2019) 4. Cinelli, M., Coles, S.R., Kirwan, K.: Analysis of the potentials of multi criteria decision analysis methods to conduct sustainability assessment. Ecol. Ind. 46, 138–148 (2014) 5. Hexmoor, H.: Computational Network Science-An Algorithmic Approach, p. 128. Morgan Kaufmann, (2016). https://doi.org/10.1016/C2013-0-19272-0>. Accessed: 29th, July, 2020 6. Jacobs, J.: The death and Life of Great American Cities, p. 472. Random House, New York (1961) 7. Konijnendijk, C. C. et al. (eds.): Urban Forests and Trees, p. 269. Springer, Berlin (2005) 8. Li, Y.Y., Wang, X.R., Huang, C.L.: Key street tree species selection in urban areas. J. Agric. Res. 6(15), 3539–3550 (2011) 9. Longaray, A.A.: Proposal for using AHP method to evaluate the quality of services provided by outsourced companies. Procedia Comput. Sci. 55, 715–724 (2015) 10. Majumder, M.: Impact of Urbanization on Water Shortage in Face of Climatic Aberrations, p. 98. Springer, New York (2015) 11. Ministry of Construction: Circular no. 20/2005/TT-BXD dated december 20, 2005, on Guidelines for Urban Forestry (2005) 12. Pham, A.T., Pham, H.P.: Roles of urban trees in Hanoi. Constr. Plann. Mag. 82–2016, 82–85 (2016) 13. Pham, A.T., Le, K.L.: Some points of view in urban tree. Archit. Mag. Vietnam Assoc. Archit. 262, 64–66 (2017) 14. Pham, A.T.: Project for development of urban tree system in Vung Tau with a vision to 2050. College Landscape Archit. Urban Tree, Hanoi (2019) 15. Pham, H. P.: Research on assessment and selection criterion of tress plant species in Hanoi Streets. Journal of Forestry Science and Technology No.1–2017 pp. 35–45 (2017) 16. Saaty, T.L.: A scaling method for priorities in hierarchical structures. J. Math. Psychol. 15(3), 234–281 (1977) 17. Saaty, T.L.: The Analytic Hierarchy Process, p. 287. McGraw- Hill, New York (1980) 18. Saaty, T.L.: How to make a decision: The Analytic Hierarchy Process. Eur. J. Oper. Res. 48, 9–26 (1990) 19. Saaty, T.L.: Decision making with the analytic hierarchy process. Int. J. Serv. Sci. 1(1), 83–98 (2008) 20. Saaty, T. L.: Chapter 10. The analytic hierarchy and analytic network processes for the measurement of intangible criteria and for decision-making. In: Greco, S., Ehrgott, M., Figueira, J. R., (eds.) Multiple Criteria Decision Analysis-State of the Art Surveys (2nd edition), pp. 363–419. Springer, New York (2016)

148

T. A. Pham and M. V. Nguyen

21. Saaty, R.W.: The analytic hierarchy process – What is it and How it is used. Mathl Modell. 9(3–5), 161–176 (1987) 22. Tullis, T., Albert, B.: Hierarchical Cluster Analysis in Measuring the User Experience Collecting, Analyzing, and Presenting Usability Metrics (Second Edition), p. 320. Morgan Kaufmann (2013). https://doi.org/10.1016/C2011-0-00016-9. Accessed 29 July 2020

Modernization of Height System in Vietnam Using GNSS and Geoid Model Ngoc Ha Hoang(B) Hanoi University of Mining and Geology, Hanoi, Vietnam [email protected]

Abstract. Recently, Vietnam has modernized the national height system, including the development and establishment of new control points, completing measurement of the national height network figure, upgrading the national Geoid model. One of the most important contents to attain the goal of modernizing the height system is to connect the national grid lines of the grades I and II with GNSS CORS stations and the fundamental gravity points. Vietnam currently uses the 2010 Geoid model which was built on the basis of the Global EGM 2008 Geoid model with additional data of more than 30,000 detailed gravity points and over 800 GPS-TC points. The combined processing of GNSS-leveling data and the gravity geoid model to upgrade the local geoid model with high accuracy (size of 4–10 cm) can allow high satellite measurement technologies to replace the traditional leveling method in determining elevations to achieve accuracy level of grades III and IV. The article presents the research and results of a newly-developed method to combine the adjustments of normal, GNSS, and anomalies heights. Keywords: GNSS · Height system · Levelling network · Geodetic computations · Geoid · Vietnam

1 Introduction The height system plays a vital role in building spatial data infrastructure for sustainable development and exploitation of natural resources. The height is determined by using the point difference measurement method. Today, leveling technology is still applied with increasing accuracy of millimeters/km. However, since GNSS has been widely used, high satellite measurement technology has been studied and applied effectively. In many countries around the world, high satellite measurement methods are gradually replacing traditional leveling methods due to faster construction time, lower costs and potentially higher accuracy [14]. The trend of modernizing the height system is associated with the application of GNSS technology in combination with the building of a highly accurate Geoid model in each country to replace traditional leveling technology gradually. In big cities of Vietnam, the process of urbanization has affected the environment, causing subsidence, surface deformation and damage to construction and transport works [14]. On the other hand, factors related to climate change make certain areas vulnerable to © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 149–166, 2021. https://doi.org/10.1007/978-3-030-60269-7_8

150

N. H. Hoang

flooding and erosion of rivers. The modernization of the national height network to serve planning, construction, socio-economic development and climate change adaptation as well as the exploitation of natural resources and nature, is essential. In this paper, we will present an overview of the process of the formation and development of height systems in Vietnam. We will also address some essential contents of the modernization of height systems using GNSS technology and the use of the Geoid model. The paper will focus on presenting issues related to processing geodetic network data with examples in the Central Highlands - South Central region of Vietnam. 1.1 The Formation and Development of Height Systems in Vietnam The geodetic leveling network in the north of Vietnam was built from 1959 to 1964 with the starting point at the Hon Dau Island, Hai Phong City. In the South, before 1975, the height system with the fundamental point at the Ha Tien town, Kien Giang province was used. The Class I and II altitude network was built in 1959, following four national roads, Hanoi - Hai Phong, Hanoi - Dong Dang, Hanoi - Lao Cai and Hanoi Vinh Linh with a total length of 1175 km. The establishment of a unified height network for the entire territory of Vietnam was carried out from 1981 to 1991. The average sea level at Hon Dau Naval Station was computed with data from 43 years from 1950 to 1992. The national height network also has more than 12,000 coordinate points of the class III measured by GPS technology. This height is computed based on the geoid model, which was completed in 2012 with an accuracy equivalent to the leveling of IV-class. In 2001, the Department of Surveying, Mapping and Geographic Information Vietnam completed the construction of the geoid model for the whole country, with standard gravity grid of 3 × 3 in the plains and midlands of Vietnam, based on more than 1038 GPS and leveling control points and nearly 30,000 detail gravity points, and EGM2008. This geoid has an accuracy of about 8 cm for the plain areas and about 20 cm for the midlands area of Vietnam. The geoid model was then used to determine the height for the whole territory of Vietnam using GNSS technology. Specifications of height networks is shown in Table 1. Table 1. Characteristics of the national height networks of Vietnam. Area

Class

I (m m) II (m m) III (m m) IV (m m) √ √ √ √ Flat topography ±2 L ±4 L ±10 L ±20 L √ √ √ √ Mountain terrain ±3 L ±5 L ±12 L ±25 L (L- is the linear length in units of Km)

1.2 The Main Contents of the Modernization of the National Height System • Determining the value “0” of the original height point through the calculation of the average sea level in Hon Dau using oceanographic data in 18.6 years. Determining

Modernization of Height System in Vietnam Using GNSS and Geoid Model

• • •

•

•

•

•

151

the value “0” of the national height and determining the value “0” depth for specific waters through monitoring data at the Naval Station. Building and improving existing stable control points with high accuracy to ensure stable and long-term use. These control points were buried deep into the rock strata and located along the I - and II - classes lines. Completing the measurement of the new national height network connecting the lines of the I - and the II- classes networks, GNSS CORS stations, state gravity points, and navigational observation stations in the restored sea areas. Developing and repeating measurement of vertical shift monitoring network in Hai Phong, Hanoi and Da Nang cities by using leveling measurement; establishment and measurement of vertical displacement monitoring networks by combining hydroleveling and satellite technologies in Ho Chi Minh City, Can Tho City, and the Mekong Delta Long for analyzing, evaluating and determining the causes and trends of subsidence to supplement updated climate change scenarios. Measuring GNSS to achieve high accuracy at some I - and II- classes and at the basic gravity I-class control points, and at the points of the oceanographic monitoring stations in sea areas of Hon Dau (Hai Phong), Hoanh Son (Ha Tinh), Son Tra (Da Nang), Quy Nhon (Binh Dinh), Vung Tau (Ba Ria-Vung Tau) and origin point of Ha Tien height. The measurement sought to ensure a density of about 10 ÷ 20 km/point for upgrading local Geoid model. At the end of 2019, Vietnam has 65 national satellite navigation stations, located mainly in Hanoi, Ho Chi Minh City and some provinces in Northern, Southern Delta and Highland areas. Covering the entire territory and the big islands need additional building about 75–80 stations. Detailing the gravity measurement of Vietnam’s mountains. Upgrading the nationwide gravity geoid model by updating the 2011 gravity geoid model with new gravity data and numerical elevation models, and combining with the global geodetic systems WGS-84, Global Geoid EGM2008 or EGM Processing combined GNSS-Leveling data and gravity geoid model to upgrade the mixed geoid model on the territory of Vietnam in the mainland and coastal areas to achieve accuracy of 5 cm in the delta area and 10 cm in the mountains area. Employing the high accuracy GNSS method to replace the traditional leveling technology of the III- and IV- classes. Calculating adjustment of the national height network and announcing the new national height system. Building a national height database.

2 Development of Algorithms for the Adjustment and Analysis of the Height Networks 2.1 Theoretical Background of Height Adjustment Networks 2.1.1 Adjustment of the Leveling Networks In Vietnam, along with the development of Information and Communication Technologies (ICT), modern theories have also been employed to solve the problem of the geodetic adjustment. In 1999, the adjustment of the plane network combined with GPS data was completed, and in 2000, the VN-2000 coordinate system was announced. Meanwhile,

152

N. H. Hoang

the I- and II- classes leveling networks were adjusted in 1996, and then, after completing the national height network (2001–2008), the adjustment was revised. Herein, the adjustment software is based on the theory of the least square method was presented in [6, 7, 9–11]. It is noteworthy that the theory of the free network adjustment has been developed and widely applied to solve the vital problem of monitoring deformation of construction works due to the advantages that it is not being affected by the errors of the original data. Since the theory of free network adjustment has been presented in various studies, including books [6, 9–12], salient features of calculating the network leveling are summarized in steps as follows (Fig. 1):

1

2

1

2

4

3

3

5

6

4 Fig. 1. Free levelling network

Normal Equation System Rx + b = 0. Where: R = AT PA; b = AT PL. ⎛ ⎞ 0.5 ⎜ 0.5 ⎟ ⎟ b=⎜ ⎝ 1 ⎠; 1

⎛ ⎞ 0 ⎜ 0⎟ ⎟ x(0) ⎜ ⎝0⎠ 0 ⎛ ⎞ 1 ⎜ 1⎟ ⎜ ⎟ Matrix C = ⎜ ⎟ ⎝ 0⎠ 0

(1)

Modernization of Height System in Vietnam Using GNSS and Geoid Model

153

Matrix R of Normal Equation System: ⎛ ⎞ 3 −1 −1 −1 ⎜ −1 3 −1 −1 ⎟ ⎟ R =⎜ ⎝ −1 −1 3 −1 ⎠; −1 −1 −1 3 x = −R∼ b. Establish the matrix R as follows:

R−1 c

(2)

R C ; Rc = CT 0 −1   ∼ R C R T = = TT O CT O 

(3) (4)

R~ calculated by the formula: −1

R∼ = R + CCT − TTT ⎛ ⎞ 1 ⎜ 1⎟

⎜ ⎟ CCT = ⎜ ⎟ 1 1 0 0 = ⎝ 0⎠

−1

T = B C T .B

0 ⎛ ⎞ 0.5 ⎜ 0.5 ⎟ ⎟ =⎜ ⎝ 0.5 ⎠; 0.5

(5)

⎛

⎞ 1100 ⎜1 1 0 0⎟ ⎜ ⎟ ⎝0 0 0 0⎠ 0000 ⎛ ⎞ 0.25 0.25 0.25 0.25 ⎜ 0.25 0.25 0.25 0.25 ⎟ ⎟ T .T T = ⎜ ⎝ 0.25 0.25 0.25 0.25 ⎠ 0.25 0.25 0.25 0.25

Matrix symbol: M = (R + CCT ). ⎛ ⎞ 24 8 16 16 ⎟ 1 ⎜ ⎜ 8 24 16 16 ⎟ = M−1 = ⎝ 64 16 16 40 24 ⎠ 16 16 24 40 ⎛ ⎞ 2222 ⎟ 1 ⎜ ⎜2 2 2 2⎟ TTT = ⎝ ⎠ 2 2 2 2 8 2222

⎛

⎞ 3122 ⎜1 3 2 2⎟ ⎜ ⎟ ⎝2 2 5 3⎠ 2235

R~ is calculated as follows: ⎛

⎞ 1 −1 0 0 −1

1 ⎜ −1 1 0 0 ⎟ ⎟ − TTT = ⎜ R∼ = R + CCT 8⎝ 0 0 3 1⎠ 0

0 13

(6)

154

N. H. Hoang

Vector Δx is determined as follows: ⎛ ⎞ 1 −1 0 0 1 ⎜ −1 1 0 0 ⎟ ⎟ x = ⎜ 8⎝ 0 0 3 1⎠ 0 0 13

⎛

⎞ 0.5 ⎜ 0.5 ⎟ ⎜ ⎟ ⎝ 1 ⎠; 1

(7)

⎛

⎞ 0 ⎜ 0 ⎟ ⎟ Vector x = ⎜ ⎝ −0.5 ⎠. −0.5 2.1.2 Combined Adjustment of the Height Networks with GNSS and Levelling Data

M Earth's surface hMg HM

hM Geoid ζM

Quasigeoid Elipxoid

Fig. 2. The normal height h and height anomaly ζ

Figure 2 depicts the relationship between normal height and ellipsoid height as follows: h = H − ζ.

(8)

where, h: normal height H: ellipsoid height (geodetic height) ζ: height anomaly Kotsakis and Sideris [8] discussed the adjustment problems, methods for combining GNSS, leveling and geoid heights, and suggested a method for evaluating the accuracy of

Modernization of Height System in Vietnam Using GNSS and Geoid Model

155

geoid models. Fotopoulos [2] considered least-squares variance component estimation (VCE) in the combined adjustment of ellipsoid, orthometric and geoid height data. In this paper, we propose an algorithm based on the method of adjustment of the conditions with unknowns. According to theory of combined adjustment of leveling, GNSS networks continue to develop in many research projects, such as [1–4]. From Eq. (8) we have: H − ζ − h = 0.

(9)

where, H = H(0) + vH h = h(0) + vh ζ = ζ(0) + f + vζ where H(0) and h(0) are determined by the adjustment of the GNSS and levelling networks, respectively. For each of the common points of the GNSS and the height networks we can form the following equation: vH − vh − vζ − f + w = 0

(10)

w = −( H(0) − h(0) −(0) ζ ) In practice, f function in the formula (10) may be approximated by different kinds of functions in order to fit the quasigeoid. In Vietnam, the problem of choosing f = f (x, y) is presented in research works such as [9, 13]. a) Linear Model

f xi , yi = a.xi + b.yi + c

(11)

xi, yi: is the coordinates of point i; b) Polynomial Model of Degree 2 f(xi, yi) = a.xi + b.yi + c.xi2 + d .xi yi + e.yi2 + f

(12)

c) Spline Function Model f(x, y), =

n i=1

2 ai rPP ln rPPi + τ1 + τ2 x + τ3 y i

(13)

Where, rPPi =



(x − xi )2 + (y − yi )2 , ai (i = 1 ÷ n)

(14)

156

N. H. Hoang

τ1 , τ2 , τ3 - is the solution of the following system of equations: ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ 0 g1,2 . . . g1,n 1 x1 y1 ζ1 a1 ⎢g ⎥ ⎢a ⎥ ⎢ζ ⎥ 0 . . . g 1 x y 2 1 ⎥ ⎢ 2⎥ ⎢ 2⎥ ⎢ 2,1 2,n ⎢ ... ... ... ... ... ... ...⎥ ⎢...⎥ ⎢...⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ gn,1 gn,2 . . . 0 1 xn yn ⎥ × ⎢ an ⎥ = ⎢ ζn ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ 1 1 . . . 1 0 0 0 ⎥ ⎢ τ1 ⎥ ⎢ 0 ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎣ x1 x2 . . . xn 0 0 0 ⎦ ⎣ τ2 ⎦ ⎣ 0 ⎦ τ3 0 y1 y2 . . . yn 0 0 0  

r2 ln rpi pj with i = j gi,i = gj,j = pi pj 0 with i = j

(15)

(16)

d) Several models can be used ranging from simple linear regression to more complicated seven parameter similarity transformation model, Kotsakis and Sideris (1999). In our study, the model selected for calculation will be a four-parameter model according to Heiskanen and Moritz [5]. fi = f (Bi , Li ) = ai x = x1 + x2 ( cosBi cosLi ) + x3 (cosBi sinLi ) + x4 (sinBi ) (17) B, L - latitude, longitude of the network point Suppose f = f (B, L) = aT x. From the n point, GNSS has a leveling height h, anomaly height ζ and geodetic height H, we have a system of conditional equations with -unknowns: BV + Ax + W = 0,

(18)

Bnx3n = (Enxn −Enxn −Enxn ); VT = (VHnx1 Vznx1 Vhnx1 ); E – Matrix unit; Anxk – Coefficient matrix. (0)

W = −(H(0) − h(0) −ζ ) . Matrix A has the form ⎛ 1 cos B1 cos L1 cos B1 sin L1 sin B1 ⎜... . ⎜ ⎜ A=⎜ · ⎜ ⎝ · · 1 cos Bk cos Lk cos Bk sin Lk sin Bk

(19) ⎞ ·⎟ ⎟ ⎟ ⎟ ⎟ ⎠

(20)

Modernization of Height System in Vietnam Using GNSS and Geoid Model

157

Also perform calculations to compare the Model 5 parameters: f = f(Bi , Li ) = ai x = x1 + x2 ( cos Bi cosLi ) + x3 (cosBi sinLi ) + x4 (sinBi ) + x5 sin2 Bi ) Then Matrix A will be: ⎞ ⎛ 1 cos B1 cos L1 cos B1 sin L1 sin B1 sin2 B1 ⎟ ⎜... . . ⎟ ⎜ ⎟ ⎜ A=⎜ ⎟ ⎟ ⎜ ⎠ ⎝ . . 2 1 cos Bk cos Lk cos Bk sin Lk sin Bk sin Bk 

AT = aT1 aT2 ... aTn ⎛ QV = ⎝

(21)

(22)

⎞

QH

⎠

Qζ

(23)

Qh x- vector of unknowns. – QH , Qζ, Qh - covariance matrixes of vectors H, ζ, h. Quantity ⎛

H

⎞

⎜ ⎟ y = H − ζ − h = (E − E − E)⎝ ζ ⎠ h ⎛ ⎞ QH ⎠(E − E − E)T = QH + Qζ + Qh Qy = BQBT = (E − E − E)⎝ Qζ Qh

(24)

(25)

System of Eqs. (18) is solved with the following conditions: −1 VH + V ζT Qζ−1 Vζ + VhT Qh−1 Vh = Min  = V T QV−1 V = VHT QH

(26)

We set up the Lagrange function  = VT PV + 2KT (BV + Ax + W) = min

(27)

Calculate derivatives in vectors: ∂φ = 2VT P − 2KT A = 0. ∂V

(28)

∂φ = − 2KT A = 0. ∂x

(29)

158

N. H. Hoang

We have the formula: V = P−1 BT K.

(30)

AT K = 0.

(31)

Substituting the expressions (30) and (31) into (18) we have:

(32)

Where N = B Q BT Q = P−1 Expression (16) can be rewritten as:    K N A W + =0 x AT O O The solution vector of system of Eqs. (33) will be   K −1 W = − Nβ O x

(33)

(34)

Here the matrix  NA =

N A . AT O

(35)

From the first equation of the system (32), we have the expression K = −N −1 Ax − N−1 W = − N−1 (Ax + W) = −N−1 W1

(36)

Here vector W1 = Ax + W.

(37)

Substituting the expression (36) into the second equation of the system (32) we get AT N−1 Ax + AT N−1 W = 0

(38)

W1 = Ax + W .

(39)

Or

Modernization of Height System in Vietnam Using GNSS and Geoid Model

159

−1 x = − AT N−1 A AT N−1 W = − [AT (QH + Qζ + Qh )−1 AAT (QH + Qζ + Qh )−1 W Vector V is calculated using the expressions (30) and (36) as follows ⎛ ⎞ ⎛ ⎞⎛ E ⎞ VH QH ⎜ ⎟ ⎟ −1 ⎠⎜ V = ⎝ Vζ ⎠ = QBT K = −⎝ Qζ ⎝ −E ⎠N W1 Qζ Vh −E ⎞ ⎛ QH

−1 ⎟ ⎜ = −⎝ Qζ ⎠ QH + Qζ + Qh W1 Qh

(40)

(41)

Or is: V = T W1

(42)

Where, the matrix ⎛

QH

⎞

⎟ ⎜ T = ⎝ Qζ ⎠ N−1 Qh

(43)

To evaluate the accuracy after adjustment, we need to calculate the covariance matrix of vector V after adjustment as follows: QV = T Qw1TT .

(44)

The covariance matrix of vector x determined from the formula (40)

−1 . Qx = AT N−1 A

(45)

Symbol matrix R = AT N−1 A substitute formula (40) into formula (39) we have

 (46) W1 = Ax + W = −A R−1 AT N−1 + E W. E is the unit matrix. Follow the formula for calculating covariance matrix of the function we have



 Qw1 = −A R−1 AT N−1 + E N −N−1 A R−1 AT + E



 = −A R−1 AT + N −N−1 A R−1 AT + E = (A R−1 AT1 A R−1 AT − N−1 A R−1 AT +1 A R−1 AT + N = N − A R−1 AT

(47)

160

N. H. Hoang

So we have the expression Qw1 = N − AQx AT .

(48)

VHi , Vζ , Vhi used to calibrate Hi , ζi , hi at the GNSS common points and levelling network. It should be noted that, in the adjustment of mixed quantities of different properties, we should use the matrix defined as follows: C = σ 2 Q. ⎛ CV = ⎝

(49) ⎞

CH

⎠

Cζ

(50)

Ch Here σ- standard deviation. The Q matrix is determined from the results of separate adjustment of geodetic networks. The calculation formulas from (40) to (44), the matrices QH , Qζ, Qh are replaced with the matrix CH , Cζ, Ch . 2.2 Methodology for the Combined Adjustment and Analysis of the Height Networks Step 1: Data preprocessing • Input: – Leveling observations. – GNSS observations. – Geoid model EGM 2008 • Output: – Normal, Ellipsoid heights and height anomalies ζ – The matrices QH , Qζ, Qh. To illustrate the theory, we proceed to adjust the GNSS and regional leveling network of the I - class and the II - class in the Central Highlands with the grid diagram shown in the Fig. 3. In this figure, there are 19 points in the GNSS network, measured and adjusted by the Natural Resources and Environment of Vietnam Corporation. The calculation and adjustment of the GNSS network with one origin point with coordinates B, L, and H were based on the VN- 2000 coordinate. The height anomaly ζ is calculated from the harmonic coefficients with data taken from the Geoid model EGM 2008. GM2008 altitude anomaly is determined in the WGS-84 coordinate system and

Modernization of Height System in Vietnam Using GNSS and Geoid Model

161

Fig. 3. The common points of the GNSS and levelling network at the Central Highlands of Vietnam

calculated from the harmonic coefficients using the HARMONIC_SYM-WGS-84 program with EGM2008_to_2190_TideFee and Zeta_To_N_2160_EGM_2008. From the calculation results, we have vector h, geodetic height vector H, and Ch and CH matrices. The data included in the calculation are given in Table 2 [9]. To illustrate the method, we choose models 4 and 5 parameters.

162

N. H. Hoang Table 2. Data of points in the Central Highlands - South Central Coast. B° 



L° 



x (m)

y (m)

H (m)

h (m)

ζEGM2008

1 12 17 49.7278 108 8 13.1651 1361403.1 84126742

533.791

531.386

0.975

2 13 3 51.8607

108 13 3.5957 1446475.1

849003.67

762.892

760.658

−0.832

3 12 44 2.6369

108 45 8.5642 1410672.6

907606.89

423.331

420.94

4 12 39 33.7426 108 1 35.6883 1401372

828787.48

433.139

431.206

−0.998

1.917

5 12 54 47.3372 108 15 57.685 1429790.8

854468.8

683.955

681.622

−0.286

6 12 48 18.5108 108 32 19.266 1418219.7

884258.82

468.914

466.566

0.897

7 12 34 55.5563 107 50 29.777 1392591.3

808770.27

360.331

358.653

−1.641

8 12 29 42.5938 107 44 18.233 1382848.8

797647.78

582.95

581.077

−1.607

9 12 33 12.6619 108 10 6.5913 1389832.8

844360.07

479.26

476.956

0.117

10 12 48 36.0408 108 26 9.7056 1418608.3

873092.14

496.481

494.14

0.501

11 11 55 21.3842 107 44 56.576 1319480.6

799449.86

595.505

593.597

0.249

12 11 16 9.3570

838836.33

212.09

210.656

2.398

13 11 49 27.3762 108 34 10.514 1309607.8

889064.06 1019.755 1016.085

4.646

14 11 38 48.6369 108 1 9.264

829234.28 1034.089 1031.218

2.406

15 11 37 13.0575 108 13 53.458 1286570.5

852437.46

774.008

770.976

3.405

16 11 52 30.3191 107 55 23.186 1314414.1

818479.88

856.696

854.127

1.354

17 11 56 11/9678 108 9 36.538

1321514.4

844251.83 1115.27

1112.005

2.762

18 12 12 54.3090 108 7 23.684

1352299

839876.04

498.455

495.959

1.197

866373.15

966.814

963.3346

3.993

19 11 46 2.4338

108 6 11.131

1247546.9 1289255.2

108 21 39.369 1303020.4

Step 2: Combined network adjustment for corrector Surface Model (CSM) Twelve common points of GNSS and leveling network were chosen to calculate the adjustment. Points 13 to 19 are control points. 4-parameter model according to formula (17) was selected, and the components of matrix A are defined in Table 3. 2 E; Cζ = σ2 ζE; C = σ2 E . To calculate we have: CH = σH h h Where: E- matrix unit (12 × 12). σH = 20 mm, σζ = 100 mm, σh = 15 mm. Calculated by formula (40), vector x would be ⎛ ⎞ −473.397 ⎜ 55.4673 ⎟ ⎟ x=⎜ ⎝ 473.631 ⎠ 244.289 Step 3: Conducting normal heights calculation for the test points according to the following formula hj = Hj − ζj + aj x (j = 13, 14 . . . 19)

(51)

Modernization of Height System in Vietnam Using GNSS and Geoid Model

163

Table 3. Matrix components A. x1 x2

x3

x4

1

−0.30415

0.9285117

0.2129815

1

−0.30453

0.9252903

0.226046

1

−0.31357

0.9236267

0.220426

1

−0.30193

0.9277971

0.2191542

1

−0.30550

0.9255964

0.2234736

1

−0.31004

0.9245298

0.2216358

1

−0.29903

0.9290476

0.2178381

1

−0.297453 0.9298984

0.216357

1

−0.304357 0.9274295

0.2173512

1

−0.308373 0.9250659

0.2217187

1

−0.298270 0.9318558

0.2065901

1

−0.304735 0.9321736

0.1954199

where: x is the parameter vector of the corrector surface estimated via the combined adjustment. As long as the GNSS observations and a geoid model are available, the normal heights of new points are achievable by using the above equation. To calculate only the normal height of the GNSS measurement points without leveling, we will use the difference

(52) hji = Hji − ζji + aj − ai x. Here, vector x is calculated in expression (40). The weight of these measurements is calculated by the following formula:



T Qhji = QHji + Qζj + aj − ai Qx aj − ai . Phji = Q−1 hji

(53) (54)

2.3 Results and Discussions Executing the calculation by formula (44) we have matrices QH , Qζ, and Qh . Diagonal components of these matrices are presented in Table 4. Estimated Variance Component (Unit [mm2 ]) σ2 = 1.422289 Analysis of values of vectors VH , Vζ , Vh for model 4 and 5 parameters is presented in Table 5.

164

N. H. Hoang Table 4. The main diagonal components of the matrices QH , Qζ, Qh. Points diag QH

diag Qζ

diag Qh

1

8.758

5474.069 2771

2

7.040

4400.026 2.227

3

8.689

5431.065 2.749

4

12.958

8099.079 4.100

5

12.348

7717.589 3.907

6

11.764

7352.73

7

11.771

7356.774 3.724

8

10.643

665.114 3.368

9

11748

7342.411 3.717

10

12.608

7880.295 3.989

11

10.549

6592.938 3.338

12

1.59204

995.025 0.504

3.722

Table 5. Results of the empirical test for the 12 GNSS/levelling points used in the combined adjustment. Unit [m] 4 ParamtrsTransf model

5 ParamtrsTransf model

VH

VH

Vζ

Vh

Vζ

Vh

Mean

0.000

0.000

0.000

0.000

0.000

0.000

Max

0.008

0.121

0.005

0.005

0.086

0.002

Min RMS

−0.004 −0.103 −0.002 −0.008 −0.187 −0.004 0.004

0.094

0.002

0.004

0.092

0.002

The deviation of the differences between the heights calculated by the formula (31) and the standard height of the test points is shown in Table 6. From Table 4 and Table 5, for the twelve participating common points, the accuracy of the GNSS height, normal height, and anomaly height all increased after the computation. However, the combined adjustment result did not increase the quality of the network leveling. From the calculation results in Table 6 for seven test points, it is evident the results of the adjustment allowed the determination of the parameters in the models (17) and (21). The results of the adjustment also enabled the determination the altitude of the GNSS that meet the III- and IV- classes standards.

Modernization of Height System in Vietnam Using GNSS and Geoid Model

165

Table 6. Error and deviation of the levelling lines related to 13, 14, 15, 16, and 17 test points. Levelling line

L (km)

4 P M (mm)

14_13

63.197

−15.94

2.699

95.396

198.741

13_12

79.840

−116.01

−138.37

107.224

223.383

13_18

65.13064

−278.01

−303.59

96.844

201.759

18_17

31.094

50.916

49.793

66.914

139.405

17_16

26.732

65.5215

82.904

62.0438

129.258

17_19

28.834

117.110

121.501

64.436

134.2426

15_17

35.88984

−201.435

−201.67

71.88975

149.7703

15_19

21.559

−84.324

−80.168

55.718

116.080

14.873

15.683

12.000

25.000

σ (1 km)

5 P M (mm)

Deviation limit (III-class) (mm)

Deviation limit (IV-class) (mm)

3 Summary and Conclusion This paper aims to present the development, modernization of the height system in Vietnam, and the algorithm for the adjustment and analysis of height systems. It included the proposed methodology for the combined adjustment of leveling and GNSS networks. The modernization of the height system in Vietnam, as in many other countries, essentially develops a new modern technological method to process the combined data measured both from the GNSS method and traditional leveling technique. It uses the best modeling geoid in every country. These new technologies shorten the time to develop height data system from several decades to two to three years. Thanks to this modernization, data accuracy has been greatly improved, and data could be stored more conveniently for easy exploitation for diverse purposes. The application of new technology allows solving complex problems such as monitoring the subsidence of construction works as well as forecasting problems related to mining. The proposed methodology for the combined adjustment of the height networks by experimental calculation has been conducted in the Central Highlands - South Central region of Vietnam. The content of the algorithm includes the following processes: (1) Preprocessing and separate adjustment of leveling and GNSS heights, (2) Combined adjustment with leveling, GNSS, and height anomalies, and (3) Evaluation of the control points. This algorithm allows inputting covariance matrices of geodetic height, leveling height, and height anomaly. If the error of height anomaly is greater than 2 dm, it will not significantly affect the adjustment of GNSS and leveling height values. Correction numbers are primarily used to correct height anomalies. The experimental results demonstrated the effectiveness of this algorithm. This algorithm can be applied to adjust the height networks in Vietnam.

166

N. H. Hoang

References 1. Eshagh, M., Berntsson, J.: On quality of NKG2015 geoid model over the Nordic countries. J. Geod. Sci. 9, 97–110 (2019) 2. Fotopoulos, G.: Calibration of geoid error models via a combined adjustment of ellipsoidal, orthometric and gravimetrical geoid height data. J. Geod. 79, 111–123 (2005) 3. Fotopoulos, G.: An analysis on the calibration of geoid, orthometric and ellipsoidal height data. UCGE Report No. 20183 (2003) 4. Gillins, D., Dennis, M.: Inclusion of leveling with GNSS observations in a single, 3-D geodetic survey network adjustment FIG working week 2017. In: Surveying the World of Tomorrow, Helsinki, Finland, 29 May–2 June 2017 5. Heiskanen, W.A., Moriz, A.: Physical Geodesy. Freeman and Company, San Franciso, WH (1967) 6. Hoang, N.H.: Adjustment of Geodetic and GPS Networks. Science and Technology Publishing House, Hanoi (2006) 7. Hoang, N.H., Truong, Q.H.: Basis of Mathematical Processing for Geodetic Data. Transport Publishing House, Ho Chi Minh City (2000). (in Vietnamese) 8. Kotsakis, C., Sideris, M.G.: On the adjustment of combined GPS/levelling/geoid networks. J. Geodesy 73(8), 412–421 (1999) 9. Le, V.H., Nguyen, X.H.: Combining EGM2008 global gravity model and leveling GPS altitude to improve the accuracy of GPS measurement results. J. Constr. Sci. Technol. Hanoi (2013). No. 3+4 10. Leick, A.: GPS Satellite Surveying, 2nd edn. A Wiley Interscience Publication. Wiley, Hoboken (1995) 11. Markuze, I.U., Hoang, N.H.: Adjustment of Terrestrial and Satellite Space Networks, p. 274. Nhedra Moscow Publishing House, Moscow (1991). (in Russian) 12. Markuze, I.U.: Basis of Geodetic Adjustment Computation. Nhedra Moscow Publishing House, Moscow (1990). (in Russian) 13. Nguyen, D.D., Dang, N.C.: Improving accuracy of abnormally EGM 2008 height using GPSlevelling data in Tay Nguyen and South Central Coasts. Earth Sci. J. (Vietnam) 34(1), 85–91 (2012) 14. Project: Modernizing the National Height Network to Serve the Planning, Construction, Socio- economic Development and Climate Change Response in a Number of Big Cities and Coastal Areas. Ministry of Natural Resources and Environment of Vietnam (2018)

Seismic Hazard Assessment for South-Central Region, Vietnam Trong Cao Dinh1(B) , Bach Mai Xuan1 , Hung Pham Nam1 , Tuan Thai Anh1 , Vuong Trong Kha2 , and Trieu Cao Dinh3 1 Institute of Geophysics, Vietnam Academy Science Technology,

A8/18 Hoang Quoc Viet Street, Hanoi, Vietnam [email protected] 2 Department of Mine Surveying, Hanoi University of Mining and Geology, 18 Pho Vien Street, Hanoi, Vietnam 3 Institute for Applied Geophysics, Vietnam Union of Science and Technology Associations, 210D Doi Can Street, Hanoi, Vietnam

Abstract. The objective of this paper is to present the studied results on determining seismic sources, maximum earthquake (Mmax ) and assess probabilistic seismic hazard for the south-central region of Vietnam. A combination of methods, including maximum gradient (Gmax ) of Bouguer gravity anomaly, Modified Gumbel Type-I, hybridizations of the Kijko and the OpenQuake-engine algorithms were employed. The results show that the seismic activity is characterized by the following quantities: b = 0.67, a = 3.87 and Mc = 2.6, whereas, the thickness of a seismic layer of about 16 km (from 2 km depth to 18 km depth). Seismic sources with the magnitude ≥ 5.0 are Tanh Linh, Thuan Hai - Minh Hai, Northern Cuu Long, Cuu Long, Southern Cuu Long, Eastern Phu Quy, Canh Duong, and Saigon River. The Mmax estimated is 5.47 with the modified Gumbel Type-I distribution, and 5.9 with the frequency - magnitude distribution. The earthquake with a maximum magnitude of 6.0 is likely to occur at Thuan Hai - Minh Hai fault. The Mmax value at the remaining faults varies between 5.0 ÷ 5.5. Overall, the typical components of the earthquake hazard in the south-central region have the following values: the peak ground acceleration (PGA) is 0.010 g ÷ 0.209 g; the peak ground velocity (PGV) is 0.0 cm/s ÷ 7.3 cm/s; the peak ground displacement (PGD) is 0.112 cm ÷ 1.680 cm at 10% probability of exceedance in 50 years. Keywords: Seismics · Hazard assessment · South-central region · Vietnam

1 Introductions Vietnam is a region of moderate seismicity. The northwestern part of the country is the most seismically active, while in the central and southern parts of Vietnam, the activity is low. The greatest strength observed magnitude of 6.7 (Tuan Giao Earthquake, 1983) occurring in Northwest Vietnam [1–3]. After this earthquake, the earthquake research is being more attention, and the most concern area is Northwest Vietnam, compared to © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 167–191, 2021. https://doi.org/10.1007/978-3-030-60269-7_9

168

T. C. Dinh et al.

other regions. Vietnam is probably still relatively separate from developed countries, such as Europe, Japan, and the United States. For example, the research orientation of the source structure of earthquakes, the doing for seismic hazards over the world, is concerned in many directions, but still at a preliminary level in Vietnam. In Vietnam, the most atlas of seismic hazard assessment is evaluated on the basis of probabilistic approach including studies such as Phuong [4, 5], Xuyen [6], Thanh [7]. These studies are based on probabilistic approaches [8], using common ground motion prediction equations such as in [9], and dedicated software CRISISS and EQRISK to calculate and construct the peak ground acceleration map in different return period with the scale of 1: 1.000.000. Also, at the same scale, Trieu [10] published the map of peak ground acceleration, velocity and displacement based on a Neo-deterministic approach for Vietnamese territories and some other areas. Before the year 2000, there were only two working seismic stations in the southcentral region of Vietnam, the first one was the Nha Trang station, which was built in 1957, whereas the second one was the Da Lat station, which was built in 1980. From 2010 to 2017, seven additional stations were built and had a total of 9 national working seismic stations. Due to the weakly active seismic activity, the research results of the earthquake in the south-central region were rarely mentioned, only some results of probabilistic seismic hazard assessment for the south-central region at the scale of 1:500 000 have been published [5]. Meanwhile, the south-central region has the highest economic growth rate in Vietnam. Some vital economic sectors such as oil and gas exploitation, petrochemical industry, energy (Ninh Thuan nuclear power plant, Ninh Thuan thermal power plant, Binh Thuan wind power plant, etc.); Vung Tau seaport; Binh Duong and Dong Nai economic zones; large cities with high population density have been built and developed. Therefore, the seismic hazard assessment at a large scale (1:200.000 scales, as in our studies in recent times) is essential and practical. In Vietnam in general and the south-central region in particular, the studies of determining seismic sources are mainly based on active faults in neo-tectonic, making the problem of determining source regions limited. In fact, not all active faults are capable of causing earthquake [5, 6, 11]. In addition, the determination of Mmax for the seismic source is mainly based on the seismic similarity problem [10]. The latest result of the maximum earthquake (Mmax ) prediction based on artificial neural network problem was obtained by authors in [12]. The FeedForward neural network with backpropagation algorithm was used to evaluate the reliability of the algorithm for some standard samples [13–15]. Input data for calculation include (1) Lineament density value; (2) Gradient value of Bouguer gravity field; (3) Gradient of an aeromagnetic anomaly; (4) Gradient of vertical movement of the Earth’s crust in the neotectonic period; (5) Gradient of sedimentary crust thickness; (6) Gradient of crystalline basement depth; and (7) Gradient of the Earth’s crust thickness. These data are assessed as directly related to the active fault and the earthquake magnitude. Standard samples are typical earthquakes with a magnitude greater than or equal to 4.5 (a total of 24 earthquakes as standard samples). The first task consists of the investigation and construction of network structure. The method of result checking - evaluation was used for each obtained network structure (number of layers, number of neurons in each layer, type of propagation function, and estimation of error of neural network). Modules of MATLAB software

Seismic Hazard Assessment for South-Central Region, Vietnam

169

were employed to develop, select and evaluate results. The correlation coefficient value was utilized for a neural network assessment, and when this value is closer to one, the network is chosen. In this paper, the maximum gradient (Gmax ) of Bouguer gravity anomaly is employed to delineate seismic sources, Modified Gumbel Type-I and hybridizations of the Kijko are applied to estimate Mmax , and the probabilistic approach is applied to assess seismic hazard by using OpenQuake-engine Software [16], which is evaluated as a modern software suite in calculating earthquake hazard, but rarely to be used in Vietnam.

2 Study Area and Data 2.1 Geological Setting of the Studied Area The study area is the south-central region of Vietnam (Fig. 1). Tectonically, this region is located within the Da Lat sub-block, belonging to the Da Lat - Can Tho modern geodynamic block, where the Saigon River fault acts as the southern boundary with the Can Tho sub-block. Meanwhile, the Da Lat - Can Tho block is located in the east - southeast of the Indosinian microplate, which is a block-type differential uplift and denuded in Cenozoic [2, 3]. The thickness of the Earth’s crust in Da Lat sub-block tends to increase gradually from the southeast (about 20 km in Vung Tau coastal area) to the northwest (Fig. 1) and reaches 32 km in Da Lat [1, 3].

Fig. 1. The thickness of the Earth’s crust and manifestations of the earthquake and geothermal activities in the south-central region: 1) Volcano in Holocene; 2) Hot water occurrence on the surface; 3) Isopachs of the Earth’s crust; 4) Fault

170

T. C. Dinh et al.

Based on the manifestation of volcanic activity in Neogene (Dong Nai, Hon Tro Island), hot water occurrence along the coast from Ninh Thuan to Vung Tau and high geothermal anomalies in Vung Tau coastal area, Trieu Cao Dinh and et al. [2] suggested that the Earth’s crust in this subzone is strongly active in the modern period. 2.2 Seismic Activity of the Study Area 2.2.1 The Earthquake Catalog Based on previous works [6, 10], the seismic activity in the south-central region is not strong compared to that in the Northwest region in Vietnam, where the maximum magnitude was less than 6.0. According to historical records, some earthquakes with M = 5.0 ÷ 5.5 also occurred in this area before the 20th century, for example in Binh Thuan province in 1715 (M = 4.0 ÷ 5.0) and in 1877 (M = 5.0 ÷ 5.5), and in Ninh Thuan province in 1882 (M = 5.0 ÷ 5.5). The Institute of Geophysics (Vietnam Academy of Science and Technology) is responsible for managing the earthquake data in Vietnam and updating annual earthquake data [6, 16]. Due to the wars and the lack of attention to earthquakes in historical records, the earthquake catalog obtained in the Institute of Geophysics (Vietnam) is insufficient in data, and the accuracy of magnitude determination is also low.

Fig. 2. Earthquake epicenter distribution and Mmax forecast for the south-central region (Vietnam) on the basis of the neural network method [9]

Seismic Hazard Assessment for South-Central Region, Vietnam

171

The error of magnitude determination according to historical records and people surveys is probably not less than 1.0 magnitude unit. Only after 2002, with seven earthquake observation stations established in this area, there has been sufficient information about earthquakes with Magnitude ≥ 2.0 [5, 10, 11]. Therefore, in order to have the best earthquake catalog for seismic hazard assessment in the south-central region, the authors have collected more earthquake data from ISC (International Seismological Centre). CN algorithm [10, 15, 17] was used to analyze, compare and produce the most reliable earthquake catalog from 1900 to 2019 (Fig. 2), to serve the seismic source determination, Mmax prediction and seismic hazard assessment in the south-central region. The number of earthquakes in the study area is 214, with a minimum magnitude of 2.5 (Appendix 1). 2.2.2 Characteristics of Seismic Activity The earthquakes in the south-central region mainly occur in the coastal zone from Nha Trang city to Vung Tau city. According to historical data, the earthquake clusters with magnitude M = 5.0 ÷ 5.5 is mainly concentrated along the coast of Binh Thuan province - Ninh Thuan province. Meanwhile, the earthquake cluster recorded in the period of 2004–2010 with the maximum observed magnitude M = 5.3 is concentrated in the sea of Vung Tau city (Fig. 2). The magnitude of completeness (Mc) is an important value, indicating the minimum threshold of seismicity in the study area. The selection of Mc threshold depends on the difference between Gutenberg-Richter (GR) function value obtained in practice and that obtained from statistics. The Mc value in the South-central region was determined to be 2.6. The difference between GR function input value and calculated value with the lowest MSE (Mean Square Error) reached 0.01183 (Table 1).

Fig. 3. Earthquake hypocenter distribution according to the depth (reflecting the seismic layer)

172

T. C. Dinh et al. Table 1. Magnitude interval and MSE for calculating GR function Magnitude interval MSE 2.5–5.3

0.01227

2.6–5.3

0.01183

2.7–5.3

0.01206

2.8–5.3

0.01312

2.9–5.3

0.01428

3.0–5.3

0.01566

GR distribution function in the study area has the following form with a calculated magnitude of completeness: logN = 3.87 − 0.67 ∗ M

(1)

Based on the function (1) we have: b value = 0.67 and a value = 3.87. The depth of the seismic layer is represented in Fig. 2, which was determined to be approximately 16 km (from 2 km depth to 18 km depth, Fig. 3), and based on the distribution between the hypocenter depth and the number of earthquakes with hypocenter located at the corresponding depth.

3 Background of the Methods Used 3.1 Determination of Maximum Horizontal Gradient The determination of maximum horizontal gradient firstly was proposed by authors Blakely and Simpson in [18] for identifying the boundaries between geological bodies with different densities. This is one of the highly effective methods applied [19, 20] in parallel with signal analytic and vertical derivative approaches [21, 22]. The maximum total horizontal gradient (Gmax) at each point (the central point compared to eight neighboring points in four directions of the square grid) on the data grid of Bouguer gravity anomaly g(x,y) was calculated by the formula (2):     2 g(x,y) 2 G= (2) + g(x,y) dx dy

3.2 Analytic Hierarchy Process (AHP) When there are many factors representing an object, but their roles are not exactly the same, the weight determination for each of these factors is essential. The analytic hierarchy process (AHP) (also known as a p-weight model) is a semi-quantitative method which was constructed and developed by Saaty [14]. For the zonation proposes, this

Seismic Hazard Assessment for South-Central Region, Vietnam

173

method is widely used [23–27]. This method involves constructing a system of pairwise comparison matrices between different factors for active fault. This approach can be described as the hierarchy of importance of factors in active fault identification; each those is compared to the others to determine their importance. The comparison value between two factors is based on the number values: 1 for equally important; 3 for slightly important; 5 for much more important; 7 for very much more important and 9 for absolutely more important. In terms of the quality of the final decision, the value is related to a judgement’s consistency, as validated by an expert. The measurement of the judgement’s consistency is obtained by calculating the consistency ratio (CR). A matrix is consistent if its elements observe the assumption of transitivity and proportionality of the preferences. A reasonable level of consistency in the paired comparisons is a CR of 0.1 or less [28]. 3.3 Modified Gumbel Type-I Extreme Value Distribution in M max Assessment Gumbel distribution or Gumbel Type-I extreme value distribution firstly was introduced by Gumbel in 1935 [29, 30]. Normally only the Gumbel asymptotic distribution with the upper limit, Gumbel Type-I, Type-III and modified Type-I distributions were used. The modified Type-I asymptotic distribution (3) is more effective in the Mmax evaluation compared to other distributions [31, 32]. With the simple constructing of the Gumbel distribution function, only the earthquake list of the study area is needed, in the M max assessment, the Gumbel problem is widely used in the world [33–36]. If X is considered as a random variable with distribution function F(X ) = P{X ≤ x}, then the probability that x is the largest among n independent samples of the distribution F(X ) will be G(x) = P{X1 ≤ x, X2 ≤ x, . . . , Xn ≤ x} = F n (x), which is the distribution function of Gumbel extreme values. The extremes values are obtained by dividing the time series into equal intervals and selecting the maximum values in each of these intervals. Afterwards, the newly obtained extreme values are arranged in ascending order, which means building a sequence of numbers with the x values as follows: x1 < x2 < x3 < . . . < xN . The values of G(x) for different extreme values x are calculated as follows. The observed time series is divided into N equal intervals, and the maximum values of x in each these intervals are selected. In seismology, if the extreme values are calculated in magnitude, they are the maximum magnitudes in each interval; so at the jth position the value of x is xj , corresponding to   j G xj = N +1 , j = 1, 2, . . . , N .  −βx − e−βv  −βu e , with β > 0; v > u > 0 (3) G(x) = exp −e 1 − e−βv where v is the upper limit of the magnitude of earthquakes occurring, β and u are the parameters of the required distribution function. 3.4 Estimation of the Maximum Earthquake Magnitude on the Basis of Frequency - Magnitude Relationship The maximum earthquake magnitude assessment based on the frequency-magnitude distribution was firstly proposed in 2004 by Kijko [31]. Following this scientific direction,

174

T. C. Dinh et al.

the author is completing his approach [37–39], and the research results are widely used by scientists around the world [40–42]. The maximum earthquake magnitude in a study area Mmax is defined as the upper limit of earthquake magnitude for a particular area, which means the magnitude of the strongest earthquake that is likely to occur. Mmax obs + , where  is the positive correction coefficient, M obs is value is estimated by Mmax max the maximum earthquake magnitude observed in the study area. Based on the characteristics of seismic activity in South-central region, the three examined cases are as follows: 1- the earthquake magnitudes are distributed according to the doubly-truncated Gutenberg-Richter relation; 2- the empirical magnitude distribution deviates moderately from the Gutenberg-Richter relation; and 3- no specific form of the magnitude distribution is assumed, and only a few of the largest magnitudes are known. In the second case, the Mmax determination based on the Gutenberg-Richter distribution with an incomplete earthquake catalog was applied in this study: Suppose that in the study area, n earthquakes with magnitude greater than or equal to Mmin (is known and is denoted as the threshold of completeness) were recorded in the period of time T . Assume further that the magnitudes are independent, identically distributed, and have random values with cumulative distribution function (CDF), FM (m). The unknown parameter Mmax is the upper limit of the range of magnitudes and is thus termed the maximum earthquake magnitude, and is to be estimated. obs + Mmax = Mmax

Mmax

[FM (m)]n dm

(4)

Mmin

in which the desired Mmax appears on both sides. An estimated value of Mmax , M max can be obtained through iteration. Maximum regional magnitude mmax is calculated according to the procedure by Kijko-Sellevoll-Bayes [43]. Based on the Bayesian distribution and the Gutenberg Richter CDF distribution of earthquake magnitude, it is possible to build the Bayesian version of Mmax estimator from the integral of correction coefficient :   q n  n Mmax p 1−  = Cβ dm (5) p + M − Mmin Mmin After applying the Cramér approximation, (4) can be expressed by:

   δ 1/q exp nr q /(1 − r q )  obs ˆ  −1/q, δr q − (−1/q, δ) Mmax = Mmax + (6) β 

 where r = p/(p + Mmax − Mmin ); c1 = exp −n 1 − Cβ ; δ = nCβ and (., .) is the complementary incomplete gamma function; Cβ is a normalizing coefficient equal 2  2  q −1 

; p = β/ σβ and q = β/σβ . The symbol β to 1 − p/(p + Mmax − Mmin ) denotes the known mean value of parameter β (β = bln(10)); and b is the parameter of the frequency-magnitude (Gutenberg-Richter) relation and σβ is the known standard deviation of β.

Seismic Hazard Assessment for South-Central Region, Vietnam

175

3.5 Seismic Hazard Calculation The initial version of the OpenQuake-engine was based on OpenSHA [44], an objectoriented based, state of a library for the hazard calculation. In the active and diverse community of probabilistic seismic hazard analysis (PSHA) modellers and users, the Global Earthquake Model (GEM) initiative [45] is promoting the creation of open and transparent tools for seismic-hazard and risk assessment [46–48]. Seismic source model (SSM) is defined as a set I of separate seismic sources (Src), and for each source [23], it is a set of rupture segments of separate earthquakes J :  SSM = {Src1 , Src2 , . . . , Srci },i = 1, I (7) Srci = Rupi1 , Rupi2 , . . . , Rupij , j = 1, J The probability that ground motion parameter X exceeds a value x, at least once in a period of time T , P(X ≥ x|T ) can be calculated as 1 minus the probability that no source causes the ground motion exceedance. By further assuming that each source that causes the rupture in the earthquake is independent, we have: P(X ≥ x|T ) = 1 −

Ji I  

Prupij (X x|T )

(8)

i=1 j=1

where Prupij (X < x|T ) is the jth probability of rupture in the ith source not causing the exceedance and Ji is the total number of ruptures caused by the ith source. By using the total probability theorem, formula (8) can be written as: Prupij (X < x|T ) =

∞ 

k  Prupij (k|T ) ∗ P X < x|rupij

(9)

k=0

where Prupij (k|T ) is the j th probability of rupture in the ith source occurring for the k th   time in a period T and P X < x|rupij is the conditional probability that the parameter X does not exceed the value x with the occurrence of rupij . Suppose that the seismic sources are independent of each other and the rupture in the earthquake is caused by each source, we can calculate the probability that ground motion exceeds a value x, at least once in a period T as follows: P(X ≥ x|T ) = 1 −

Ji  ∞ I  

k  Prupij (k|T ) ∗ P X < x|rupij

(10)

i=1 j=1 k=0

Equation (10) allows the probabilistic seismic hazard analysis (PSHA) over time.

4 Methodology for the Seismic Hazard Assessment in the South-Central Region, Vietnam Based on the seismic hazard assessment, the proposed methodology consistes of two main steps: determination of the seismic source; prediction of maximum earthquake and assessment of the seismic hazard.

176

T. C. Dinh et al.

Step 1: Determination of the Seismic Source In this analysis, three following points were used to determine the seismic source in south-central region, including (1) Determination of deep geological fault zones; (2) Identification of active fault zones; and (3) Determination of the seismic zone. For the determination of deep geological fault zones, in this research, based on geological-geophysical data, we only calculated the maximum gradient (Gmax ) of Bouguer gravity anomaly, combined regarding the previous research results. The linear distribution of Gmax will reflect the existence of deep geological faults [1–3]. Mmax values at upward continuation to 2 km, 6 km and 10 km were calculated with the aim of investigating the dip direction of the fault. It is suggested that: (a) The upward-continued field component will reflect the structure at the corresponding depth; (b) The movement trend of Gmax , according to the depth, will reflect the dip direction of the fault [1]. Regarding the identification of active fault zones, typically, a fault that is determined to be active must have at least one of the following indications [3]: (1) Manifestations of various vertical movements towards both walls of the fault in the present: strong (greater than or equal to 5 mm/year); moderate (from 1 to 5 mm/year) and weak (less than or equal to 1 mm/year); (2) Manifestations of horizontal slip (sinistral strike-slip or dextral strike-slip): strong (greater than or equal to 5 mm/year); moderate (from 1 to 5 mm/year) and weak (less than or equal to 1 mm/year); (3) Manifestations of land cracking - subsidence, landslide, and erosion due to endogenous causes (for strongly active faults in the last period); (4) Manifestations of seismic activity and coinciding with the boundary zone of modern geodynamic blocks; (5) Manifestations of young folding: anticline, syncline, flexure, and alteration zone of topographic and geomorphological elements; (6) Manifestations of volcanic activity (in Holocene and Quaternary) and gasemitting area related to seismic activity or area of concentrated young tectonic fractures; and (7) Manifestations of hot water activity, zone with high geothermal gradient. For the establishment of criteria for active fault identification: the identification problem was employed with the aim of using the standard sample (the active fault that generated the earthquake), thereby comparing the studied fault with the standard sample to establish criteria for active fault identification. The standard sample used for the problem is Son La fault, particularly the segment running through Tuan Giao town, where the Tuan Giao earthquake occurred in 1983 [15]. The magnitude of this earthquake was determined to be 6.7, and it is the only earthquake in Vietnam whose focal mechanism was most detailed and quantitatively studied. The results of this earthquake study can be used as a model in assessing the applicability of empirical formulas in the world to Vietnam as well as in analyzing the criteria for active fault identification. Based on the study conducted by the author in [15] on fault segmentation in this earthquake, a system with 9 criteria was introduced for active fault identification in order to calculate the maximum earthquake in Vietnam. In southcentral region, there are only seven typical criteria to identify the active fault as follows: SG1 - Topographic slope signifying topographic dissection; SG2 - Density of hot water occurrence points in the area of a grid cell divided according to the coverage area; SG3 - Length-density of rivers; SG4 - Lineament density according to the total length; DH5 - Density of landslide points; SG6 - Density of Gmax points, and SG7 - Length-density of geological faults.

Seismic Hazard Assessment for South-Central Region, Vietnam

177

The classification of the activity level of the fault in the south-central region indicates the areas with different characteristics of an active fault. The fault activity was classified into three levels: (1) Inactive (or weakly active), (2) Moderately active fault, (3) Strongly active fault. The component factors such as lineament density, geological fault, hot water point, etc. were all classified into three levels according to the above levels of an active fault. Determination of the seismic zone is defined as the active fault zone, which generated the earthquake with magnitude M ≥ Mc - the magnitude of completeness [3, 10]. 4.1 Prediction of Maximum Earthquake and Assesment of the Seismic Hazard Because earthquake data in the south-central region is limited and unevenly distributed spatially and temporally, earthquake prediction by normal statistical approach encounters many difficulties. In an attempt to estimate the maximum earthquake magnitude with the highest reliability, the following approaches were utilized: (1) The modified Gumbel Type-I extreme value distribution [32, 49]; (2) Estimation of the maximum earthquake magnitude on the basis of frequency - magnitude relationship [31, 37–39] combined with reference to the results of Mmax prediction using artificial neural network [12]. OpenQuake-engine software (using the ground motion attenuation function from Campbell-Bozorgnia 2008 and seismic source model-SSM) was employed for the first time in seismic hazard assessment in the South-central region [43, 50, 51].

5 Results and Discussion 5.1 Determination of Seismic Zones The calculation results are shown in Fig. 4, and the difference between the previously determined fault system and Gmax distribution can be easily recognized [3]. This difference is due to the scale of the study and the detailed level of data. The study conducted by authors in [3] was carried out for the whole territory of Vietnam at a scale of 1:1.000.000; meanwhile, the detailed level of data is at 1:200.000. From the viewpoint that Gmax reflects the location of a deep geological fault, it was utilized as one of the important quantities for determining active fault. Based on the component maps, through the superposition, and using ArcGIS 10.0 software, the authors have established the map of active faults at a scale of 1:200.000 (Fig. 5). The weight of component factors was calculated correlatively according to the formula of landslide suitable index: LSI = 0.3215 * SG1 + 0.2313 * SG2 + 0.2172 * SG3 + 0.1040 * SG4 + 0.0644 * SG5 + 0.0398 * SG6 + 0.0217 * SG7. The reliability of weight evaluation is expressed by CR (consistency ratio) = 0.025. The result in Fig. 5 indicates the existence of 19 active fault zones in the southcentral region, i.e., 10 northeast-southwest faults: Dong Xoai, Tanh Linh, Gia Ray - Van Gia, Thuan Hai - Minh Hai, Northern Cuu Long, Cuu Long, Southern Cuu Long, Phu Quy, Eastern Phu Quy, and Canh Duong, 8 northwest-southeast faults: Nha Trang, Ninh Phuoc, Cho Lau, Duc Pho - Ham Minh, Loc Ninh - Ham Tan, Thien Tan - Binh Son, Dong Nai River, and Saigon River, and one longitudinal fault: Loc Ninh - Saigon.

178

T. C. Dinh et al.

Fig. 4. Distribution of deep geological faults [3] and maximum gradient value of Bouguer gravity field at different depths: 1-Normal fault; 2-Reverse fault; 3-Gmax at upward continuation to 2 km; 4-Gmax at upward continuation to 6 km; and 5-Gmax at upward continuation to 10 km

Fig. 5. Earthquake epicenter and activity level of faults: 1-Inactive; 2-Moderately active; 3-Strongly active; 4-Active faults identified by geomorphic principle

Seismic Hazard Assessment for South-Central Region, Vietnam

179

The active fault zone that generated the earthquake with M ≥ Mc is accepted as the seismic zone (see Fig. 7 below). 5.2 Maximum Earthquake Prediction The calculation results by Modified Gumbel Type-I extreme value distribution in Mmax assessment are represented in Fig. 6, and the maximum magnitude value is 5.47.

Fig. 6. Graph of modified Gumbel Type-I distribution and Mmax value

The calculation results are: β = 0.2371; u = 4.742; v = 5.47; R-squared (Coefficient of determination) = 0.9712. The result of the estimation of the maximum earthquake magnitude based on frequency - magnitude relationship indicate Mmax = 5.90 ± 0.68 obs = 5.30 ± 0.25). (for Mmax The research results conducted by authors in [6, 10] showed that Mmax in southcentral region does not exceed 6.0 (Mmax ≤ 6.0). Mmax prediction based on artificial neural network problem indicated Mmax = 5.5 [12]. The result of modified Gumbel Type-I asymptotic distribution shows Mmax = 5.47. Based on the frequency - magnitude distribution, Mmax = 5.90 ± 0.68. The above-mentioned results allow estimating the maximum earthquake magnitude in the sources of the south-central region as follows: Tanh Linh (Mmax = 5.0); Thuan Hai - Minh Hai (Mmax = 6.0); Northern Cuu Long (Mmax = 5.0); Cuu Long (Mmax = 5.5); Southern Cuu Long (Mmax = 5.5); Eastern Phu Quy (Mmax = 5.5); Canh Duong (Mmax = 5.5); and Saigon River (Mmax = 5.5).

180

T. C. Dinh et al.

Table 2. Seismic hazard parameters estimated for the seismic sources in the south-central region Seismic source

Rake Dip Depth to the top of Depth to the Maximum the seismogenic bottom of the earthquake layer seismogenic layer magnitude (M max )

Tanh Linh

180°

60° 2 km

18 km

5.0

70°

90° 2 km

18 km

6.0

Northern Cuu Long

180°

60° 2 km

18 km

5.0

Cuu Long

180°

60° 2 km

18 km

5.5

Southern Cuu Long

180°

60° 2 km

18 km

5.5

Eastern Phu Quy

180°

60° 2 km

18 km

5.5

Canh Duong

180°

50° 2 km

18 km

5.5

Saigon River

180°

70° 2 km

18 km

5.5

Thuan Hai – Minh Hai

Fig. 7. Seismic sources in the south-central region: 1)-Seismic source; 2)-Maximum earthquake magnitude (M max ) in the source

Seismic Hazard Assessment for South-Central Region, Vietnam

181

5.3 Probabilistic Seismic Hazard Analysis The 10% probability of exceedance in 50 years was used in PSHA in the south-central region (Table 2), and the results are represented in Figs. 8, 9, and 10. a) PSHA (10% probability of exceedance in 50 years) in the south-central region show the high values in the coastal zone, from Ninh Thuan to Vung Tau: peak ground acceleration PGA = 0.160 g ÷ 0.209 g (corresponding to levels VII-VIII on MSK-64 scale); peak ground velocity PGV = 5.0 cm/s ÷ 7.3 cm/s; peak ground displacement PGD = 1.40 cm ÷ 1.68 cm. Meanwhile, in the coastal zone of the South-central region, there is an explosion of petrochemical industrial zones, energy industry (Ninh Thuan nuclear power plant, Ninh Thuan thermal power plant, Binh Thuan wind power plant, etc.) and seaports (Vung Tau, Cam Ranh). Therefore, the seismic microzoning (at 1:50.000 or greater scales) in this zone is extremely essential and practical.

Fig. 8. Peak ground acceleration (PGA) in the south-central region

b) In Cuu Long basin - the area of oil and gas exploitation, the values of PSHA (10% probability of exceedance in 50 years) reach level VII (MSK-64): peak ground acceleration PGA = 0.120 g ÷ 0.140 g; peak ground velocity PGV = 4.0 cm/s ÷ 6.0 cm/s; peak ground displacement PGD = 1.30 cm ÷ 1.50 cm. c) On the mainland of south-central provinces, the values of PSHA (10% probability of exceedance in 50 yr) only reach levels VI-VII (MSK-64): peak ground acceleration PGA = 0.090 g ÷ 0.110 g; peak ground velocity PGV = 2.0 ÷ 4.0 cm/s; peak ground displacement PGD = 0.80 cm ÷ 1.20 cm.

182

T. C. Dinh et al.

Fig. 9. Peak ground velocity (PGV) in the south-central region

Fig. 10. Peak ground displacement (PGD) in the south-central region

6 Concluding Remark This study assessed the probabilistic seismic hazard for the south-central region of Vietnam. Based on the findings, some conclusions are summarized as follows:

Seismic Hazard Assessment for South-Central Region, Vietnam

183

• Earthquakes in the south-central region are mainly concentrated in the coastal area and the sea of Vung Tau, where the oil and gas exploitation has been carried out. The maximum observed earthquake magnitude was M = 5.3 (2005). Meanwhile, according to the historical data, the magnitude of earthquakes in coastal areas of Binh Thuan (in the years of 1715, 1877) and Ninh Thuan (in the year 1882) did not exceed 5.5. The seismic activity is characterized by the following quantities: b = 0.67; a = 3.87, Mc = 2.6, the seismic layer thickness of about 16 km (from 2 km depth to 18 km depth). • A total of 19 active faults were determined: Dong Xoai, Tanh Linh, Gia Ray - Van Gia, Thuan Hai - Minh Hai, Northern Cuu Long, Cuu Long, Southern Cuu Long, Phu Quy, Eastern Phu Quy, Canh Duong, Nha Trang, Ninh Phuoc, Cho Lau, Duc Pho - Ham Minh, Loc Ninh - Ham Tan, Thien Tan - Binh Son, Dong Nai River, Saigon River and Loc Ninh – Saigon. If the seismic zone is defined as the active fault zone that generated the earthquake with M ≥ Mc, there are eight seismic sources in south-central region (M ≥ 5.0), including 1) Tanh Linh; 2) Thuan Hai – Minh Hai; 3) Northern Cuu Long; 4) Cuu Long; 5) Phu Quy; 6) Eastern Phu Quy; 7) Canh Duong, and 8) Saigon River. • The maximum earthquake (Mmax ) estimation by several different methods obtains the following results: a) According to the artificial neural network, Mmax = 5.5; b) According to the modified Gumbel type-I distribution, Mmax = 5.47; c) According to the frequency - magnitude distribution, Mmax = 5.9. The earthquake with the maximum magnitude is likely to occur at Thuan Hai - Minh Hai source (Mmax = 6.0). The Mmax values at the remaining sources vary between 5.0 ÷ 5.5, with the lowest values recorded at Tanh Linh and Northern Cuu Long sources (Mmax = 5.0). • PSHA (10% in 50 years) in south-central region shows the typical values as follows: peak ground acceleration PGA = 0.010 g ÷ 0.209 g; peak ground velocity PGV = 0.0 cm/s ÷ 7.3 cm/s; and peak ground displacement PGD = 0.112 cm ÷ 1.680 cm. The values of PSHA are high in the coastal zone, reaching levels VII-VIII from Ninh Thuan to Vung Tau; whereas those on the mainland only reach levels VI-VII (MSK-64).

Acknowledgements. The research team would like to convey their sincere thanks to the Ministry of Science and Technology for their funding for the implementation of the research themes of Governmental Science and Technology Project coded KC.09.38/16–20.

Appendix 1: Earthquake Catalog for South-Central Region, Vietnam Number 1 2 3 4 5

Year 1923 1923 1966 1966 1990

Month 2 5 2 2 10

Date 15 02 22 21 15

Longitude 109 109 109.9 109.9 107.48

Latitude 10.1 10.1 12.8 12.8 10.4

Depth 10 17 16 18 17.4

Magnitude 5.1 5.1 3.3 3.3 3.7 (continued)

184

T. C. Dinh et al.

(continued) Number 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 39 40 41 42 43 44 45 46 47

Year 2002 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2006 2006 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007

Month 8 12 9 9 11 5 10 12 9 12 8 10 11 8 11 11 8 12 7 3 1 7 6 11 3 11 11 11 11 2 1 11 11 11 2 1 2 2 8 9 2 2

Date 26 16 12 19 11 10 17 27 26 18 11 17 07 05 08 07 05 04 03 18 26 22 15 28 23 29 29 28 29 15 20 28 28 28 10 07 15 20 24 27 03 09

Longitude 107.23 108.45 108.57 108.6 108.28 108.61 108.31 108.53 108.41 108.5 108.54 108.47 108.31 108.38 108.26 108.26 108.37 108.316 108.39 108.471 108.742 108.834 108.319 109.92 107.422 109.664 109.225 107.899 108.297 108.285 108.227 108.273 108.388 108.109 108.115 108.271 108.302 108.216 108.328 109.156 108.22 108.359

Latitude 10.3 10.02 10.39 9.96 9.64 9.94 10.39 9.99 9.94 10.07 10.31 10.34 10.02 9.98 10.12 10.08 9.99 10.218 9.87 11.027 9.914 9.935 9.991 10.542 9.601 10.354 10.083 10.053 9.967 10.077 10.087 10.014 10.037 10.061 10.138 10.06 10.459 10.332 9.688 9.961 10.166 10.617

Depth 10 7 7 7 10 10 10 10 10 10 10 10 10 10 10 12 16 10 17.7 2.1 2.7 4.1 4.4 4.6 7.8 8.5 8.7 8.9 11.4 14.3 4.2 4.9 5.4 6.8 8.4 9.5 10 10 10 10 10 10

Magnitude 3.7 3.6 3.7 3.7 3 3.1 3.1 3.1 3.3 3.3 3.6 4 4 4.5 5.3 5.2 4.4 2.5 4.3 2.7 3.3 3 3.8 3.9 2.7 3.3 3.5 3.6 4 3.5 3.2 4.5 4.9 2.9 2.7 2.8 2.5 2.5 2.5 2.5 2.6 2.6 (continued)

Seismic Hazard Assessment for South-Central Region, Vietnam (continued) Number 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

Year 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2008 2008 2008 2008 2008 2009 2009 2009 2009 2009 2009 2009 2009 2009 2010 2010 2010 2010 2010

Month 3 5 6 2 2 3 4 11 12 1 7 7 1 5 8 9 3 4 1 4 11 1 3 4 3 8 8 9 4 7 4 6 9 7 9 6 7 1 9 6 11 11

Date 23 15 15 04 27 23 29 28 12 13 07 07 21 03 10 08 12 29 08 29 29 26 30 09 10 23 27 07 18 25 20 28 20 28 22 16 25 01 25 14 04 05

Longitude 108.275 108.248 108.295 108.208 108.227 108.248 107.732 108.039 108.136 108.383 108.54 108.368 108.264 108.563 108.666 108.78 108.286 107.597 108.257 107.435 108.865 107.939 107.421 107.964 108.833 108.31 109.568 108.182 108.187 108.288 108.266 108.04 108.167 108.188 108.184 108.022 108.161 107.913 108.23 108.174 108.126 108.249

Latitude 10.298 9.922 10.058 10.568 10.052 10.689 9.726 9.819 9.787 10.299 9.651 10.036 10.425 10.871 9.653 9.804 10.498 9.678 10.118 9.692 10.152 10.172 9.906 10.195 9.869 9.994 11.189 10.089 9.975 10.033 10 9.907 9.908 10.065 10.201 9.826 10.119 9.645 9.864 9.773 9.746 9.862

Depth 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 11.2 16.4 16.5 17.1 17.1 10 10 10 12.4 17 10.2 3.7 5 5 5 10 16.8 16.9 17 5.2 5.5 5.7 6.3 7.5

185

Magnitude 2.6 2.6 2.6 2.7 2.7 2.7 2.7 2.7 2.7 2.8 2.8 2.8 2.9 2.9 2.9 2.9 3.2 3.3 2.8 3.5 2.9 2.6 2.6 2.7 2.8 3 2.8 2.7 2.7 4.1 2.5 2.5 2.7 4 4.3 2.7 3.7 3.1 3.3 3.7 2.6 3.7 (continued)

186

T. C. Dinh et al.

(continued) Number 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131

Year 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010

Month 11 7 2 8 3 6 1 3 3 3 3 4 4 4 4 4 5 5 6 6 6 6 6 7 12 2 4 4 4 4 5 5 6 8 9 10 11 11 12 5 6 6

Date 05 31 27 27 03 23 07 11 19 20 22 08 09 09 10 25 14 21 15 19 20 21 28 03 31 05 07 13 19 28 22 24 14 16 22 20 04 04 20 27 10 25

Longitude 108.197 108.225 108.122 107.944 108.123 108.136 108.056 108.233 108.269 108.204 108.194 108.247 108.276 108.227 108.324 108.28 108.209 108.113 108.159 108.297 108.292 108.243 108.213 108.207 108.264 108.189 108.243 108.213 108.189 108.253 108.275 108.259 108.254 108.281 108.203 108.236 108.222 108.23 108.269 108.296 108.244 108.26

Latitude 9.998 9.979 9.85 9.661 9.82 9.872 9.824 9.888 9.955 9.83 9.827 9.905 10.007 9.912 10.059 10.007 9.805 9.745 9.802 10.024 10.009 9.914 9.853 9.854 9.942 10.057 9.903 9.85 9.867 9.922 9.975 9.949 9.929 9.959 9.966 9.9 9.89 9.899 9.949 10.016 9.909 9.944

Depth 10.5 10.8 11.9 13.8 3.3 3.9 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

Magnitude 2.7 2.9 2.6 2.6 2.6 4.2 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.7 2.7 2.7 (continued)

Seismic Hazard Assessment for South-Central Region, Vietnam (continued) Number 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173

Year 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2010 2011 2011 2011 2011 2011 2011 2011

Month 7 7 9 4 6 6 8 11 4 5 10 6 2 9 11 11 12 7 12 12 7 12 1 12 12 7 6 11 11 11 6 11 11 10 7 2 2 5 1 1 1 2

Date 15 22 06 09 02 24 16 18 20 28 08 30 25 15 06 12 13 09 31 04 24 31 18 05 31 09 29 05 28 05 08 04 06 27 02 19 20 12 10 10 11 07

Longitude 108.211 108.247 108.241 108.274 108.114 108.24 108.141 108.253 108.258 108.16 108.044 108.143 108.077 108.237 108.199 108.229 108.229 108.282 108.232 108.234 108.253 108.199 108.219 108.273 108.269 108.281 108.224 108.214 108.234 108.204 108.292 108.208 108.195 108.201 108.238 108.288 108.146 108.241 108.332 108.289 108.314 108.26

Latitude 9.84 9.921 9.956 10.018 9.816 9.893 9.748 9.902 9.969 9.864 9.717 9.841 9.991 9.871 9.915 9.885 9.885 9.953 9.895 9.899 9.925 9.903 10.073 9.951 9.96 9.96 9.925 9.928 9.993 9.935 10.015 9.968 9.976 10 9.93 10.002 9.74 9.899 10.09 10.04 10.045 9.905

Depth 5 5 5 5 5 5 5 5 5 5 5 5 5 5 6.2 7.3 7.3 9 9.4 9.6 9.9 9.9 10 10.7 10.9 11.5 11.6 12.6 13.5 13.7 13.9 16.8 17 17 14.7 12.2 12.5 3.4 5 5 5 5

187

Magnitude 2.7 2.7 2.7 2.8 2.8 2.8 2.8 2.8 2.9 2.9 2.9 3.1 3.2 4 3.8 2.6 2.6 2.6 3.2 2.5 2.5 3.1 3.8 2.5 2.6 2.9 2.9 3.1 2.7 3.3 2.6 3 2.5 2.9 2.5 2.5 2.5 2.9 2.5 2.5 2.5 2.5 (continued)

188

T. C. Dinh et al.

(continued) Number 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214

Year 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2012 2013 2013 2013 2014 2014 2014 2015 2016 2016

Month 4 4 1 1 3 2 2 2 6 1 2 1 1 1 6 2 4 3 7 2 1 2 1 2 1 1 3 5 2 1 1 2 7 7 6 12 12 12 1 11 7

Date 05 19 10 24 28 17 26 25 06 27 26 23 14 22 21 26 25 14 02 16 21 11 14 28 26 17 28 28 03 26 14 15 28 28 19 13 11 19 19 23 25

Longitude 108.265 108.213 108.203 108.226 108.24 108.223 108.253 108.271 108.143 108.201 108.253 108.26 108.229 108.232 108.223 108.221 108.224 108.243 108.143 108.249 108.237 108.257 108.262 108.243 108.237 108.247 108.295 108.291 108.269 108.17 108.294 108.34 108.724 108.245 108.024 108.329 108.419 108.292 108.175 107.88 108.03

Latitude 9.921 9.865 9.83 9.871 9.914 9.87 9.927 9.985 9.838 9.865 9.893 9.889 9.882 9.853 9.875 9.876 9.879 9.932 9.88 9.92 9.911 9.909 9.947 9.916 9.946 9.919 10.018 10.016 9.944 10.066 9.943 9.77 12.057 11.785 9.738 10.001 10.281 10.717 10.863 9.543 10.912

Depth 5 5 5 5 5 5 5 5 5 6.5 6.6 6.9 7 7.1 7.3 8.1 8.1 9.3 9.5 9.7 9.7 10 10.3 10.4 10.9 11.1 11.8 12.7 13 17 17.8 17.5 3.1 10 15.2 17.1 17.1 17.1 15 10 11.7

Magnitude 2.5 2.5 2.6 2.6 2.6 2.7 2.7 2.9 3 3 2.6 2.5 2.6 2.8 2.6 2.6 2.6 2.7 3.1 2.7 2.9 2.7 2.9 3.2 2.8 3 2.8 2.8 2.6 4.4 2.9 2.5 2.7 2.5 3.2 2.9 3.1 3.2 3.1 4.5 2.9

Seismic Hazard Assessment for South-Central Region, Vietnam

189

References 1. Dinh, T.C.: Geophysical Fields and Lithosphere Structure of Vietnam Territory, , vol. 330. Publishing House for Science and Technology, Hanoi (2005) 2. Trieu Cao Dinh, L.P.H., Van, L.D., Van, D.L., Dinh, T.C.: Modern Geodynamics of Vietnam Territory, vol. 242. Publishing House for Science and Technology, Hanoi (2013) 3. Trieu Cao Dinh, L.P.H., Viet, C.D., Thanh, H.D., Van, D.L., Dinh, T.C., Anh, T.T., Nam, H.P., Xuan, B.M.: Tectonic Deformation of the East Vietnam Sea and Adjacent Area in Cenozoic, vol. 308. Publishing House for Science and Technology, Hanoi (2019) 4. Nguyen, H.P., Pham, T.T.: Probabilistic seismic hazard assessment for the South Central Vietnam. Vietnam J. Earth Sci. 36, 451–461 (2014) 5. Nguyen, H.P., Pham, T.T., Nguyen, T.N.: Investigation of long-term and short-term seismicity in Vietnam. J. Seismol. 23, 951–966 (2019) 6. Xuyen, N.D.: Tectonic geology and strong earthquake-generating areas in the Vietnam’s territory, p. 330. Institute of Geophysics, Hanoi (2004) 7. Thanh, T.T.M.: Seismic hazard Assessment in Vietnam and surouding (2002) 8. Cornell, C.A., Engineering seismic risk analysis, pp. 1583–1606 (1968) 9. Campbell, K.W.: Empirical Near-Source Attenuation Relationships for Horizontal and Vertical Components of Peak Ground Acceleration, Peak Ground Velocity, and Pseudo-Absolute Acceleration Response Spectra, pp. 154–179 (1997) 10. Dinh, T.C.: Seismic Hazards in Vietnam. Publishing House for Science and Technology, Hanoi (2010) 11. Nguyen, H.C.: Earthquake microzoning in Ho Chi Minh City. Final project report, Ho Chi Minh City, p. 360 (2009) 12. Bui, X.N., Cao, T.D., Nguyen, L.Q., Mai, B.X., Trinh, L.H., La, H.P., Goyal, R., Thai, T.A., Nam, H.P.: Assessment on maximum magnitude of natural and triggered earthquake when water is impounded in the mining pit: a case study in Nui Nho quarry, Vietnam based on gravity and magnetic data. Russ. J. Earth Sci. 20, 1–10 (2020) 13. Dinh, T.C.: The application of neural network for earthquake prediction in North-West Vietnam. In: 9th General Assembly of Asian Seismological Commission: Extended Abstract of the International Conference (2012) 14. Saaty, T.L.: Decisions Making for Leaders: The Analytic Hierarchy Process for Decisions in a Complex World. RWS Publications, Pittsburgh (1995) 15. Cao Dinh Trieu, N.D.V.: Fault segmentation for estimation of maximum earthquake in Vietnam. J. Geol. (2012). Series A 16. GEM: The OpenQuake-engine User Manual. Global Earthquake Model (GEM) Open-Quake Manual for Engine version 3.8.1, p. 183 (2020) 17. Peresan, A., Panza, G.F., Costa, G.: CN algorithm and long lasting changes in reported magnitudes: the case of Italy. Geophys. J. Int. 141, 425–437 (2000) 18. Blakely, R.J., Simpson, R.W.: Approximating edges of source bodies from magnetic or gravity anomalies. Geophysics 51, 1494–1498 (1986) 19. Boschetti, F.: Improved edge detection and noise removal in gravity maps via the use of gravity gradients. J. Appl. Geophys. 57, 213–225 (2005) 20. Cooper, G.R.J., Cowan, D.R.: Edge enhancement of potential-field data using normalized statistics. Geophysics 73, H1–H4 (2008) 21. Fedi, M., Florio, G.: Detection of potential fields source boundaries by enhanced horizontal derivative method. Geophys. Prospect. 49, 40–58 (2001) 22. Hansen, R.O., Pawlowski, R.S., Wang, X.: Joint use of analytic signal and amplitude of horizontal gradient maxima for three-dimensional gravity data interpretation, pp. 100–102. Society of Exploration Geophysics (1987)

190

T. C. Dinh et al.

23. Abe, S., Yoshimatsu, H.: A review of landslide hazards in Japan and assessment of their susceptibility using an analytical hierarchic process (AHP) method. Landslides 3, 149–158 (2006) 24. Bhatt, B.P., Awasthi, K.D., Heyojoo, B.P., Silwal, T., Kafle, G.: Using geographic information system and analytical hierarchy process in landslide hazard zonation. Appl. Ecol. Environ. Sci. 1(2), 14–22 (2013) 25. Das, B., Chakrabortty, R., Pal, S.C., Malik, S.: Modeling and mapping of groundwater potentiality zones using AHP and GIS technique: a case study of Raniganj Block, Paschim Bardhaman, West Bengal. Model. Earth Syst. Environ. 4, 1085–1110 (2018) 26. Gaikwad, R., Bhagat, V.: Multi-criteria watershed prioritization of Kas Basin in Maharashtra (India): AHP and influence approaches. Hydrospatial Anal. 1, 41–61 (2017) 27. Stathis, D., Stefanidis, S.: Assessment of flood hazard based on natural and anthropogenic factors using analytic hierarchy process (AHP). Nat. Hazards 68, 569–585 (2013) 28. Alonso, J.A.: Consistency in the analytic hierarchy process: a new approach. Int. J. Uncertainty Fuzziness Knowl.-Based Syst. 14(4), 445–459 (2006) 29. Gumbel, E.J.: The return period of flood flows. Ann. Math. Stat. 12, 163–190 (1941) 30. Gumbel, E.J.: Les valeurs extrêmes des distributions statistiques. Annales de l’Institut Henri Poincaré 5, 115–158 (1935) 31. Kijko, A.: Estimation of the maximum earthquake magnitude, mmax . Pure Appl. Geophys. 161, 1655–1681 (2004) 32. Kim, S.G.: On the estimation of parameters in the statistical prediction of earthquakes. J. Phys. Earth 31, 251–264 (1983) 33. Phuong, N.H.: Probabilistic assessment of earthquake hazard in Vietnam based on seismotectonic regionalization. Tectonophysics 198, 81–93 (1991) 34. Adnani, A., Hendriyawan, A., Marto, A.: Seismic hazard assessment for peninsular Malaysia using Gumbel distribution method. Jurnal Teknologi 42, 57–73 (2005) 35. Goda, K.: Nationwide earthquake risk model for wood-frame houses in Canada. Front. Built Environ. 5, 128 (2019) 36. Tsapanos, T.M.: A quantitative appraisal of earthquake hazard parameters computed from Gumbel I method for different regions in and around Turkey. Nat. Hazards 47, 471–495 (2008) 37. Kijko, A., Smit, A., Sellevoll, M.A.: Estimation of earthquake hazard parameters from incomplete data files. Part III. Incorporation of uncertainty of earthquake-occurrence model. Bull. Seismol. Soc. Am. 106(3), 1210–1222 (2016) 38. Kijko, A.: On Bayesian procedure for maximum earthquake magnitude estimation. Res. Geophys. 2, e7 (2012) 39. Kijko, A., Sellevoll, M.A.: Estimation of earthquake hazard parameters from incomplete data files. Part II, incorporation of magnitude heterogeneity. Bull. Seismol. Soc. Am. 82, 120–134 (1992) 40. Salamat, M., Zare, M.: Bayesian and frequentist approaches for the estimation of the maximum expected earthquake magnitude in Iran. J. Seismol. Earthq. Eng. 21, 55–63 (2019) 41. Ezzelarab, M., Shokry, M.M.F., Mohamed, A.M.E., Helalb, A.M.A., Mohamed, A.A., ElHadidy, M.S.: Evaluation of seismic hazard at the northwestern part of Egypt. J. Afr. Earth Sci. 113, 114–125 (2016) 42. Haarala, M., Orosco, L.: Generalized proofs of the Kijko-Sellevoll functions Pruebas generalizadas de las funciones Kijko-Sellevoll (2019) 43. Pagani, M., Monelli, D., Weatherill, G.A., Garcia, J.: The OpenQuake-engine book: hazard. Global Earthquake Model (GEM) Technical report, p. 67 (2014) 44. Field, N., Jordan, T.A., Cornell, C.A.: OpenSHA: a developing community-modeling environment for seismic hazard analysis. Seismol. Res. Lett. 4, 406–419 (2003)

Seismic Hazard Assessment for South-Central Region, Vietnam

191

45. Crowley, H., Pinho, R., Pagani, M., Keller, N.: Assessing global earthquake risks: the Global Earthquake Model (GEM) initiative. In: Handbook of Seismic Risk Analysis and Management of Civil Infrastructure Systems (2013) 46. Abbott, E., Van Houtte, C.: OpenQuake implementation of the canterbury seismic hazard model. Seismol. Res. Lett. 90, 2227–2235 (2019) 47. Kalakonas, P., Silva, V., Mouyiannou, A., Rao, A.: Exploring the impact of epistemic uncertainty on a regional probabilistic seismic risk assessment model. Nat. Hazards 104, 997–1020 (2020) 48. Weginger, S., Jian, Y., Papi-Isaba, M., Lenhardt, W. (2019). 21 49. Kuo, J.T., Yegulalp, T.M.: Statistical prediction of the occurrence of maximum magnitude earthquakes. Bull. Seismol. Soc. Am. 64(2), 393–414 (1974) 50. Coppersmith, K., Bommer, J., Hanson, K., Unruh, J., Coppersmith, R., Wolf, L., Youngs, R., Rodriguez Marek, A., Al Atik, L., Toro, G.: Hanford Sitewide Probabilistic Seismic Hazard Analysis, Report PNNL-23361, Pacific Northwest National Laboratory, Richland, Washington (2014) 51. Pagani, M., Garcia, J., Poggi, V., Weatherill, G.: Probabilistic seismic hazard analysis: issues and challenges from the GEM perspective (2016)

Secondary Processes Associated with Landslides in Vietnam Pham Van Tien1(B) , Le Hong Luong2 , Tran Thanh Nhan3 , Do Minh Duc1,4 , Dinh Thi Quynh1 , Nguyen Chau Lan5 , Nguyen Quoc Phi6 , Do Canh Hao7 , Nguyen Huu Ha4,8 , Dang Thi Thuy1 , and Vu Ba Thao9 1 Institute of Geotechnology and Environment, Hanoi, Vietnam

[email protected] 2 Institute of Transport Science and Technology, Hanoi, Vietnam 3 University of Sciences, Hue University, Hue City, Vietnam 4 VNU University of Science, Vietnam National University, Hanoi, Vietnam 5 University of Transport and Communication, Hanoi, Vietnam 6 University of Mining and Geology, Hanoi, Vietnam 7 Institute of Training and Science Application, Thuyloi University, Ninh Thuan, Vietnam 8 Department of Science and Technology, 208 Dien Hong, Quy Nhon, Binh Dinh, Vietnam 9 Academy for Water Resources, Hydraulic Construction Institute, Hanoi, Vietnam

Abstract. Landslides are one of the most dangerous geohazards in tropical monsoon countries. Various impacts of landslides on lives and property not only result from the destruction of the down movement itself but is also due to secondary effects including the formation of landslide-dammed lakes and the generation of tsunami-like waves. This paper presents a study on secondary processes associated with landslides hazards in Vietnam through site surveys, air photos, and data collection and analysis. First, the paper reports a comprehensive investigation of the study on landslides and their consequential hazards in recent 30 year. Then, three typical cases of landslides in the Van Hoi reservoir, Khanh waterfall, and Song Bung hydropower reservoir are characterized in terms of geological features, causes, and sliding mechanisms. Besides, landslide hazard assessment for disaster risk reduction is briefly discussed. Study results significantly indicate that heavy rainfall is the main trigger for landslides and its cascading effects (i.e., river damming and dam breach, and landslide-generated waves). While the geological structures of high fractured, deformed, and weathered rocks are the main preparatory factor of the landslides. Landslides associated with secondary hazards has been rarely analyzed in Vietnam, this study will, therefore, bring a significant understanding for planning and management of multiple disaster risk in the river-hillslope system. Keywords: Landslides · Secondary processes · Dam reservoir · Cause · Mechanism · Vietnam

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 192–209, 2021. https://doi.org/10.1007/978-3-030-60269-7_10

Secondary Processes Associated with Landslides in Vietnam

193

1 Introduction Landslide phenomena are globally one of the most frequent natural hazards that cause a lot of significant damage to people and properties [1, 2, 3]. The human and economic losses result from the destruction of the mass movement of earth materials itself and the potential effects of its secondary processes that include the dam formation due to river blockage and the generation of tsunami-like waves [4–6, 7, 8]. The cascading effects to upstream and downstream areas due to the secondary hazards are presented in detail by Korup [8]. In river valleys, the large amount of the sliding materials can completely or partially fill the rivers to create natural reservoirs behind the landslide dams [4–8]. If the water table due to the impoundment process increases; the landslide dam may be highly vulnerable to instability or breaching because of various phenomena of upstream inundation, dam erosion phenomena, and the continuous effects overtopping and piping failures [6, 8]. The landslide dam breach associated with debris flows and outburst flood will pose serious hazards to downstream communities. While upstream reservoir bank slopes saturated by impounded water are prone to the failure to generate impulse waves and overtop that may cause cascading effects [4–6]. Several historical records of catastrophic landslides associated with secondary processes, including dam formation, landslide-generated waves and flash floods, are presented in Table 1. Table 1. Historical records of catastrophic landslides associated with its secondary effects over the world No Event

Time

Country Casualties Ref

1

The overtopping and flooding due to the landslide dam failure

1786

China

100,000

[9]

2

The failures of three landslide-dammed lakes and its flooding

1933

China

20,000

[10]

3

Landslide lake outburst flooding and landslide dam 2013 failures in Uttarakhand

India

5,000

[11]

4

The landslide induced waves in the Vajont reservoir 1963

Italy

2,000

[12]

5

The Shiaolin landslide dam and severe outburst flood event

2009

Taiwan

400

[13]

6

The Jure landslide dam and its dam overtopping and failures

2014

Nepal

156

[14]

7

The large-scale landslide in the Canelles reservoir

2006

Spain

-

[15]

8

The Qianjiangping landslide in the Three Gorge Reservoir (TGR)

2003

China

24

[16]

9

The large-scale deep-seated landslide in the Aratozawa reservoir

2008

Japan

-

[17]

10

The Shuping and Outang landslides in the TGR

Active China

-

[18]

194

P. Van Tien et al.

Landslides and its secondary processes have been well studied by many authors. First, hazard assessment of landslide dams by an investigation of various effects resulting from dam formation and failure, (e.g., backwater inundation, dam breaches, outbursts of impoundment waters, downstream flooding, debris flows, and long-term river aggradation) has been conducted by [4–6, 8, 19–27]. In this point of view, heuristic, deterministic, statistical, and inventory approaches are employed to identify the risks and explore their mechanisms [1, 28]. Tien [28] present that numerical models using the heuristic method were often carried out by trial-and-error analyses (e.g., the LSFLOW model by Sako [29] and a distinct element method (a two-dimensional (2D) particle flow code (PFC) by Li [30]). Deterministic approaches are commonly based on slope stability analysis using soil properties [18, 31]. For statistical methods, regression analysis is used to assess the susceptibility of the landslides and its sequential effects [4, 8, 24, 26]. Inventorying landslide dams for building regional and global databases of landslide dams is the most common approach for qualitative identification of landslide dam hazards [32]. There are several large datasets of landslide dams in the world, which consist of 31 cases in Switzerland; 232 landslide dams in New Zealand; 300 landslide dams from the Alps to the Southern Apennine and Sicily areas; and a complete and unique inventory of 828 landslide dams triggered by May 12, 2008 Wenchuan earthquake in China [28, 32]. In these studies, several geomorphological indexes, which integrate two or more morphological parameters of the landslides (e.g. the volume, depth, velocity of mass movement) and the river valley (e.g. catchment area, valley width, slope parameters), were also proposed to evaluate dam formation and stability for its risk assessment and management [28]. Notably, modeling approaches, including deterministic models, an empirical model, numerical simulations, or laboratory and field experiments have been widely employed to evaluate secondary hazards of dam formation, backwater inundation, dam failure, dam-breach floods, etc. [22, 25, 28, 33, 34]. Secondly, slope failures into reservoirs may create impulse waves that pose the most disastrous hazards to dam structures and downstream areas. Landslide-induced waves have been studied by many authors worldwide, e.g., Panizzo [35]; Biscarini [36]; AtaieAshtiani and Yavari-Ramshe [37]; and Glimsdal [38]. Hazard assessment of landslideinduced waves aims to evaluate some parameters such as the velocity, height and run-up of waves, the run-up impact areas, and time of the propagation. These parameters are investigated from mathematical theories [39], physical model experiments [35, 40], and numerical simulations [36]. As for numerical models, the authors carried out their studies based on Boussinesq equations [41] and Reynolds-averaged Navier–Stokes (RANS) equations [42]. Vietnam has a mainland area of 330,000 km2 , of which 70% is mountainous. Vietnam, with a 3,260 km long coastline, located in the eastern margin of the Indochinese peninsula that is strongly affected by typhoons and tropical depressions. In the period 1961–2010, the country was on average hit by about 12 typhoons per year which always brought heavy rainfall and floods [43]. Besides, the complexity of topography, geology and monsoon climate with extreme rainfall makes the country extremely prone to geo-hydrological hazards such as landslides and flash floods. Landslides frequently take place on cut slopes during tropical cyclones in which rainfall plays the role of a triggering factor [27, 44, 45, 46]. In recent decades, landslides and flash floods have

Secondary Processes Associated with Landslides in Vietnam

195

become very intensive and destructive, causing huge losses of life and property [47]. The damages to the transportation system due to landslides and floods were estimated at 100 million USD per year [48]. In mountainous regions, landslides may block the river channels to form natural reservoirs that are vulnerable to failure to induce flash floods while the sudden falling of the landslides into the reservoirs can generate displacement waves inducing downstream flash floods and overtopping. Although these secondary hazards of landslides have been reported many times by the media and authorities, landslides associated with secondary hazards have rarely addressed for proposing disaster risk reduction measures. This paper, therefore, aims to firstly review the episodes of landslides in association with secondary hazards, and then to present preliminary results on a study of three landslide cases in Vietnam through site surveys and the analysis of air photos and data collection.

2 Secondary Processes Associated with Landslides in Vietnam: A Review In recent years, landslides, mudflows and flash floods frequency occur in the mountainous regions, causing serious and extensive damages to lives and properties in Vietnam. From 2000 to 2015, landslides, mudflows and flash floods happen every year with a total of 250 events, killed 779 people, and caused a huge economic loss [49]. Flash floods are mainly induced by heavy rainfall in the large watersheds with steep channels in the Northern and Central regions and frequently accompanied by landslides [47, 50]. The provinces, which are the most frequently affected by landslides, mudflows and flash floods are such as Lai Chau, Lao Cai, Ha Giang, Yen Bai, Son La, Hoa Binh, Thanh Hoa, Nghe An [13, 51]. Many landslides and flash floods, which simultaneously were induced by heavy rainfall, were recorded in Vietnam, e.g. the 1996 flash flood due to landslide dam breach killed over 89 people in Lai Chau [51]; the 2005 flash flood claimed 50 people in Van Chan, Yen Bai; the flash flood caused 88 deaths in Lao Cai in 2008; the Ban Khoang flash flood and extensive landslides killed 11 people in 2013; the Sa Na outburst flood and landslide disaster claimed ten casualties in 2019 (Table 2). As can be seen in Table 2, many flash floods occurred as a cascading effect due to the formation and failure of landslide-dammed lakes. The landslide dam breach causing the 1996 Muong Lay flash flood has been so far one of the worst sediment-related disasters in Vietnam. The landslide formed a natural lake in 1 km length at Nam He river. Due to continuous heavy rainfall, the landslide-dammed lake failed to release a flash flood with a flooding water depth of 15 m, which swept away the Muong Lay town. This disaster claimed 89 people and completely demolished Muong Lay town. The town was forced to relocate to a new site after the disaster. In most of the cases, landslide dams are formed in the regions where the terrain is narrow valleys bounded by high and steep slopes [50–52]. This kind of regional setting is mainly the preparatory condition for blocking river valleys to form landslide-dammed lakes in Vietnam. In this regard, narrow valleys facilitate the river damming even a relatively small volume of the landslides [4, 28]. On 26 September 2012, a landslide with a total volume of 10,000 m3 completely blocked the Coc stream to form a natural reservoir in Ta Phoi commune, Lao Cai city (Fig. 1). The landslide-dammed lake had a maximum depth of 20 m, 25 to 30 m wide

196

P. Van Tien et al.

Table 2. Recorded history of some flash flood and landslide events triggered by heavy rainfalls No

Time

Location

Flash floods due to dam breach

Death

Ref

1

27 Jul. 1991

Son La city

x

32

[53]

2

27 Jul. 1991

Mai Son, Son La

x

16

[53]

3

Aug. 1996

Muong Lay, Lai Chau

x

89

[51]

4

Sep. 2005

Van Chan, Yen Bai

-

50

[54]

5

5 Jul. 2007

Moc Chau, Son La

-

10

[53]

6

8 Aug. 2008

Lao Cai city

-

88

[47]

7

26 Sep. 2009

Mai Chau, Son La

-

35

[53]

8

26 Apr. 2010

Ha Giang city

-

5

[47]

9

14 Aug. 2010

Van Yen, Yen Bai

-

7

[47]

10

31 Aug. 2012

Bac Ha, Lao Cai

x

11

[47]

11

4 Sep. 2013

Ban Khoang, Sapa, Lao Cai

x

11

[55]

12

5 Aug. 2016

Bat Xat, Lao Cai

-

11

[49]

13

3 Aug. 2017

Nam Pam, Son La

-

15

[49]

14

3 Aug. 2017

Mu Cang Chai, Yen Bai

x

14

[49]

15

3 Aug. 2019

Sa Na, Thanh Hoa

x

10

[49]

Note: Mark “x” presents the landslide dam formation and breach to generate flash floods

with an impoundment volume of 1.0 million m3 . The upstream inundation area was estimated at 0.2 km2 [56]. The dam formation and its potential outbreak threatened to 400 households and a large area of agricultural land downstream. The dam failed on the night of 26 September when urgent countermeasures to drain out impounded water were conducted by the Lao Cai government. Therefore, there was no sudden failure of the landslide dam and outburst flood that could cause a severe disaster to downstream communities. In July 2013, a landslide dam reservoir was created due to a blockage of the Muong Hum stream in Bat Xat, Lao Cai city [52]. The dam breached in a short time, it might be because a small volume of the landslide material was eroded by a large volume of impounded water. Notably, about 1:30 AM on 12 October 2017, a landslide associated with mudflows severely destroyed Khanh village in Phu Cuong, Tan Lac, Hoa Binh province. The catastrophic disaster killed 18 deaths, buried 10 houses and partly damaged to other tens of houses and local roads (Fig. 22) [57]). The landslide was preceded by local failures and ground settlements in the village just one day before, but no evacuation and any measures were forced to prevent and mitigate its potential impacts [58]. In Vietnam, landslides not only dam the river channels to form natural lakes, but also the falling of the landslide body into the water can generate displacement waves. Two events of landslide-generated waves were recorded in the recent years, including impulse waves generated by a large deep-seated landslide in the Van Hoi reservoir dam

Secondary Processes Associated with Landslides in Vietnam

(a)

(b)

(c)

(d)

197

Fig. 1. Landslide dam formation in Ta Phoi, Lao Cai (photos from Radio The Voice of Vietnam - VOV): (a) Landslide body blocked the Coc stream; (b) Upstream inundation due to landslide dam formation; (c) Countermeasures to drain out impounded water; and (d) Landslide dam after the breach at Muong Hum, Bat Xat, Lao Cai [52].

in 2016 and landslide-induced water waves in Truong river. In the latest event, the landslide was induced by extremely intense rainfall with cumulative 48-h precipitation of 947.2 mm (Fig. 3). The landslide produced a soil volume of 150,000 m3 and its downslope movement instantly generated a surge wave of 8.5 m high across the Truong river destroyed 6 houses, claimed 1 person, and injured 3 others in the opposite slope in Tra Giang commune, Bac Tra My district, Quang Nam province (Fig. 4, [59]). Duc et al. [59] studied the landslide and its impulse wave by using numerical models of LS-RAPID and LS-TSUNAMI developed by Sassa et al. [60, 61] for landslide simulation and landslidetsunami simulation, respectively (as shown in Fig. 5). The numerical simulations of Tra Giang landslide and landslide-generated tsunami, which were first applied in Vietnam, were well verified to interpret the entire processes of the landslide and its consequential event.

3 Case studies 3.1 Landslide in Van Hoi reservoir dam Located in An Tin commune, Hoai An district, Binh Dinh province, Van Hoi reservoir with a storage capacity of 14.5 million m3 is one of the very big irrigation reservoirs and plays a vital role in agricultural development in Vietnam. Recently, landslides frequently take place in the rainy season, which seriously affect the safe operation of dam facilities

198

P. Van Tien et al.

(a)

(b) Waterfall

Paddy field

Resident area buried by debris materials

River flow (c)

Landslide-dammed lake

Landslide debris Landslide debris Fig. 2 (a) Photo of Khanh waterfall before the failure by a local resident [37], and (b, c) landslide and its impacts to resident areas in Khanh village (photos from a video by Tan Lac Radio Television [62]).

and threaten the lives of thousands of households downstream. On 17 December 2016, heavy and prolonged rainfall triggered several landslides on the upstream slope opposite the dam facility. The biggest landslide entered the lake and generated high-water waves of about 20 m which extensively destroyed the dam structure and its facilities before causing spills over the dam crest and operation house (Fig. 6) [64]. This serious disaster has never happened in Vietnam. Precipitation data before and after the landslide occurrence monitored at Hoai An rain gauge station in Tang Bat Ho district, approximately 6 km from the reservoir site, are shown in Fig. 7. It shows that the landslide was triggered by heavy rainfall. The accumulative rainfall prior to the sliding was over 800 mm in December 2016. Field surveys and UAV investigations were carried to examine the causes and characteristics of the landslide. An orthomosaic photo of the Van Hoi reservoir that was generated by a series of aerial photos is shown in Fig. 8a. The landslide occurred on the slope that densely covered by a forest. The slope failed at elevation 196 m and deposited

Secondary Processes Associated with Landslides in Vietnam

199

Landslide triggered by rainfall at 4 pm on 5 November 2017

Fig. 3. Landslide occurrence time and hourly precipitation data at Tra My meteorological station.

(a)

(b)

(c)

Fig. 4. (a) 3D view of the Tra Giang landslide [59]; (b) Landslide after sliding (photo by Dinh Thi Quynh); and (c) Collapsed houses due to tsunami waves [63].

200

P. Van Tien et al.

(a)

(b)

Fig. 5 (a, b) Simulations of rainfall-induced landslide and landslide-generated tsunami [59].

Fig. 6. Damages to the Van Hoi dam and its facilities due to landslide-generated waves (Photo courtesy by Do Canh Hao).

at the reservoir floor at an elevation ranging from 25–28 m. The displaced mass has a length of about 510 m and a width ranging from 120 m at the top and 280 m at the toe. The landslide occurred on the slope of 21°. The displaced materials of the slope entirely lied on the floor of the lake and severely reduced the operating capacity of the dam (Fig. 8b). Preliminary results of site surveys show that the landslide slid along the bedrock of granite rocks and the landslide deposits mainly consisted of weathered granite materials of sand, silty sand, clayed sand, and clay (Fig. 8c). 3.2 Landslide in Khanh waterfall In Phu Cuong, Tan Lac, Hoa Binh province, a large portion of the Khanh waterfall slid down on 12 October 2017 and buried a large area in Khanh village (Figs. 9 and 10). The landslide is geologically characterized by Dong Giao and Co Noi Formations with limestones, sandstone, tuffaceous siltstone, clay shale, and marl. The area is located in

Secondary Processes Associated with Landslides in Vietnam

201

Landslide occurred on 16 Dec. 2016

Fig. 7. Daily rainfall in December 2016 and landslide occurrence.

(a)

Landslide source area

Debris

Van Hoi dam

Reservoir

Resident areas (b)

Weathered rocks

Sliding surface Landslide deposits in Van Hoi reservoir Fig. 8. (a) Orthomosaic photo of the Van Hoi reservoir and its landslide; (b) Van Hoi reservoir and landslide in June 2018 and October 2019; (c) Sliding surface exposed in the flank of the landslide.

202

P. Van Tien et al.

the North-West of Vietnam, which is strongly affected by active faults in the east-west direction [65]. Some lineaments as active faults can be delineated on satellite images around the Khanh village slope in Fig. 9b. Based on the site survey, a geological slope profile of the landslide was created in Fig. 9c. It presents that the landslide area consisted of weathered materials of limestones in Devonia-Slurian Age and sedimentary rocks. The slope slid along the sliding surface which was the geological boundary of sedimentary rocks and limestones. The slope with a height of 120 m and width 200 m was triggered by heavy rainfall. The cumulative hourly precipitation that triggered the landslide was 437 mm between 9 to 11 October 2017 (Fig. 11).

(c) Sliding surface

Landslide site

(a)

Limestones

Sedimentary rocks

Fault line

(b) Lineaments (faults)

National Highway No.6

Reservoir Landslide

Affected resident area Khanh village

Fig. 9. (a) Landslide location and its geological settings [65], (b) 3-D view of Khanh waterfall area, and (c) geological slope profile of the landslide

We conducted a site survey in January 2018 to study geological and morphological features and sliding mechanism. Topographical and geological features of the landslide area is presented in Fig. 11. The landslide was characterized by a compound type initiated by rockfalls and then the slope failed as earth fall and earth slide. A rapid massive movement as a debris flow was then formed along the river valley with a long distance

Secondary Processes Associated with Landslides in Vietnam

203

of 350 m. The landslide debris moved down and blocked the stream to form a small natural lake with an impounded water volume of about several thousand cubic meters (Fig. 11a and b). Site investigations indicated that the sliding surface mainly lied on an active fault in a northwest-southeast direction. The landslide debris consisted of materials of strongly fractured, deformed and weathered limestones (Fig. 11c, d, and e). These features of limestones are due to the strong influence of active tectonic faults in the study area. The geological features of fractured and sheared rocks favored the water infiltration deeply into the slope strata to generate the high pore water pressure that triggered the sliding. As can be seen in Fig. 11, the water always flowing out from the top mountain indicates the abundant condition of water runoff and groundwater table in the study site. This factor is also a favorable condition for accelerating both physical and mechanical weathering processes of sedimentary and limestone rocks in the landslide area.

Landslide occurred at 1:30 AM on 12 October 2017

Fig. 10. Landslide occurrence time and hourly rainfall data at Mai Chau meteorology station.

3.3 Landslide nearby Song Bung hydropower reservoir No. 5 A large-scale deep-seated landslide near Song Bung No. 5 Hydropower is one of the most dangerous hazards in Quang Nam province. The landslide is located near the national highway No. 14 and on the left bank of the Song Bung reservoir (Fig. 12a). Landslide characteristics were presented on the topographic map using the method of aerial photo interpretation (Fig. 12a, b, [66]). The boundary and main scarp of the landslide are apparent from aerial photos. The landslide is about 1,000 m in width and 1,500 m in length, extending from the top of the mountain to the shoreline of the reservoir (Fig. 12b, c). The landslide is reactivated to move down, therefore, many cracks have been developing along the road and retaining wall (Fig. 12d). In the case of sliding, the landslide can block the stream to form a dam on the upstream of the reservoir or mass movement falling the water body can induce impulse waves and overtop on the dam.

204

(a)

P. Van Tien et al.

(b)

(c)

(d)

(e)

Fig. 11. (a) Top-view of the landslide at Khanh waterfall; (b) Closed-view of the sliding surface; (c) Head scarp and main body of the landslide, and (d, e) Landslide strata with deformed, fractured, and weathered limestones.

Secondary Processes Associated with Landslides in Vietnam

205

(a) Song Bung Hydropower Dam No. 5 Landslide boundary

(c)

(b) Reservoir

Highway

Reservoir

Cracks on highway

(d)

Deformed retaining wall Landslide body Fig. 12. (a) Overview of the landslide by UAV photos; (b, c) Aerial photo and topographic map with the landslide scarp and boundary [66]; (d) Cracks on retaining wall and highway due to sliding.

4 Discussion and Conclusion Landslides associated with secondary processes have severely caused damages to socioeconomic infrastructure and human living throughout the world. Landslide dam formation and landslide-induced waves have rarely addressed so far in Vietnam, therefore, this paper presenting a literature review on the study of landslides and its secondary hazards is helpful not only for understanding the landslides and its secondary processes but also for disaster risk reduction and disaster preparedness. It shows that landslides and flash floods frequently co-occur during rainfall and these two geo-hydrological hazards cause a lot of human and economic losses. Flash floods were recognized as consequential effects of the failure of landslide dams. Landslides associated with river damming have studied since the 1990s and the 1996 Muong Lay flash flood and landslide disaster presented by Minh et al. [51] has documented to be one of the most disastrous cascading geomorphic events in Vietnam. Based on an investigation of various cases, it indicates that the phenomena of landslides followed by river blockages often take place in the narrow channels and steep slopes in the North and Central region of Vietnam. These features are in agreement with previous studies. In the study areas, geological features of fractured, deformed, and weathered rocks are one the most important preparatory conditions for the landslide occurrence. While heavy rainfall is the main trigger of the landslides in Vietnam, which is commonly characterized by high intensity and a short-time period.

206

P. Van Tien et al.

Specifically, rainfall triggered the Truong Giang landslide and Khanh waterfall landslide are very extreme. Tsunami-like waves generated in the Truong river and Van Hoi reservoir are specific and uncommon in Vietnam. In this study, three cases of landslides in association with secondary hazards, e.g., of dam formation (at Khanh waterfall and in Song Bung No. 5 reservoir) and landslide-generated waves (in Van Hoi reservoir), are briefly presented through site investigations, aerial photos, and data analysis. The problems of landslides and their hazards in dam reservoirs have been outlined as an increasingly considerable challenge during the building and operation periods. However, this kind of research topic has still been under development, particularly the investigation of its sliding mechanisms has not been conducted. Therefore, it is imperative to study the initiation mechanism and processes as well as to assess the landslide hazards in Vietnam. The understanding of the mechanisms and processes of landslides and its secondary hazards are very crucial for safely planning and managing the dams and their reservoirs. Acknowledgement. This research is funded by Vietnam National Foundation for Science and Technology Development (NAFOSTED) under grant number 105.08–2019.14.

References 1. Dai, F.C., Lee, C.F., Ngai, Y.Y.: Landslide risk assessment and management: an overview. Eng. Geol. 64, 65–87 (2002) 2. Nadim, F., Kjekstad, O., Peduzzi, P., Herold, C., Jaedicke, C.: Global landslide and avalanche hotspots. Landslides 3, 159–173 (2006) 3. Petley, D.: Global patterns of loss of life from landslides. Geology 40(10), 927–930 (2012) 4. Costa, J.E., Schuster, R.L.: The Formation and Failure of Natural Dams, pp. 87–392. US Geological Survey, Open-File Report (1987) 5. Dal Sasso, S.F., Sole, A., Pascale, S., Sdao, F., Bateman Pinzón, A., Medina, V.: Assessment methodology for the prediction of landslide dam hazard. Natural Hazards Earth Syst. Sci. 14, 557–567 (2014) 6. Evans, S.G.: The formation and failure of landslide dams: an approach to risk assessment. Ital. J. Eng. Geol. Environ. 1 (2006) 7. Korup, O.: Recent research on landslide dams – a literature review with special attention to New Zealand. Progress Phys. Geogr. 26(2), 206–235 (2002) 8. Korup, O.: Geomorphic hazard assessment of landslide dams in South Westland, New Zealand: fundamental problems and approaches. Geomorphology 66, 67–188 (2005) 9. Dai, F.C., Lee, C.F., Deng, J.H., Tham, L.G.: The 1786 earthquake-triggered landslide dam and subsequent dam-break flood on the Dadu River, southwestern China. Geomorphology 65(3–4), 205–221 (2005) 10. Xu, F.G., Yang, X.G., Zhou, J.W.: Experimental study of the impact factors of natural dam failure introduced by a landslide surge. Environ Earth Sci. 74, 4075–4087 (2013) 11. Ray, P.K.C., Chattoraj, S.L., Bisht, M.P.S., Kannaujiya, S., Pandey, K., Goswami, A.: (2016) Kedarnath disaster 2013: causes and consequences using remote sensing inputs. Nat. Hazards 81, 227–243 (2016) 12. Genevois, R., Ghirotti, M.: The 1963 Vajont landslide. Giornale di Geologia Applicata 1, 41–52 (2005)

Secondary Processes Associated with Landslides in Vietnam

207

13. Tsou, C.Y., Feng, Z.Y., Chigira, M.: Catastrophic landslide induced by Typhoon Morakot, Shiaolin. Taiwan. Geomorphology 127, 166–178 (2011) 14. MOI.: Report on Jure landslide, Mankha VDC, Sindhupalchowk district. Nepal Government Ministry of Irrigation (MOI), 2071/06/05 BS, 29 p. (2014) 15. Pinyol, N.M., Alonso, E.E., Corominas, J., Moya, J.: Canelles landslide: modelling rapid drawdown and fast potential sliding. Landslides 9(1), 33–51 (2012) 16. Wang, F., Zhang, Y., Huo, Z., Peng, X.M.: Mechanism for the rapid motion of the Qianjiangping landslide during reactivation by the first impoundment of the Three Gorges Dam reservoir. China. Landslides 5(4), 379–386 (2008) 17. Setiawan, H.: Landslide Hazard Assessment on the Upstream of Dam Reservoir. PhD thesis, Kyoto University (2017) 18. Yin, Y., Huang, B., Wang, W., Wei, Y., Ma, X., Zhao, C.: Reservoir-induced landslides and risk control in Three Gorges Project on Yangtze River, China. J. Rock Mech. Geotech. Eng. 8(2016), 577–595 (2016) 19. Ermini, L., Casagli, N.: Prediction of the behaviour of landslide dams using a geomorphological dimensionless index. Earth Surf. Proc. Land. 28, 31–47 (2003) 20. Fan, X., van Westen, C.J., Xu, Q., Gorum, T., Dai, F.: Analysis of landslide dams induced by the 2008 Wenchuan earthquake. J. Asian Earth Sci. 57, 25–37 (2012) 21. Fan, X., van Westen, C.J., Korup, O., Gorum, T., Xu, Q., Dai, F., Huang, R., Wang, G.: Transient water and sediment storage of the decaying landslide dams induced by the 2008 Wenchuan earthquake. China. Geomorphology 171, 58–68 (2012) 22. Fan, X., Tang, C.X., Van Westen, C.J., Alkema, D.: Simulating dam-breach flood scenarios of the Tangjiashan landslide dam induced by the Wenchuan Earthquake. Nat Hazards Earth Syst Sci 12, 3031–3044 (2012) 23. Hermanns, R.L.: Landslide dam. In: Bobrowsky, P.T. (ed.) Encyclopedia of earth sciences series, pp. 602–605. Springer, Dordrecht (2013) 24. Korup, O.: Geomorphometric characteristics of New Zealand landslide dams. Eng. Geol. 73(1), 13–35 (2004) 25. Peng, M., Zhang, L.M.: Breaching parameters of landslide dams. Landslides 9(1), 13–31 (2012) 26. Stefanelli, C.T., Segoni, S., Casagli, N., Catani F.: Assessing landslide dams evolution: a methodology review. Advancing Culture of Living with Landslides, pp. 253–258 (2017) 27. Stefanelli, C.T., Catani, F., Casagli, N.: Geomorphological investigations on landslide dams. Geoenviron. Disasters 2, 21 (2015) 28. Tien, P.V.: Mechanisms and Hazard Assessment of Rainfall-Induced Landslide Dams. PhD thesis, Kyoto University (2018) 29. Sako, Y., Mori, T., Nakamura, H., Kamee, K., Hanaoka, M., Yusa, N., Fukuda, M., Ohno, R.: Estimating the shape of a landslide dam (river blockage), attributed to a deep catastrophic landslide, using the LSFLOW model. In: Proceedings of the Interpraevent International Symposium 2014 in Nara Prefecture, Japan (2014) 30. Li, X., He, S., Luo, Y., Wu, Y.: Simulation of the sliding process of Donghekou landslide triggered by the Wenchuan earthquake using a distinct element method”. Environ. Earth Sci. 65(4), 1049–1054 (2012) 31. Yang, Q., Ye, Z., Ding, W., Gao, Y.: Impact of water level fluctuation on the reservoir landslide stability. In: Sassa, K., et al. (eds.) Landslide Science for a Safer Geoenvironment, vol. 3 (2014) 32. Tien, P.V., Luong, L.H.: Landslide Dam’s Hazard assessment: a literature review. In: VACI2019 Conference Proceedings. pp. 178–188 (2019). ISBN: 978-604-67-1216-9 33. Kofler, C., Comiti, F., Gems, B., Thiebes, B., Schneiderbauer, S., Schlögel, R.: Assessment of Rockslide Dam Scenarios at Catchment Scale in the Context of Cascading Hazards. Advancing Culture of Living with Landslides, pp. 685–691 (2017)

208

P. Van Tien et al.

34. Tien, P.V., Sassa, K., Takara, K., Fukuoka, H., Khang, D., Shibasaki, T., Hendy, S., Ha, N.D., Loi, D.H.: Formation process of two massive dams following rainfall-induced deep-seated rapid landslide failures in the Kii Peninsula of Japan. Landslides (2018) 35. Panizzo, A., Girolamo, P., De Risio, M., Di Maistri, A., Petaccia, A.: Great landslide events in Italian artificial reservoirs. Nat. Hazards Earth Syst. Sci. 5, 733–740 (2005) 36. Biscarini, C.: Computational fluid dynamics modelling of landslide generated water waves. Landslides 7(2), 117–124 (2010) 37. Ataie-Ashtiani, B., Yavari-Ramshe, S.: Numerical simulation of wave generated by landslide incidents in dam reservoirs. Landslides 8(4), 417–432 (2011) 38. Glimsdal, S., L’Heureux, J.S., Harbitz, C.B.: The 29th January 2014 submarine landslide at Statland, Norway—landslide dynamics, tsunami generation, and run-up. Landslides 13(6), 1435–1444 (2016) 39. Kranzer, H.C., Keller, J.B.: Water waves produced by explosions. J. Appl. Phys. 30, 398–407 (1960) 40. Walder, J.S., Watts, P., Sorensen, O.E., Janssen, K.: Water waves generated by subaerial mass flows. J. Geophys. Res. 108(5), 2236–2255 (2003) 41. Hanes, D.M., Inman, D.L.: Experimental evaluation of a dynamic yield criterion for granular fluid flows. J Geophys Res 90(B5), 3670–3674 (1985) 42. Heinrich, P.: Nonlinear water waves generated by submarine and aerial landslides. ASCE J. Waterways Port Coastal Oc. Eng. 118, 249–266 (1992) 43. Srex.: The Vietnam special report on managing the risks of extreme events on disasters to advance climate change adaptation. IMHEN and UNDP (2015) 44. Dieu, T.B., Binh, T.P., Phi, Q.N., Nhat, H.D.: Spatial prediction of rainfall-induced shallow landslides using hybrid integration approach of Least-Squares Support Vector Machines and differential evolution optimization: a case study in Central Vietnam. International Journal of Digital Earth 9, 1077–1097 (2016) 45. Duc, D.M.: Rainfall-triggered large landslides on 15 December 2005 in Van Canh district, Binh Dinh province. Vietnam. Landslides 10, 219–230 (2013) 46. Lan, C.N., Tien, P.V., Do, T.N.: Deep-seated rainfall-induced landslides on a new expressway: a case study in Vietnam. Landslides 17(2), 395–407 (2019) 47. Thu.c, T., Ha, L.T.: Flashfloods-Background and methodologies. Publisher of Natural Science and Technology in Hanoi (2012) 48. Tam, D.M.: Flooding and landslides at the highways of Vietnam. In: Proceedings of the International Workshop on “Saving Our Water and Protecting Our Land”, Hanoi, 20–22 October, 2001, pp. 18–27 (2001) 49. Vietnam Disaster Management Authority (VDMA) Flash floods and landslides in Vietnam. A presentation report, Scientific Meeting on October 2019, Hanoi, Vietnam (2019) 50. Tu, T.V., Duc, D.M., Tung, N.M., Cong, V.D.: Preliminary assessments of debris flow hazard in relation to geological environment changes in mountainous regions, North Vietnam. Vietnam J. Earth Sci. 38(3), 277–286 (2016) 51. Minh, V.C., Chuong, P.D., Minh, T., Thang, T., Tu, D.V., Tu, T.V., Dan, N.L., Can, N., Chat, V.V., Hai, T.Q., Kha, T.V., Hoan, N.T., Linh, P.D., Hai, N.P., Cuc, L.T., Nhan, P.T.: Report on the assessment of landslides and debris flows and the proposals of countermeasures in Lai Chau. Institute of Geological Sciences - Vietnam Academy of Science and Technology (1997) 52. SFLP: Report on landslides in Laocai, the State-Funded Landslide Project (SFLP) for Investigation, assessment and warning zonation for landslides in the mountainous regions of Vietnam (2014) 53. Nghi, H.Q., Khuong, D.V., Linh, N.M.: An application of GIS for landslides and flash floods hazard forecast maps in Son La. Journal of Water Resources Science and Technology, Vietnam Academy for Water Resources (2012)

Secondary Processes Associated with Landslides in Vietnam

209

54. Nga, P.T.T., Duy, N., Archana, R.S., Matthias, G.: Vulnerability assessment of households to flash floods and landslides in the poor upland regions of Vietnam. Clim. Risk Manage. 28, 100215 (2020) 55. Nhandan Online News. https://www.nhandan.com.vn 56. Radio The Voice of Vietnam (VOV) (2012). https://vov.vn/xa-hoi/hinh-anh-sat-lo-nui-chandong-suoi-coc-tai-lao-cai-226451.vov 57. Hoa Binh Online News. https://www.baohoabinh.com.vn/274/119827/Tro-lai-noi-xay-ra-lodat-kinh-hoang-xom-Khanh.htm 58. Lao Dong Online News. https://laodong.vn/xa-hoi/can-canh-thac-khanh-truoc-khi-sat-lochon-lap-18-nguoi-o-hoa-binh-570142.ldo 59. Duc, D.M., Khang, D., Duc, D.M., Ngoc, D.M., Quynh, D.Q., Thuy, D.T., Giang, N.K.H., Tien, P.V., Ha, N.H.: Analysis and modeling of a landslide-induced tsunami-like wave across the Truong river in Quang Nam province, Vietnam. Landslides (2020) 60. Sassa, K., Nagai, O., Solidum, R., Yamazaki, Y., Ohta, H.: An integrated model simulating the initiation and motion of earthquake and rain induced rapid landslides and its application to the 2006 Leyte landslide. Landslide 7(3), 219–236 (2010) 61. Sassa, K., Dang, K., Yanagisawa, H., He, B.: A new landslide-induced tsunami simulation model and its application to the 1792 Unzen-Mayuyama landslide-and-tsunami disaster. Landslides 13(6), 1405–1419 (2016) 62. Tan Lac Radio and Television: A published video on Khanh village before and after the landslide disaster (2017) 63. Lao Dong Online News. https://laodong.vn/xa-hoi/tai-nan-hy-huu-sat-nui-ben-kia-song-xoaso-lang-ben-nay-574774.ldo 64. Binh Dinh Irrigation Works Operation Limited Company. A report on Landslides in Van Hoi reservoir released on 19 September 2016 (No. 150/BC-KTCTTL). https://khaithacthuyloibin hdinh.com.vn 65. Department of Geology and Mineral Resources (DGM) Geological and mineral resources map of Vietnam on 1: 200,000 (1999) 66. Luong, L.H.: Large scale landslide risk evaluation by aerial photograph interpretation and integrated ahp approach for humid tropical region based on Japan and Viet nam field surveys. Ph.D. thesis. Tohoku Gakuin University (2016)

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional: A Case Study in Sapa, Vietnam The Viet Tran1(B)

, Viet Hung Hoang1

, Huy Dung Pham1

, and Go Sato2

1 Department of Civil Engineering, Thuyloi University, 175 Tay Son, Dong Da, Hanoi, Vietnam

[email protected] 2 Graduate School of Environmental Informations, Teikyo Heisei University, Tokyo, Japan

Abstract. Slope failures in nature are in the three-dimensional form; therefore, it is generally thought that the traditional one-dimensional or two-dimensional slope stability analyses cannot well consider the characteristics of the actual landscape in many situations and commonly produce more conservative results. In this study, the deterministic Scoops3D - a fully three-dimensional, physical-based landslide model was deployed. The program employs the three-dimensional column limit – equilibrium techniques and the digital elevation model (DEM) to perform a comprehensive three-dimensional slope stability analysis. Scoops3D evaluates the stability of a rotational, spherical slip surface encompassing many DEM cells, producing the least-stable potential landslide for each cell throughout the entire digital landscape alongside the related volumes and areas. For the evaluation of the performance of Scoops3D, a severe landslide event that took place on August 05, 2019, following a historical rainstorm event in Sapa, Lao Cai, Vietnam was taken into account. The Success Rate (SR) and the Modified Success Rate (MSR) were employed to compare the actual landslide scar with that predicted by Scoops3D. The results show that with reliable input data, the approach is capable of predicting the locations of future landslides with moderate accuracy, and the updated topographical conditions simulated by Scoops3D can be used for further studies on the occurrence of future landslides in the study area. Keywords: Three-dimensional · Limit-equilibrium · Slope failure · Scoops3D · GIS · Sapa · Vietnam

1 Introduction Slope stability assessment is a crucial branch in the field of geotechnical engineering [1]. At present, most approaches and tools for slope stability analyses are based on the two-dimensional (2D) theories with many simplified assumptions. This simplification might reduce the accuracy and reliability of their performance as the planar analyses do not well reflect the actual situations. It has been claimed by many researchers that the 2D approach has been accepted in slope stability analyses not only because they are © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 210–229, 2021. https://doi.org/10.1007/978-3-030-60269-7_11

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

211

more straightforward than the 3D, but also because they give more conservative results than the 3D does [2, 3, 4, 5]. Besides, the plane-strain models are only suitable for the cases where the slip surfaces are wide enough in comparison with the cross-sectional dimension [5]. Thus, 1D or 2D simplification is sometimes mostly an intuitive approach, which might lead to non-conservative values of the back-calculated soil shear strength [6]. The literature review reveals that the assessment of slope stability over a regional scale is becoming more and more critical in the context of disaster prevention and mitigation [4]. Many two- and three-dimensional slope stability methods have been developed into or integrated with Geographic Information System (GIS). GIS is a powerful tool for spatially distributed data processing and the technology that has recently shown substantial improvement. This tool has been widely used in landslide assessment [7]. In literature, the stability of a natural slope depends very much on conditions that are related to the topography, strata, as well as pore-water pressure (PWP). Those quantities are all spatially distributed, and therefore building a landslide model under a GIS environment is, as a result, possible [8]. However, the deterministic models of slope stability always require complicated algorithms and iteration processes. Thus, only simple models that allow simulating the factor of safety (Fs ) for the individual pixel of GIS raster data that have been implemented inside the GIS system [9] and GIS-based application of the 3D slope stability model in practical situations has rarely been developed. Regarding the limit equilibrium method (LEM), most of the 3D slope stability LEMs are derived from 2D LEMs with similar simplified assumptions [2]. In general, three popular approaches are commonly used for column-based 3D slope stability analyses in literature: 1) the ordinary column-based model developed by Hovland (1977) [10]; 2) The 3D extension of the Bishop’s simplified method introduced by Hungr (1987) [11]; and 3) The 3D extension of Janbu’s simplified method suggested by Hungr et al. (1989) [12]. Amongst these three approaches, the extension of Bishop’s method typically provides reliable factor of safety (Fs ) results that are very close to more recent LEMs [4, 11, 13, 14, 15]. In the effort to combine GIS and 3D column-based slope stability analysis, several models have been introduced such as 3DSLOPEGIS [16, 17], r.slope.stability [18, 19] in GRASS GIS [20]. However, these models are probably developed for the individual pre-existing sliding area but not for a regional landscape. Also, the knowledge of the slip surface is essential for the initial model development. In the 3DSlopeGIS, for example, the Monte Carlo-based random searching method is utilized to find the potential slip surface. The search is performed by minimizing the 3D factor of safety with the exploitation of the Monte Carlo random simulation method to identify the 3D critical failure plane. In the beginning phase, the initial slip surface is assumed to be the lower part of an ellipsoid slip. Then each randomly produced slip surface is changed according to the strength of different existing strata. Finally, the relative minimization of the 3D factor of safety is achieved. With this complexity, there is an increased need for more reasonably GIS grid-based 3D deterministic models to identify the critical slip surface and to model the actual mechanism of landslides in reality. In this study, Scoops3D [14], a three-dimensional limit equilibrium model developed by the U.S. Geological Survey was used for the prediction of potential landslides.

212

T. V. Tran et al.

Scoops3D is one of many programs benefiting from the development of GIS. The model analyzes spherical failure surfaces by expanding Bishop’s Simplified Method or the Ordinary (Fellenius) method in the 3D space and employing the factor of safety to assess the landslide potential for an area represented by a DEM. The sliding event that took place in Sapa, Lao Cai Province, Vietnam on August 05, 2019, was considered to validate the performance of Scoops3D. Frequent landslides that occur along the hillside roads in this area have caused considerable concerns and more studies should be conducted to find out mitigation measures for this geohazard [21]. The quality of the factor of safety map simulated by Scoops3D will be evaluated by comparing the actual sliding scar with the predicted landslide using the Modified Success Rate (MSR) proposed by Huang and Kao (2006) [22]. The paper is organized as follows: Sect. 1 gives a brief overview of the advances of the application of 3D slope stability analysis compared to the traditional 1D or 2D approach and the use of Scoops3D in the slope stability assessment on a regional scale. The integration of GIS and 3D column-based slope stability analysis is also discussed. An introduction to the study area is summarized in Sect. 2. Section 3 and Sect. 4 briefly introduces Scoops3D and the major corresponding data used in slope stability assessment. Section 5 looks at the MSR – a tool to evaluate the performance of Scoops3D in landslide prediction in the selected area. The next section, the Result, and Discussion examine the performance of Scoops3D in the prediction of landslide location for the study area. Finally, conclusions are drawn in Sect. 7.

2 The Study Area In Lao Cai, landslides are a recurrent phenomenon due to its particular geological and geomorphological conditions [23]. Nguyen and Dao (2007) in their study [24] have pointed out eight major reasons for the occurrence of a landslide in the North-Western part of Vietnam including 1) Relief slope: landslides often occur at the slope of > 25o and most at the interval of 30o to 45o ; 2) weathering process of rocks: many landslides with sliding surface in the boundary between the original rocks and uncompleted weathering zones; 3) Modern tectonic movement (caused by earthquakes and active faults): many locations of landslide are closely connected to active tectonic faults; 4) Hydro-system (surface stream, especially groundwater): All major locations of landsliding locations are concerned with groundwater. Landslides often occur in the regions with a high level of rain and the probability of occurrence increases during rainy seasons; 5) Vegetation density: landslides do not often occur in areas with a high percentage of vegetation cover, in the areas with low vegetation coverage, landslides occur more often in greater magnitude; 6) Striking and dipping of original rocks: conditions that the relief slope direction coincides with the dipping of rocks or with the rock foliation is the favorable condition for landslide occurrences; 7) Physical property and structure of original rocks: it is well acknowledged that the slip surface will always go through the weaken and strongly broken up zones; 8) Human activity: human activities such as deforestation, farming, the creation of artificial lakes, road construction, mining, and others may directly or indirectly cause a landslide. Rainfall-induced landslides have been a significant problem in mountainous areas in the North of Vietnam in general and in Lao Cai province in particular [25, 26, 27]. A

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

213

large number of geological hazards have been reported during rainy seasons, especially, landslide-related hazard in Lao Cai province, Vietnam. This hazard has received considerable attention in the last few decades due to significant losses of human lives, infrastructural damage, as well as environmental devastation [25]. The study area is located along the provincial road No. 152, at Km9 + 100 which locates in Lao Chai, Hau Thao, Ta Van, and Su Pan communes of Sapa District (Fig. 1a) in Lao Cai province, between longitudes 103◦ 54 03 E and 103◦ 54 06 E and latitudes 22o 18 08 N to 22o 18 11 N. On August 05, 2019, at around 09:00 am (local time), a serious landslide occurred on the provincial road No. 152 (Fig. 1b) resulting in more than 300 m3 of loose soil and rock to slide down and lead to the death of one person. The incident was the result of continuous torrential rain that lasted for several days after a severe storm attacked Lao Cai province.

Fig. 1. a) Location of the study area, b) aerial photo shows the study site, the yellow dotted line presents the actual slide scar.

Sapa has experienced a large number of landslides, slope failures, and soil erosion events compared to other districts in northern Vietnam [28]. The major reason might be

214

T. V. Tran et al.

because the town belongs to a high precipitation zone of the Hoang Lien Son mountain range where the total annual rainfall can reach between 2000 up to 3600 mm. The rainfall amount is not evenly distributed with space and time when about 80% - 85% of the total annual rainfall is often recorded in summer. The period with the maximum rainfall amount is usually observed in the period from June to August.

(a)

(b)

Fig. 2. a) Landslide in the field and b) a close look at the conditions on the surface of rupture.

The study area has a very complicated geological structure consisting of debris deposits of granite with many large and small tectonic faults cut through [22, 26]. Referring to the geological conditions, the stratigraphic units contain sedimentary, metamorphic, and igneous rocks formed at different ages [25]. Large cleavage and distribution of geological formations with different ages and origins are directly correlated to tectonic faults in the area [29]. The cover soil layer is highly weathered and most of the slopes in the study area are composed of landslide landforms. Thus, one can be seen in Figs. 2a and 2b, a closer look reveals that landslides took place on the boundary between the impervious hard layer and the cover soil layer. The soil on the slip surface was loose and observed to be in the saturated condition at the time of failure.

3 Application of Scoops3D for the Three-Dimensional Stability Assessment of Slopes at the Regional Scale 3.1 Three-Dimensional Slope Stability Assessment Using Scoops3D Scoops3D is one of the many programs that have enjoyed the benefits of the development of GIS. The program applies the 3D LEM and the “method of columns” to find the potential sliding masses which are assumed to be the intersections of spherical surfaces with the soil columns defined by the raster grid cells [14]. For each potential failure, Scoops3D calculates the stability against the rotation along the portion of the spherical

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

215

surface. One advantage of the program is that Scoop3D is at variance with many 3D models so that the program can consider the topography described by the whole DEM and the ability to locate all the potential sliding masses throughout the landscape, but Scoop3D is not for an individual landslide in a small area. In short, the merit of Scoops3D as opposed to other existing programs is the slip surface searching technique. In Scoops3D, the equilibrium of forces and moments is assured for every column as well as for the total slip surface mass. In general, all LEMs define the factor of safety as the ratio of the average shear resistance (strength), s, to the shear stress τ, needed for maintaining the limiting equilibrium along a predetermined trial surface: Fs =

s τ

(1)

The shear strength of the soil on the trial surface is calculated based on the linear Coulomb-Terzaghi failure criterion in Eq. 2. s = c + [σn − u]tanφ 

(2)

where: c is the effective cohesion; φ  is the effective internal friction angle; σn is the normal stress, and u is the pore-water pressure acting on the shear surface. In Eq. (1), limiting equilibrium results when Fs = 1. A value of Fs < 1 points out that the slope is theoretically unstable. In an unsaturated soil environment, PWP presents negative values and thus increasing the shear resistance [30]. Scoops3D is able to consider the influence of negative PWP. However, in this study, due to the limitation of input data, the influence of negative PWP is ignored. Scoops3D analyzes spherical failure surfaces by expanding Bishop’s Simplified Method [31] into three dimensions. In this approach, the side-forces on soil columns are horizontal (with no net shear stress between slices) and are assumed to be ignored. By summing for all columns within the potential slip surface, the governing equation for the Scoops3D analysis is given in Eq. (3) [32].      Ri,j ci,j Ahi,j + Wi,j − ui,j Ahi,j tanφi,j /mαi,j  (3) Fs = Wi,j (Ri,j sinαi,j + keq ei,j ) Where Ahi,j is the horizontal area of the trial surface at the base of the column (i, j); R(i,j) is the resisting force arm or the failure surface radius; W i,j represents the weight of the column (i, j) above the slip surface; α i,j stands for the apparent dip of the column base in the direction of rotation; keq is the horizontal pseudo-acceleration coefficient from earthquake shaking; ei,j is the horizontal driving force moment arm for a column (equal to the vertical distance from the center of the column to the elevation of the axis of rotation), and mα i,j = cosε i,j + (sinα i,j tanφ i,j )/Fs3D ; with ε i,j = the true dip angle of the trial surface and the horizontal surface as described in Fig. 3. It should be noted that in 3D, the value of all the parameters with subscript (i, j) may vary from column to column. More details about the extended Bishop’s Simplified

216

T. V. Tran et al.

Fig. 3. Schematic diagram showing the slip direction, ε – apparent dip in the direction of slip, α, and δ - azimuthal direction of slip, (modified from Reid et al. (2015) [14])

Method are provided in the Scoops3D manual book [14]. Figure 4 illustrates the conceptual framework of the application of Scoops3D in landslide prediction. As can be seen, like other physical-based models of slope stability calculation, Scoops3D requires various types of input data corresponding to the spatial distribution of the soil engineering properties, soil thickness, groundwater table, the size criteria for potential failures, and the topographical conditions. The accuracy of the predicted results by Scoops3D depends significantly on the quality of the input data [14].

Soil depth

DEM

Soil unit weight

Shear strength parameters

Limit equilibrium Method

Groundwater table condition

Size criterias for potential failures

Fs map

Landslide validation

Fig. 4. Conceptual framework for the use of Scoops3D in landslide prediction

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

217

3.2 Determination of the Potential Slip Surfaces in Scoops3D In 3D slope stability analysis, a search with numerous trial surfaces is required due to the variation in the local topography, the 3D distribution of material properties as well as the PWP [14]. Depending on the size criteria defined by users, Scoops3D broadly examines the DEM for potential failure masses. It simulates the stability of millions of potential slip surfaces [33]. The search grid points located above the DEM (Fig. 5) is systematically searched. Each point is the centroid of some potential failure surfaces. From this point, Scoops3D detects an initial radius, and the intersection between the spherical surface and the DEM (Fig. 6) is found. When the search is completed, every DEM grid point of interest is included in some potential landslides [14]. Therefore, the advance of Scoops3D in comparison with other current programs is the slip surface searching technique. Depending on the time and the power of the computer; users can set a reasonable range of the 3D grid of centers above the DEM and the center of this sphere can be any arbitrary point above the DEM. In this study, the potential failures are limited by a volume between 60 to 1000 m3 . Landslides of this magnitude are estimated to occur in the study site based on the assumed depth of the cover soil layer and the dimensions of the site. In order to consider all potential slip surfaces, the “search-box” technique is applied. Following this technique, the vertical extent of the search lattice is set from the lowest elevation on the DEM to the elevation when there is no change on the corresponding predicted stability map. The horizontal extent is followed by the boundary of the DEM. The radius at each search grid point is incremented by 0,5 m until the volume reaches the maximum specified value (1000 m3 ). Figures 5 and 6 illustrate the principle of how potential slip surfaces are found using the search box.

Fig. 5. 3D search lattice above a DEM. Each dot denotes the center of multiple spherical trial surfaces (modified from Reid et al. (2015) [14])

218

T. V. Tran et al.

Fig. 6. Three-dimensional overview of a DEM with one potential trial slip surface. Above is one layer of the search grid. Each dot denotes the center of multiple spherical trial surfaces (modified from Reid et al. (2015) [14])

3.3 Determination of the Slip Direction The 2D simplification typically corresponds to the worst situation when the direction of the slip surface is often not considered. SCOOPS3D is able to simulate the stability for a slip surface in any direction anywhere within the DEM, not just the slip aligned with x-y orthogonal coordinates. Different potential slip directions may lead to a differently simulated Fs. According to Reid et al. (2015) [14], Scoops3D always computes the stability of the sliding mass in the overall fall direction, defined as the average ground surface for all full DEM cells encompassed by the potential sliding mass. The fall direction is simulated using the arctangent of the sum of the slopes for the raster DEM cells in the xand y-directions, converted to degrees, translated to a range of 0 to 360 degrees where 0 degree is the positive x-axis. The slip direction that yields the lowest Fs is a function of the distribution of stresses.

4 Data Used for Slope Stability Assessment on a Regional Scale 4.1 DEM and DEM Resolution DEM representations are easily reachable due to the increase in the development of satellite images and aerial photography [34]. In 3D slope stability analysis, a search with numerous trial surfaces is required due to the variation in local topography, the 3D distribution of material properties, and the water pressure [14]. Scoops3D systematically examines the DEM for potential failure masses based on the size criteria and the search

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

219

grid defined by the user. To consider all potential slip surfaces, the vertical extent of the search lattice is from the lowest elevation on the DEM to the elevation when there is no change on the stability map is found. The horizontal extent is followed by the boundary of the DEM. The radius at each search grid point is incremented by 0,5 m until the volume reaches the maximum specified value (1000 m3 ). DEM is a convenient 3D representation for the modeling of slope stability problems because it can manually extract other information from the GIS such as spatial and image data [33]. The resolution of the DEM used for landslide modeling is selected depending on the quality and density of input data, the size of the study area, the required resolution of the output maps as well as the relative size of the sliding scars [14, 35]. In the case of Scoops3D, increasing the resolution of the DEM provides more active columns for a given trial surface, thereby increasing the accuracy of calculated landslide volume and area [14]. According to Reid et al. (2010) [36], Scoops3D typically produces reasonable estimates of potential landslides using as few as about 200 active columns in a potential failure mass or a trial failure with at least 200 columns. This number is enough to produce a predicted factor of safety map results within 1,0% of failures represented by many columns. In this study, we utilized the AW3D 1,0 m DEM, which was calculated by an image matching process that employed a stereo pair of optical images taken with Digital Globe satellite constellation. Figure 7 illustrates the DEM of the study area.

Fig. 7. Digital elevation model of the study area

220

T. V. Tran et al.

4.2 The Spatial Distribution of Soil Depth Defining an accurate soil profile for the input of Scoops3D plays an important role in deciding the quality of the predicted Fs map. However, the spatial distribution of soil thickness is a function that depends on the complicated interactions of numerous factors such as the underlying lithology, the slope gradient, the hillslope curvature, the upslope contributing area, and the vegetation cover [3, 37]. Therefore, the estimation of the soil thickness distribution remains a challenging and expensive task [3]. Scoops3D can represent the subsurface distribution of material properties by a series of layers. The elevation of the lower boundary of each layer is defined by using an ASCII raster grid which is contained in a separate file. However, for the site in this study, the field survey shows that the soil profile can be represented by two layers: the residual soil layer and the bedrock layer. The former is observed to have a thickness of about 4,0m at the deepest part. Therefore, in this study, for a conservative reason, a uniform thickness of 4,0m of the cover residual soil layer was used for the entire area. The soil profile is assumed to include two layers: the cover soil layer and the bedrock layer as presented in Fig. 8.

Fig. 8. Cross-section that cuts through section A-A in the study area

4.3 Determination of Critical Hydrological Conditions Within the Soil Caused by a Rainstorm The difficulty in determining the spatial distribution of the PWP field is due to the complicated, spatially- and chronologically-dependent groundwater flows [3]. Thus, slope stability evaluation often uses simplified assumptions to define the pore-pressure field [38]. Scoops3D permits users to select from several different options to take the effects of PWP into account in the slope stability calculation [14]. These include: 1) Ignore the role of groundwater pressure, 2) Use the pore-pressure ratio (ratio of pore pressure to vertical stress at a point), Ru,

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

221

3) Input a piezometric surface (input a piezometric surface that represents the water table with vertically hydrostatic pressure heads beneath the surface [13]) 4) Input a 3D distribution of saturated pore-pressure heads: for this option, Scoops3D requires a separate 3D file containing a list of locations with corresponding values of the pressure head 5) Input a 3D distribution of variably saturated pore-pressure heads: For this option, a 3D map containing locations of corresponding variably saturated pressure head and water content and parameters for van Genuchten soil-water characteristic curve is required. Each of these options makes different assumptions about the distribution of PWP, which results in small variations in the Fs equation employed in Scoops3D. In this study, based on the data available as discussed in the previous section, option (3) was selected. In the study area, rainfall might be the most common factor that triggers landslides [25]. As rainwater infiltrates into the unsaturated zone of the soil slope, it leads to the rise in the groundwater table, which in turn reduces the shear strength of the soil [39]. Based on the information observed in the field after the occurrence of the landslide as presented in Fig. 2b, the spatial distribution of the groundwater table at the time of simulation (the time when the landslide was observed to occur) is assumed to coincide with the ground surface at the time of failure. This assumption is acceptable since the landslide was caused by a historical rainstorm in the last 20 year. Figure 9 illustrates the relationship between rainfall intensity versus duration and the cumulative rainfall lasting from 04.00 on August 03, 2019, to 13.00 August 05, 2019, in Sapa, Lao Cai.

10.0

180.0 Rainfall intensity

160.0

cumulative rainfall

140.0 120.0

8.0

100.0

6.0

80.0 60.0

4.0

40.0

2.0

Cumulative rainfall [mm]

Rainfall intensity [mm]

12.0

20.0 0.0

0.0 4 7 10 13 16 19 22 1 4 7 10 13 16 19 22 1 4 7 10 13 Time of the day [hr]

Fig. 9. Relationship between rainfall intensity versus duration and the cumulative rainfall lasting from 04.00 on August 03, 2019, to 13.00 August 05, 2019, in Sapa

222

T. V. Tran et al.

After the piezometric surface option is selected, a map containing the elevations of the piezometric surface to determine pore pressures acting on the trial surface at the base of each column is required as one of the input data. This option assumes that pore pressures are hydrostatic with vertical depth. Therefore, the pore water pressure at depth zpz can be defined by: u = zpz γw

(4)

where zpz is the vertical depth below the piezometric surface and can vary in different columns, γw is the unit weight of water. 4.4 Determination of Soil Parameters for Slope Stability Assessment The elevation of the lower boundary of each layer is defined by an ASCII raster grid, each of which is in a separate file. As discussed in Sect. 4.2, the soil profile in this site is assumed to have two layers: the cover soil where the landslide took place and the bedrock layer. Scoops3D uses strength properties (including the cohesion and the friction angle) of the layer intersected by the trial surface for each column in a potential failure mass for computation of slope stability [14]. Specifically, Scoops3D uses the spatial distribution of soil cohesion (c) and the friction angle φ values of the material intersected by the trial surface in the center of the column for computation of slope stability. For this study site, the cover soil layer is classified as ML (inorganic silt with low to medium compressibility) according to the Unified Soil Classification System. The index properties and some grain sizes of the cover soil layer are presented in Table 1. Table 1. Physical properties of the residual soil layer Gravel (%)

Sand (%)

Silt (%)

Clay (%)

Liquid limit

Plastic limit

6,4

22,8

50,5

20,3

43,5

31,8

Major soil parameters for the application of Scoops3D including the unit weight and shear strength parameters were identified by laboratory tests: the drive-cylinder test to define the unit weight and direct shear test to define the shear strength parameters of soil. Each soil parameter is defined by averaging the values of 4 tests. The soil samples were taken from the surrounding sliding scar, their values are illustrated in Table 2.

5 Landslide Evaluation The performance measure is essential in landslide modeling and prediction. In this study, to validate how well the predicted landslide and the actual landslide scar fit each other, the Modified Success Rate (MSR) suggested by Huang and Kao (2006) [22] was used. The MSR is the modification of the Success Rate (SR) developed by Montgomery & Dietrich (1994) [40] to deal with the over- and under predicted problems of model

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

223

Table 2. Soil parameters used for slope stability assessment using Scoops3D Parameters

Symbol

Unit

Value range

Selected value

Unit weight

γ

kN/m3

18,3 ÷ 19,0

19,0

Friction angle

ϕ

(o )

15,0 ÷ 16.5

15,0

Cohesion

c

kN/m2

11.0 – 13,0

11,0

performance. In general, comparing the Total number of Actual Unstable cells (TAU), the Total number of Actual Stable cell (TAS) and those in the predicted raster Fs map, there are four kinds of possible outcomes: 1) 2) 3) 4)

the Actual Unstable cells are predicted as Unstable cells (AUU), the Actual Unstable cells are predicted as Stable cells (AUS), the Actual Stable cells are predicted as Unstable cells (ASU), and the Actual Stable cells are predicted as Stable cells (ASS).

It is obvious that a higher number of raster cells in types 1 and 4 results in betterpredicted results [35]. According to Huang and Kao (2006) [22], the value of MSR can be simulated as follows: MSR = 0.5

ASS ASS AUU + 0.5 = 0.5SR + 0.5 TAU TAS TAS

(5)

where TAU = total number of actual unstable cells, TAS = total number of actual stable cells, AUU = actual unstable cells are predicted as unstable cells (number of successfully predicted landslides), ASS = actual stable cells are predicted as stable cells (number of successfully predicted stable cells), In Eq. 5, the performance value derived by MSR ranges from 0.0 to 1.0. According to Huang and Kao (2006) [22], the simulation derived by MSR would be the best when the MSR value is in the range from 80% to 90%. When MSR < 80%, landslide overprediction is likely to occur in the predicted Fs map while an MSR value that is higher than 90% would result in the under-prediction of the predicted Fs map.

6 Results When all the input data is available, Scoops3D will simulate the factor of safety for all raster cells within the digital landscape. Except for the Fs map, one of the very important outputs of Scoops3D is the map showing the updated topographical conditions when all unstable masses (area with Fs < 1) are removed. Specifically, a new ground surface

224

T. V. Tran et al.

that is coincident with the trial surfaces of these specified potential failure masses is created. Figure 10 presents the updated topographical conditions of the study site when all unstable masses are removed. It is clear that compared to the original DEM (Fig. 7), most changes on the topographical map mostly occur in the actual sliding scar. The predicted sliding mass is located near the foot of the slope. This achieved map can be used for the simulation of future geological disasters in the same site or estimation of the shape and size of the potential sliding masses.

Fig. 10. Updated topographical conditions when all unstable masses are removed

Figure 11 illustrates the predicted Fs map for the study site. In this Figure, the Fs value was classified according to the classification system suggested by Mandal & Maiti (2005) [41] who divides the stability results into four different classes: 1) Stable (Fs ≥ 1.5): only major destabilizing factors lead to instability; 2) Moderately stable (1.25 ≤ Fs < 1.5): moderate destabilizing factors result in instability; 3) Quasi-stable (1 ≤ Fs < 1.25): minor destabilizing factors lead to instability, and 4) Unstable (Fs < 1), stabilizing factors are needed for stability. For the validation of the predicted results, with the definition of the factor of safety, we maintain that the only important distinction is strictly between Fs < 1 (for unstable cells) and Fs ≥ 1 (for stable cells).

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

225

As can be seen, the predicted sliding area does not completely fit the landslide scar; however, it somehow shows the exact location of the actual landslide when almost half of the actual landslide body was accurately predicted. This predicted result is understandable since numerous simplified assumptions were used for the input data and the hydrological conditions within the slope at the observed time of failure.

Fig. 11. Fs map predicted by Scoops3D using Bishop simplified method, the dotted line shows the sling scar

Table 3 shows the value of TAU, TAS, AUU, ASS, and the result of SR and MSR. As can be seen, within the actual landslide scar, about 48% of all the cells were accurately predicted and the value of MSR in Table 3 shows an under-predicted result. It is understandable since numerous simplified and conservative assumptions were applied in the Scoops3D model, such as assumptions related to the spatial distribution of soil depth, the subsurface hydrological conditions caused by the rainstorm as well as the input shear strength parameters of soil.

226

T. V. Tran et al. Table 3. Simulation of MSR Parameters Value TAU

649

TAS

6461

AUU

308

ASS

6337

SR

0.48

MSR

0.73

7 Conclusion This study adopted the U.S. Geological Survey Scoops3D slope stability model, which is based on the column limit-equilibrium method to evaluate landslide potential in threedimensional space. Comparing the actual landslide scar and the predicted unstable area, the simulated SR value shows that 48% of the total raster cells within the actual landslide scar were accurately predicted. MSR = 0.73 shows an under-predicted result of the predicted unstable area. This outcome is understandable since numerous simplified and conservative assumptions were employed during the establishment of the model such as assumptions related to the spatial variation of the soil depth, soil physical properties, as well as the spatial-time varying groundwater table. Therefore, it can be concluded that with reliable input data, Scoops3D alone is a reasonably good predictor of slope failures within the study area. The approach is capable of predicting the locations of future landslides with moderate accuracy. Therefore, for future works, more studies are needed to improve the accuracy of the predicted Fs map both in the spatial and temporal aspects. One can evaluate the effects of the uncertainty of the input data on the predicted Fs map using Scoops3D or find a suitable tool for the simulation of the rainfall-induced pore water pressure change for the slope stability assessment. Further studies should also be conducted to make use of the updated DEM when all unstable masses are removed for the prediction of the occurrence of future landslides in the study area. Acknowledgment. The support from the Vietnam Ministry of Science and Technology under the Grant NÐT 67/e-Asia19 is gratefully acknowledged.

References 1. Johari, J., Javadi, A.A.: Reliability assessment of infinite slope stability using the jointly distributed random variables method. Scientia Iranica. 19(3), 423–429 (2012) 2. Chakraborty, A., Goswami, D.: State of the art: three dimensional (3D) slope-stability analysis. Int. J. Geotech. Eng., 1–6 (2016). https://doi.org/10.1080/19386362.2016.1172807 3. Peng, W., Mo, J., Xie, Y.: Comparison for the results from 2D and 3D analysis for slope stability. Appl. Mech. Mater. 90–93, 255–259 (2011)

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

227

4. Tran, T., Alvioli, M., Lee, G., An, H.U.: Three-dimensional, time-dependent modeling of rainfall-induced landslides over a digital landscape: a case study. Landslides 15, 1071–1084 (2018). https://doi.org/10.1007/s10346-017-0931-7 5. Zhang, K., Cao, P., Liu, Z., Hu, H., Gong, D.: Simulation analysis on three-dimensional slope failure under different conditions. Trans. Nonferrous Met. Soc. China 21, 2490–2502 (2011). https://doi.org/10.1016/S1003-6326(11)61041-8 6. Kalatehjari, R., Ali, N.A.: Review of three-dimensional slope stability analyses based on limit equilibrium method. EJGE 18, 119–134 (2013) 7. Trigila, A., Iadanza, C., Spizzichino, D.: Quality assessment of the Italian Landslide Inventory using GIS processing. Landslides 7(4), 455–470 (2010) 8. Jia, N., Yang, Z., Xie, M., Mitani, Y., Tong, J.: GIS-based three-dimensional slope stability analysis considering rainfall infiltration. Eng. Geol. Environ. 74(3) (2014). https://doi.org/10. 1007/s10064-014-0661-1 9. Quiu, C., Xie, M., Esaki, T.: Application of GIS technique in three-dimensional slope stability analysis. Paper presented at the International Symposium on Computational Mechanics, Beijing, China (2007) 10. Hovland, H.J.: Three-dimensional slope stability analysis method. J. Geotech. Geoenviron. Eng. 103(9), 971–986 (1977). https://ojps.aip.org/gto 11. Hungr, O.: An extension of Bishop’s simplified method of slope stability analysis to three dimensions. Geotechnique 37(1), 113–117 (1987) 12. Hungr, O., Salgado, F.M., Byrne, P.M.: Evaluation of three-dimensional method of slope stability analysis. Can. Geotech. J 26, 679–686 (1989) 13. Lam, L., Fredlund, D.G.: A general limit-equilibrium model for three-dimensional slope stability analysis. Can. Geotech. J. 30, 905–919 (1993) 14. Reid, M.E., Christian, S.B., Brien, D.L., Henderson, S.T.: Scoops3D—software to analyze three-dimensional slope stability throughout a digital landscape (Version 1.0). U.S. Geological Survey, Virginia (2015) 15. Spencer, E.: A method of analysis the stability of embankments assuming parallel inter-slice forces. Geotechnique 17(1), 11–26 (1967) 16. Xie, M., Esaki, T., Cai, M.: GIS-Based Implementation of three-dimensional limit equilibrium approach of slope stability. J. Geotech. Geoenviron. Eng. 132, 656–660 (2006). https://doi. org/10.1061/ASCE1090-02412006132:5656 17. Xie, M., Esaki, T., Qiu, C., Wang, C.: Geographical information system-based computational implementation and application of spatial three-dimensional slope stability analysis. Comput. Geotech. 33, 260–274 (2006). https://doi.org/10.1016/j.compgeo.2006.07.003 18. Mergili, M., Marchesini, I., Alvioli, M., Metz, M., Schneider-Muntau, B., Rossi, M., Guzzetti, F.: A strategy for GIS-based 3-D slope stability modelling over large areas. Geosci. Model Dev. 7, 2969–2982 (2014). https://doi.org/10.5194/gmd-7-2969-2014 19. Mergili, M., Marchesini, I., Rossi, M., Guzzetti, F., Fellin, W.: Spatially distributed threedimensional slope stability modelling in a raster GIS. Geomorphology 206, 178–195 (2012). https://doi.org/10.1016/j.geomorph.2013.10.008 20. Neteler, M., Mitasova, H.: Open Source GIS: A GRASS GIS Approach. Springer, New York (2007) 21. Pham, H., Bui, X.: Analysis of post-landslide electric imaging data at a site in Sapa, Vietnam. Paper presented at the EAGE-GSM 2nd Asia Pacific Meeting on Near Surface Geoscience & Engineering, Kuala Lumpur, Malaysia (2019). https://doi.org/10.3997/2214-4609.201900395 22. Huang, J.C., Kao, S.J.: Optimal estimator for assessing landslide model performance. Hydrol. Earth Syst. Sci. 10(6), 957–965 (2006). https://doi.org/10.5194/hess-10-957-2006

228

T. V. Tran et al.

23. Nguyen, Q.P., Phuong, N., Nguyen, K.L.: Statistical and heuristic approaches for spatial prediction of landslide hazards in Laocai, Vietnam. In: International Symposium on Geoinformatics for Spatial Infrastructure Development in Earth and Allied Sciences, Ho Chi Minh city, p. 7 (2012) 24. Nguyen, V.C., Dao, V.T.: Investigation and research of landslide geohazard in north-western part of Vietnam for the sustainable development of the territory, pp. 269–280. Osaka Univ. Knowl. Arch. OUKA, Osaka (2007) 25. Bui, D., Tran, A., Hoang, N., Thanh, N., Nguyen, D.: Spatial prediction of rainfall-induced landslides for the Lao Cai area (Vietnam) using a hybrid intelligent approach of least squares support vector machines inference model and artificial bee colony optimization. Landslides 14, 447–458 (2017). https://doi.org/10.1007/s10346-016-0711-9 26. Tran, T., Pham, H., Hoang, V., Trinh, M.: Soil type, rainfall infiltration and the stability of unsaturated cut-slopes. Paper presented at the International Symposium on Lowland Technology, Hanoi (2018) 27. Tran, T., Trinh, M., Lee, G., Oh, S., Nguyen, T.: Effect of extreme rainfall on cut slope stability: case study in yen Bai city, Viet Nam. J. Korean Geo-Environ. Soc. 16(4), 23–32 (2015). https://doi.org/10.14481/jkges.2015.16.4.23 28. Dang, K., Burkhard, B., Muller, F., Dang, V.: Modelling and mapping natural hazard regulating ecosystem services in Sapa, Lao Cai province. Vietnam. Paddy Water Environ. 16, 767–781 (2018). https://doi.org/10.1007/s10333-018-0667-6 29. Nguyen Duc, M., Tran, Q.H.: Features of large-scale landslide at Hau Thao area, Sa Pa town, Lao Cai Province. Paper presented at the Geotechnics for Sustainable Infrastructure Development, Hanoi (2019). https://doi.org/10.1007/978-981-15-2184-3_119 30. Tran, T.V., Lee, G., An, H.U., Kim, M.: Comparing the performance of TRIGRS and TiVaSS in spatial and temporal prediction of rainfall-induced shallow landslides. Environ. Earth Sci. 76(315), 1–16 (2017). https://doi.org/10.1007/s12665-017-6635-4 31. Bishop, A.W.: The use of the slip circle in the stability analysis of slopes. Geotechnique 5(1), 7–17 (1955). https://doi.org/10.1680/geot.1955.5.1.7 32. Brien, D.L., Reid, M.E.: Modeling 3-D slope stability of coastal bluffs, using 3-D groundwater flow, Southwestern Seattle. Washington. Retrieved from U.S, Geological Survey (2007) 33. Reid, M.E., Christian, S.B., Brien, D.L.: Gravitational stability of three-dimensional stratovolcano edifices. J. Geophys. Res. 105(B3), 6043–6056 (2000) 34. Tun, Y., Marcelo, A., Pedroso, D., Scheuermann, A.: Multimodal reliability analysis of 3D slopes with a genetic algorithm. Acta Geotech. 14, 207–223 (2019). https://doi.org/10.1007/ s11440-018-0642-9(0123456789(),-volV)(0123456789().,-volV) 35. Tran, T.V., Lee, G., Kim, M.: Shallow landslide assessment considering the influence of vegetation cover. J. Korean Geo-Environ. Soc. 14(4), 17–31 (2016). https://doi.org/10.14481/ jkges.2016.17.4.17 36. Reid, M.E., Keith, E.C., Kayen, R.E., Iverson, N.R., Iverson, R.M., Brien, D.L.: Volcano collapse promoted by progressive strength reduction: New data from Mount St. Helens. Bull. Volcanol. 72, 761–766 (2010). https://doi.org/10.1007/s00445-010-0377-4 37. Tesfa, T.K., Tarboton, D.G., Chandler, D.G., McNamara, J.P.: Modeling soil depth from topographic and land cover attributes. Water Resour. Res. 45(10), 1–16 (2009). https://doi. org/10.1029/2008WR007474 38. Tran, T.V., Lee, G.H., Trinh, M.T., An, H.U.: Effect of digital elevation model resolution on shallow landslide modeling using TRIGRS. Nat. Hazards Rev. 18(2), 1–12. https://doi.org/ 10.1061/(ASCE)NH.1527-6996.0000233 39. Tran, T.V., Lee, G.H., Oh, S., Kim, M.: Effect of rainfall patterns on the response of water pressure and slope stability within a small catchment: a case study in Jinbu-Myeon, South Korea. J. Korean Geo-Environ. Soc. 17(12), 5–16 (2016). https://doi.org/10.14481/jkges. 2016.17.12.5

Use of Scoops3D and GIS for the Assessment of Slope Stability in Three-Dimensional

229

40. Montgomery, D.R., Dietrich, W.E.: A physically-based model for the topographic control on shallow landsliding. Water Resour. Res. 30(4), 1153–1171 (1994) 41. Mandal, S., Maiti, R.: Semi-Quantitative Approaches for Landslide Assessment and Prediction. Springer Natural Hazards, Singapore (2015). https://doi.org/10.1007/978-981-287146-6

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment of the Cliff at Chau Thoi Quarry Nguyen Quang Tuan(B) Thuyloi University, 175 Tay Son, Dong Da, Hanoi, Vietnam [email protected]

Abstract. Chau Thoi stone quarry is an old open pit mine in Binh Duong province of Vietnam, where the cut slopes are the sources of prone rockfall. Several rockfalls have occurred in this area and rockfall threatens the tourists. This case is also typical of a number of stone mines in the region. In the study, geological measurements are taken to obtain the field condition with the focus on the rock fractures. Based on the data, kinematic analysis is carried out using the DIPS software for the identification of rock slope failures. The results are then used in the analysis of rockfall. Herein, rockfall simulations are carried out by the rigid body method using the RocFall software for the analysis of bounce height, maximum runout distance, trajectories, velocity, and intensity of falling rock block. Observation at the site may be used for the back analysis. The result shows the modes of slope failures that may occur and sites that are prone to failure. The model parameters can be investigated. The vulnerable distance of rock traveling, bounce height and trajectories, velocity, and energy of rock block will be in the results. The result of this study may be useful for selection and designing protection measures and for the planning of land use hazard areas for further development. The study will be useful reference for other stone mines in the region for both exploiting work and closure of the mine. Keywords: Rockfall · Slope failure · Kinematic analysis · Trajectories · Rigid body model

1 Introduction Rockfalls are the fastest type of landslide and they are geological hazards that occur in mountainous terrains. It causes damage to infrastructures, properties and even loss of lives [1]. A rockfall refers to quantities of rock falling freely from a high position on slope or cliff surface. The term rockfall is also used for the collapse of rock from roof or walls of mine or quarry workings. A rockfall is a fragment of rock (a block) detached by sliding, toppling, or falling, that falls along a vertical or sub-vertical cliff, proceeds downslope by bouncing and flying along ballistic trajectories or by rolling on talus or debris slopes. The traveling of rock fragments depends on various stochastic parameters © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 230–249, 2021. https://doi.org/10.1007/978-3-030-60269-7_12

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

231

which include rock block geometry, slope morphology, and the material properties of both rock block and slope. Rockfall is a common geohazard in many parts of Vietnam. This hazard is also frequently met in open cast mines. Rockfall threatens mining safety and operations. It may cause safety concerns for workers, hauling activity and impact mine production. The impact of rockfall becomes more serious when the mining operation is not well planned nor regulated. Sometimes, the exploitation does not strictly follow the regulations. The mining area is not beyond the limit of safe zone from dwelling area or traffic route. In Vietnam, many quarries are near the roads for the benefit of low transportation cost. Therefore, the rockfall hazard becomes more obvious. Besides, some open-pit mines after closure are not well protected nor some trespass. People and traffic means may access the unsafe zone. Serious accident may happen due to rockfalls. A large number of accidents have occurred due to rockfalls in Vietnam. A number of recent accidents are examples of the danger of rockfalls. To choose countermeasures for slope protection or rockfall mitigation, one must know how the rockfall is, for example, the runout distances and kinetic energy caused by a rockfall. This paper focuses on the slope stability evaluation and rockfall analysis of an old quarry, which had been closed but the rockfall hazards do exist and threaten people. The rockfall analysis was carried out for the section of the bench using the simulation program RocFall version 8.0 [2] to predict the trajectory, velocity, kinetic energy, bounce height, and the runout distances of the falling rock blocks.

2 Study Area 2.1 Description of Chau Thoi Quarry The case of study is about rockfall hazard of a section of cut rock slope at an old quarry, at the foot of Chau Thoi hill, in Binh An commune, Di An town, Binh Duong province. Chau Thoi hill area is a national historical-cultural relic. This place is also a valuable geographical heritage of stratigraphy, geomorphology and the hill peak is the only checkpoint that is able to view the landscape and three major cities of the southern key economic region of Vietnam: i.e. Binh Duong province, Bien Hoa province and Ho Chi Minh City [3]. On top of the hill there is a Chau Thoi Pagoda, the oldest pagoda in Binh Duong. It is one of the oldest pagodas in the South of Vietnam. Therefore, Chau Thoi hill area is also a place for tourists. Every day, many tourists come to visit the pagoda and go around for sightseeing. Therefore, it is necessary to ensure the safety of visitors from the risk of landslides at the Chau Thoi foothills. This is also a typical example of quarries that have been exploited in this region. In the future, after the closure of these quarries, it is necessary to have the plan to use the old mine area for economic and conservation purposes. The utilization of mining areas for various purposes, such as the construction of resort facilities or planning of forest development areas, requires the assessment of the risk of rockfall as well as identifying the zone of danger. A plan of land use for these stone quarries was proposed [4]. Therefore, the solutions to the rockfall problem and the area protection from rockfall are also necessary. The investigated area, where is composed of cut slopes and natural slopes, is located in the western part of Chau Thoi hill. Those are the benches of the old stone quarry. The

232

N. Q. Tuan

old excavated pit is divided into two benches. The upper bench has a height of 10 to 20 m. The lower bench is currently partially submerged in water and belongs to the lake. The upper bench has a steep slope, which is about 70 to 85°. Some parts of the bench are nearly vertical because the bench surface is identical to the fractures. The bench floor, which used to be a hauling road, is a relatively flat surface. This area is the very place where tourists usually enter for sightseeing. There exist layers of debris, i.e. talus, next to the toe of the benches. These materials accumulated due to landslides or due to erosion activity by runoff water and they are called the talus. The material of the talus is mainly composed of rock fragments and is mixed with some fine material. Fig. 1 and shows the general views of the study area.

Direction to highway 1K`

Planned housing area (a)

Rockfall area

Chau Thoi Pagoda

The direction to Núi Nhỏ stone quarry

Rock cliff (bench of the Chau Thoi quarry)

(b) Fig. 1. (a) Arieal view of the study area (Google Earth image); (b) rock cliff where rockfall occured; (the photo was taken in April 2019 by the author)

At some locations on the bench floor, fallen rock blocks are found, including fragments from cut slope and boulders. These blocks have different sizes, from a small size

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

233

of tens centimeters to a large size of about 2 m. Some large blocks moved beyond the middle of the bench floor. Some small block moved to the edge of the bench floor. Among them, some blocks fell very recently. The newly taken photos (Fig. 3) show the shapes and the sizes the fallen rocks observed at the site.

(a)

(b)

Fig. 2. (a) Measurement of joint attitudes; (b) A rock block is prone to detach from rock cliff (the photos were taken in April 2019 by the author)

Fig. 3. An old fallen rock with subrounded shape (left); A new fallen rock with angular shape (right) (the photographs were taken in May 2020)

2.2 Geological Setting According to the geological and mineral resource map [5] and the report [6], the Chau Thoi hill lies in the area of two main formations of sedimentary rocks, i.e. Buu Long (T2 bl) formation and Chau Thoi formation (T2 ct). Also referred to the geological map (Fig. 4), there are two main tectonics fault groups in this region and nearly perpendicular to each other. One fault group aligns Northeast to Southwest, and the other aligns Northwest to Southeast. The tectonic activities caused the rock in this area highly fractured.

234

N. Q. Tuan

Many large tectonic fractures exist and were exposed as a vertical or nearly vertical cliff by the excavation.

Fig. 4. Geological map of the study area (Modified from the Geological and Mineral Resources Map of Vietnam, Piece C-48-XI) [5]

Based on the site investigation at Chau Thoi foothill, the sandstone is found. The sandstone is greenish light gray, clastic textured with angular grains, the particle sizes are of medium to coarse. The main composition includes quartz and feldspar. The rock is thickly bedded to massive, slightly weathered to fresh. The rock mass is moderately jointed. However, according to observation, the joints distribute unevenly. The rock slope is divided into different zones with different fracture densities. The spacing of joints varies in a wide range from a few centimeters to tens of centimeters. Hence the rock blocks between joints have different sizes. The largest size of fallen rock observed at the site is estimated about 3 tons. At the site of the investigated area, it was observed that the slope dip angles of are mainly over 75°. At some locations, the measured slope can be up to 85°. The main dip direction of the slope is southwest. Nevertheless, at some locations, the bench was overcut into the foothills, and local slopes were made with the dip direction to the south. Also as observed on the rock cliff, there are a number of rock blocks lying in an unstable position and being prone to detach from the cliff, for example, in Fig. 2 (b). Besides, there are rolling boulders lying on the natural surface of the hillslope above the cut bench. Many places where rock slides already occurred, mostly wedge-shaped sliding between fractures. There are also locations, where transitional sliding occurred along with the bedding or joint surface. There are also many rolling boulders of different sizes lying on the surface of the bench floor in which some rock blocks are very large. The size can be up to 2 m. These rocks are considered as the results of rockfall processes that occurred in the past, i.e., sliding, falling, and rolling down from the slopes. There exists a layer of material which is mainly composed of rock fragments accumulated next to the cliff toe. This material is the debris from slope failure, which is moved by surface

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

235

runoff water. No groundwater seepage was observed in the study area. Bushes and small trees are found randomly at the foot of the cut slope and on the natural slope surface. The current condition of the rock slope is shown in Fig. 5. Rock blocks tend to detach from slope Locations that rock detached

Accumulated material (talus): rock fragments with soils Fig. 5. Field photo of the investigated rock slope

The fracture characteristics of the rock mass at the slope were surveyed to assess the fracture density and attitudes. The measurements were made right at the rock outcrop on bench face. The joint surfaces are slightly weathered and slightly rough. According to observations and measurements in the field, it can be seen that the fissures can be divided into different groups (Fig. 6). The surface of the joints is slightly weathered, medium roughness. Joint measurement data were gathered to analyze and assess the possibility of slope instability.

Fig. 6. The traces of discontinuities on rock slope

236

N. Q. Tuan

3 Methodology and Data Used Beside the site identification of rock fall source, kinematic analysis was used to evaluate the stability of rock slope. The data needed for kinematic analysis are joint orientations, i.e. dip and dip direction, slope direction, and joint shear strength, i.e., joint friction. The orientations of joints were simply measured using geological compass. The joint shear strength was estimated based on the joint roughness referring to the empirical determination [7]. To evaluate the rockfall behavior, rockfall analysis was carried out using rigid body model. The information needed for the analysis include locations of rockfall sources, physical properties of falling rock material, size and shape of falling blocks; topographical characteristics (profile geometry, roughness, vegetation), hardness of material along the profile. The geometries of profile and the roughness were directly input in the rockfall model. The following sections present in detailed the kinematic analysis and the rockfall analysis, including the methods and the determination of different parameters. 3.1 Kinematic Analysis Rockfalls mainly occur due to the existence of unfavorable discontinuities. The orientation of geological discontinuities is the major factor that controls rock stability. The kinematic analysis is to identify the type of failure and to determine the possibility of rock slope failure governed by a set or sets of unfavorable discontinuities in the slope rock mass. During site investigation, the orientation data of discontinuities, i.e. dip angle and dip direction, were measured at an outcrop at the investigated rock slope face. Kinematic analysis is a method used to analyze the potential for the various modes of rock slope failures, i.e., plane, wedge, toppling failures, that occur due to the presence of unfavorably discontinuities (Fig. 7). The method of kinematic analysis is described in detailed in Hoek and Bray (1981) [8]. Stereographic projection was used to represent the three-dimensional orientations of rock joints, i.e., dip and dip direction of the planes. One plane can be presented on a stereonet as a great circle, a pole or a dip vector. Based on the survey with discontinuity measurement at the investigated rock cliff, the data were input in DIPS software. Using DIPS software, after input the data the scatter plot of joint poles was generated. A scatter plot permits visual analysis of pole distribution by plotting symbols representing the number of approximately coincident poles at a given orientation. For a more straightforward interpretation, the contour plot was used for analyzing pole concentrations. It is used to visualize the clustering of orientation data. Using such a contour plot from given measurement data, five groups of joints were clustered. Observations also shows that the study area has bedding planes and four sets of different systematic joints. Also the major planes for each group were plotted, as shown in Fig. 8. The dips and dip directions for each joint set were determined by the DIPS analysis as shown in Table 1. Kinematic analyses of the discontinuities reveal that the possible types of failure that may occur are planar sliding and wedge sliding. The analysis shows that four major joint sets prevail in the concerned slope. J3 is the critical joint set causing planar failure if the slope plunges to the south (Fig. 9a). The orange curve indicates the slope face. The red area in stereographic projection is

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

237

Fig. 7. Main types of rock slope failures and the corresponding identifications on the stereone: (a) plane failure; (b) wedge failure; (c) toppling failure; and (d) circular failure. (modified after Hoek and Bray [8])

Table 1. Dip and dip direction of rock joint groups Group 1 (J1 ) Group (J2 ) Group 3 (J3 ) Group 4 (J4 ) Group 5 (J5 ) 33/291

73/292

50/182

85/165

14/231

the critical zone. The direction of failure is in the range from NNW to NNE. This dip direction of slope face is not typical for the study section, but it does exist in some local locations where the slope was over cut in the past. This was confirmed by observation at site.

Fig. 8. Stereonets showing the contour plots and clusters of the major discontinuity sets at the study.

238

N. Q. Tuan

(a)

(b)

Fig. 9. Stereo-net plot showing potential planar failures (a) and wedge failure (b)

Figure 9b is the result of kinematic analysis for the case that the rock slope face plunges to southwest with the dip direction of 220°, the dip angle of 85° and the assumed friction angle of 20°. The plot shows that wedge failure may occur due to the sliding between two pairs of major joint sets, i.e. (J2, J3) and (J1 and J3). It is found that wedge failure between J1 and J3 is still able to occur when the friction angle of the joint is up to 40°. Many local locations of wedge slides were observed at the studied slope, and they are the main sources of rockfalls.

Fig. 10. Critical percentage of sliding vs slope dip direction: (a) Wedge failure; (b) Planar failure

The kinematic sensitivity analysis option in RocFall was used to investigate the effect of different parameters on the slope instability corresponding to different failure

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

239

type types. The results show that the study area is likely to occur planar and wedge sliding. Sliding possibility depends mainly on the direction of the slope. According to the analysis results (Fig. 10), with the position of the joint sets as measured at the rock outcrops, the wedge sliding can occur when the slope face has dip direction from 150° to 300°. The entire bench surface in the study area has slope directions in this range and wedge-shaped sliding certainly happens. Also, the planar sliding can occur with slope surface dip direction angles from 160 to 200° and from 270 to 320°. The study area has a number of locations that slopes are plunged to the south, so planar can occur. The traces of slope failures that occurred in the past observed at the site show that the analysis results are very reasonable. The results of these assessments bring forwards the need for a rock fall analysis. 3.2 Rockfall Analysis 3.2.1 Rockfall Modeling Approaches There are different approaches to assess the rockfall hazard and estimate the kinematic movement of rockfall. The simplest way is to use the empirical model. The empirical model bases on the statistical data for different geological and geomorphological conditions. This is usually used to estimate the runout distance of rockfall. A number of studies about the empirical model are briefly described by [9] and [10]. Experimental approach is usually used to investigate the mechanism of rockfall and the influence of different factors. This approach normally gives good results but it is costly. The experiments are usually carried out to aid or validate the other modelling approaches. There are different modeling approaches to simulate the process of rockfall. They can help to predict the trajectory and kinematic energy of rock block. Modelling can be performed using numerical method or mathematical method. Discrete element method (DEM) is the numerical method which is widely used to simulate the rockfall behavior. The current advantage of DEM allows to simulate the real behavior of rock fall, including the interaction between the rock blocks and even the breakage of rock block during moving process. The real shape of the falling rock block can be described. Nevertheless, this method is very complicated and time consuming. The mathematical method is more effective and simpler than DEM. This method uses the equations of motion and kinematic energy to describe the movement of rock block. This approach can use one of two main models, i.e., the lumped mass model and the rigid body model. The description of these two models are summarized by Volkwein et al. [11] and Leine et al. [12]. The lumped mass model considers a falling rock as a very small spherical particle with mass. The mass of the falling rock is used to calculate the trajectory and the kinematic energy. No size nor shape is considered by this model despite the fact that size and shape may affect the trajectory. Three algorithms are assigned to the model, i.e., the particle algorithm, the projectile algorithm, and the sliding algorithm. The first algorithm is to set up the initial conditions. The second one is to calculate the trajectory of the rock. The last one is used to calculate the movement of rock while it is in contact with the slope. Lumped mass models can only represent sliding motion and mimics rotation with a zero friction angle [13].

240

N. Q. Tuan

The rigid body model takes into account the size and the shape of the rock block. This model was applied by Ashayer [14] and then applied in 3D by Basson [13]. The mass of the rock, shape, and given rock density determines the size of the rock block. The use of rock shape allows us to calculate the rolling and sliding movement. The rigid body model uses the assumption that contact between a falling rock and the slope surface is instantaneous and the contact area between the colliding bodies is very small. A compressive comparison between two models, i.e., the lumped mass model and the rigid body model, was presented by Dadashzadeh et al. [15]. 3.2.2 Rockfall Model In this study, the movement of rock blocks after failure was simulated by a 2D model using the rigid body model with the aid of the software RocFall from RocScience Inc. [2] to evaluate their distribution of out-run and assessing the dynamics of the rockfalls. RocFall is a useful computer program that uses mathematical models based on the laws of motion and the theory of collision. The trajectories, endpoints (fall out distances), bounce heights, velocities, and kinetic energies of the falling rock at any point can be calculated. RocFall can also be used to aid in designing rockfall solutions, i.e., ditch, barrier, berm, fence, etc. The material properties can be calibrated to reproduce the results conforming with the actual rockfalls. The rigid body model was selected for analysis to take into account the size and the shape of a falling block. The falling rocks have different sizes and shapes those are chosen based on common fallen rock observed at the field. Since reproducing the real shapes is not feasible, especially when using a 2D model, the selected rock shapes are some of the relatively simple and typical ones. The sizes were selected based on the dimensions of loosen rock blocks in the jointed rock mass, rock blocks that have fallen to the bench floor, and the boulders (rounded and subrounded) at risk of separation from the slope. The sizes of rock block were automatically calculated depending on the shape, material density and mass. The rock shapes were selected from the library of rock type in RocFall. Rounded shapes were selected for the isolated boulders on the natural slope surface. Polygon shapes were selected for rock blocks that may detach from joint failures. The different selected shapes are shown in Fig. 12. Different typical ranges of size were chosen by defining the mass and rock density. The ranges of rock block sizes were determined by estimation of the real size of rock blocks observed at the site. Large rocks have a mass of 3000 kg, medium rocks have mass of 500 kg and small rocks have a mass of 200 kg. To consider the probability distribution of rock size, normal distribution was also selected for mass variations. The detailed geometries of rock blocks are given in Table 2. The analyses were performed for two profiles. The investigated profiles were selected for the most likely slides that may occur as a result of the kinematic analysis. The first one, i.e., profile 1, is for a typical cross-section of the study area. The second one, i.e. profile 2, is for the special section where the bench was overcut and planar sliding may occur. The elevations of slope geometry were gathered from the topographic map with a scale of 1/2000 in combination with the measurement of typical profiles at the site. Two profiles for rockfall analysis are shown in Fig. 11.

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

241

The sources of falling rocks were specified based on the sources that were identified via field investigation accompanied by kinematic analysis. Two different sources were identified for rock falls including the potential sliding rock blocks in the cut slope and the displaced stones (boulders) lying on the natural slope surface near the bench crest. The positions of the fallen boulders were placed at the peak of the bench for conservative analysis. Line seeders were used to specify the range of locations. Each seeder was assigned to 100 rock blocks, i.e. in total of 300 rock blocks. Table 2. Sizes and shapes of falling rocks Rock body

Shapes

Mass (kg)

Number of blows

Large rock

Triangular, Square, Pentagon, Rectangular

3000 ± 500

100

Medium rock

Triangular, Square, Pentagon, Rectangular

500 ± 50

100

Small rock (fallen boulders)

Smooth square, smooth triangular, smooth pentagon, egg, rhombus, smooth rectangular

200 ± 50

100

Fig. 11. Geometry and materials with the locations sources of falling rocks of the profile 1 (a) and the profile 2 (b)

3.2.3 Input Parameters The following slope material properties can be defined when using the rigid body analysis method: normal and tangential restitution coefficients, dynamic and rolling friction, forest/vegetation damping, slope roughness, and advanced friction parameters.

242

N. Q. Tuan

The restitution parameters are the most crucial parameters that influence the result of rockfall analysis. The tangential damping was considered as in CRSP (Colorado Rock Fall Simulation Program) in the analysis. The restitution coefficients are selected according to the characters of rock material and the material of terrain surfaces. These parameters are also selected according to the analysis method. In this model, there are four different types of materials, including hard rock at bench face, the soil at the natural slope, hard rock at bench floor covered by a thin layer of debris, and talus (debris material) that accumulates at the toe of the bench. It must be noted that the coefficients of restitution are not always easy to determine, and the coefficients of normal restitution for rigid body model are significantly lower than the corresponding values for lumped mass model [16]. Unfortunately, Rocscience does not have a comprehensive list of recommended values of these coefficients for rigid body analysis. Moreover, the recommended values of normal restitution coefficients given in RocFall are for 3D simulation. Therefore, this study used these parameters based on different references [17, 18]. To consider the influence of moving velocity of rock block on collision behavior, i.e., the normal restitution coefficient is not independent of velocity. The normal restitution coefficient a transition from nearly elastic conditions at low velocities to highly inelastic conditions caused by increased fracturing of the rock and cratering of the slope surface at higher impact velocities. The scaling factor K was used as recommended by Pfeiffer and Bowen [19]. This factor adjusts for the decrease in the normal coefficient of restitution as the impact velocity increases. RN (scaled ) = RN ∗ scaling factor scaling factor =

1+



1 Vrock K

2

(1) (2)

where K = velocity at which scaling factor = 0.5 and Vrock = velocity of the rock, immediately before impact, measured normal to surface. The dynamic friction and rolling resistance were taken from the table provided by the Rocscience [2]. The natural slope which is composed of weather soil was considered as medium soft material. The bench face of fresh sandstone was considered as hard material. The bench floor which is composed of sandstone covered by the thin layer of debris was considered as medium hard material. The dynamic friction coefficient was selected based on the values recommended by RocFall those are value proposed by Bar, Nicoll and Pothitos [16]. The values of input parameters were selected, as shown in Table 3.

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

243

Table 3. Parameters used in rockfall analysis Parameters

Natural slope (weathered soil)

Hard rock bench face (sandstone)

Bench floor/ Haul road (sandstone)

Talus (debris)

Normal restitution (Rn )

0.15 ± 0.02

0.45 ± 0.04

0.40 ± 0.04

0.32 ± 0.04

Tangential restitution (Rt )

0.5 ± 0.04

0.85 ± 0.04

0.9 ± 0.04

0,8 ± 0.04

Dynamic Friction

0.30 ± 0.04

0.55 ± 0.04

0.4 ± 0.04

0.4 ± 0.04

Rolling resistance

0.15 ± 0.02

0.15 ± 0.02

0.15 ± 0.02

0.15 ± 0.02

Fig. 12. Different shapes of rock boulders used in analysis

4 Results and Discussions From the results of analysis, it has been found that the motion of falling rock block is observed to be the slide, roll, fall and bounce. The results of rocfall analysis were interpreted in terms of trajectory, end location, bounce height, translational velocity and kinetic energy. The falling trajectories of rock for two profiles are shown in Fig. 13. Figure 14 shows the different results of rockfall analysis for the profile 1. The out-run of rock can reach beyond the edge of bench floor. It indicates that the rock blocks can move to the lake and the whole bench floor is under the risk of rockfall. There are some of rock blocks that stop near the crest of the bench and do not fall into the lake. The main amount of rocks would stop at the middle of the bench floor. They concentrate at the tapered edge of the talus. The end locations from calculation distribute relatively close to the real locations of fallen rocks at site. The bounce height can be up to 17.0 m above the slope surface and it might occur when a rock hit the bench crest. After reaching the bench floor, the bounce height decrease remarkably to the height of less than 1 m above the bench floor. The translational velocity may be up to about 18.9 m/s. The maximum kinetic energy may be up to 612 kJ at the location near the bench toe. According to the

244

N. Q. Tuan

Fig. 13. Rockfall trajectories with variable shapes and sizes for the profiles

Swiss Federal Guidelines [20], this value implicates high rockfall intensity as well as high risk to people.

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

245

Fig. 14. The distribution of rock end locations, bounce height, total kinetic energy along the profile 1

Figure 15 shows the different results of rockfall analysis for the profile 2. As the results about the end path distribution, a large number of rock boulder stop right at the source. It can be explained that due to the shape of rock block, sliding friction keep the rock stable source location. The out-run of rock can reach beyond the edge of bench floor. It indicates that the falling rock can move to the lake and the whole bench floor is under the risk of rockfall. There are also some of rock blocks stop near the crest of bench and do not fall into the lake. The main amount of rocks would stop at the middle of the bench floor and concentrate at the tapered edge of the talus. The calculated distribution of end location is relatively close to the real distribution of fallen rocks at site. The maximum bounce height is 10.8 m above slope and it might occur when rock hit the bench crest. After reaching the bench floor, the bounce height decrease remarkably to

246

N. Q. Tuan

Fig. 15. The distribution of rock end locations, bounce height, total kinetic energy along the profile 2

the height of less than 1m above the bench floor. The translational velocity may be up to about 6.0 m/s. The maximum kinetic energy may reach up to 389 kJ at the location near the bench toe. This value shows lower kinetic energy in comparison to rockfall energy from the profile 1. However, it still implicates high rockfall intensity level. This means the whole flat area at this local location is under danger. In both two profiles, the end locations of rock blocks tend to concentrate at the middle of the bench floor, and at the edge of the talus. Steeper cliff result in a fall movement and generally bring about shorter runout distance. It was also found that the angularity of rock block has remarkable on the travelling trajectories. Rounded block results in longer trajectory than angular block.

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

247

The size of falling block is the main factor that affect the kinetic energy. Investigation was performed for different shapes of rock. Trial simulations were carried out for every shape, i.e. triangular, square, and pentagon and rectangular (2:3) meanwhile other for input parameters were kept the same. Trials were done with the rock mass of 3000 ± 500 kg. The summarized results of trials are tabulated in Table 4. It was found that four shapes result in similar end locations, translational velocities, and maximum bounce height. The polygon with more angles tends to travel longer. In both two profiles, rounded small boulders results in longest travelling distance. Many of them would travel beyond the area of bench floor. Most of rock block with polygon shape would stop on the bench floor. Table 4. Results of rockfall analysis for four different shapes of rock block Shape

End point location of highest distribution (m)

Farthest end location (m)

Translational velocity (m/s)

Maximum bounce height (m)

Maximum total kinematic energy (kJ)

Triangle

18.0

19.8

18.9

17.8

502.3

Square

19.8

21.6

18.9

17.7

626.0

Pentagon

19.8

25.9

18.6

17.6

540.2

Rectangle (2:3)

18.9

20.6

18.5

17.9

695.5

5 Concluding Remarks The study was carried out to assess the rock slope stability and the rockfall hazards in the study area at the foothill of Chau Thoi. Two modes of rock failures were identified from the kinematic analysis and those would be the sources of rockfall. The most common mode of failure is wedge sliding between joint planes. The joint opening in the rock mass at the rock bench makes this area more susceptible to rockfalls. Some joints near the bench crest keep developed due to the tree roots and tend to make the rock blocks to fall. Some random isolated boulders lying on the terrain surface are also the source of rockfall. There exists the hazard of rockfall. The events were reproduced by the rockfall simulation. Rockfall analyses performed in the study show that the whole flat area of the bench floor at Chau Thoi foothill is prone to rockfall hazard. A number of rockfall events have already occurred. The existences of the talus debris at the bench toe, the fallen boulders, and the rock blocks in the study area are the proof of rockfall event in the past. The rockfall was performed using rigid body model considering the effect of rock shape and size. The restitution coefficients were just selected based on the experience

248

N. Q. Tuan

recommended by Rocscience and other references. Therefore, further study considering the effect of restitution coefficient should be performed. For further study, detailed investigations about rock shape, rock size and the distribution of fallen rocks should be made to calibrate the rockfall model. The results of rockfall end locations from the analysis are similar to the location of falling block observed at the field. This confirms the appropriate values of restitution coefficients. The blocks have medium to high energy and velocity to travel very far from the slope face and consequently, there is a high probability to create extensive risks to commuters if they fall. Due to the rockfall problem as the result of this study, this area should be protected or suitable measures are needed to protect the slope against rockfall. The damage capacity was classified as high for this site. Following solutions for rockfall problem at this study area should be taken: – – – –

Removal of small unstable block in the cut slope and free boulders on natural slope; Rock bolting for large block; Using protection net for the area of highly jointed rock; Keeping the talus layer at the toe of the slope to absorb the kinetic energy of falling rock.

Acknowledgments. The author thanks Rocscience Inc. for providing the use of DIPS and RocFall software. Thanks also to my cousin, Mr. Nguyen Manh Hung, and my colleague, Dr. Nguyen Trung Kien, for helping me to take the newest photos at the site of study area. The author would like to thank the reviewers for their very valuable corrections and comments.

References 1. Cruden, D.: Cruden,D.M.,Varnes, D.J.,1996, Landslide Types and Processes, Transportation Research Board, U.S. National Academy of Sciences, Special Report, 247: 36–75. Special Report - National Research Council, Transportation Research Board 247, 36–57 (1996) 2. RocScience: Rocfall v8.0 ‘Computer program for risk analysis of falling rocks on steep slopes’. RocScience Inc. (2020) 3. Ha,i, H.Q.: Chau Thoi Mountain – Geostratigraphical and Topographical Relics - View point of three cities. (2018) 4. Ha.nh, H.T.H., Tú, T.A.: A Proposal to enhance the effect for post-ming land use of Tan Dong Hiep, Nui Nho and Binh Thung quarries in Di An, Binh Duong province. Science & Technology Development 13, 84–93 (2010) 5. C`âu, D.V., Phan, Ð.N., Ky,, H.N., Thu,y, L.M., Hoa, N.N., Quang, N.V., Ðu,o.´ng, T.C.: Geological and Mineral Resources Map of Vietnam. In: Tri., T.V. (ed.) C-48-XI. Department of Geology and Minerals of Vietnam, Hanoi (1995) 6. M˜y, B.P., Khúc, V., Ðu,o.´ng, T.C., Co., M.C.: Discussion about Chau Thoi formation. Vietnam Journal of Geology 5–6, (1994) 7. Barton, N.: A relationship between joint roughness and joint shear strength. Proc. Int. Symp. on Rock Mechanics 1–8 (1971) 8. Hoek, E., Bray, J.D.: Rock slope engineering. CRC Press (1981)

Analysis of Rock Slope Failure and Rockfall for Preliminary Hazard Assessment

249

9. Dorren, L.K.A.: A review of rockfall mechanics and modelling approaches. Progress in Physical Geography: Earth and Environment 27, 69–87 (2003) 10. Frattini, P., Crosta, G.B., Agliardi, F.: Rockfall characterization and modeling. In: Stead, D., Clague, J.J. (eds.) Landslides: Types, Mechanisms and Modeling, pp. 267–281. Cambridge University Press, Cambridge (2012) 11. Volkwein, A., Schellenberg, K., Labiouse, V., Agliardi, F., Berger, F., Bourrier, F., Dorren, L.K.A., Gerber, W., Jaboyedoff, M.: Rockfall characterisation and structural protection – a review. Nat. Hazards Earth Syst. Sci. 11, 2617–2651 (2011) 12. Leine, R.I., Schweizer, A., Christen, M., Glover, J., Bartelt, P., Gerber, W.: Simulation of rockfall trajectories with consideration of rock shape. Multibody Sys.Dyn. 32, 241–271 (2014) 13. Basson, F.R.P.: Rigid Body Dynamics For Rock Fall Trajectory Simulation. 46th U.S. Rock Mechanics/Geomechanics Symposium, p. 7. American Rock Mechanics Association, Chicago, Illinois (2012) 14. Ashayer, P.: Application of Rigid Body Impact Mechanics and Discrete Element Modeling to Rockfall Simulation. Library and Archives Canada = Bibliothèque et Archives Canada (2007) 15. Dadashzadeh, N., Duzgun, S., Yesiloglu-Gultekin, N., Bilgin, A.: Comparison of Lumped Mass and Rigid Body Rockfall Simulation Models for the Mardin Castle, Turkey (2014) 16. Bar, N., Nicoll, S., Pothitos, F.: Rock fall trajectory field testing, model simulations and considerations for steep slope design in hard rock. In: Dight, P.M. (ed.) Proceedings of the First Asia Pacific Slope Stability in Mining Conference, pp. 457–466. Australian Centre for Geomechanics, Brisbane (2016) 17. Verma, A.K., Sardana, S., Sharma, P., Dinpuia, L., Singh, T.N.: Investigation of rockfallprone road cut slope near Lengpui Airport, Mizoram, India. Journal of Rock Mechanics and Geotechnical Engineering 11, 146–158 (2019) 18. Nagendran, S.K., Ismail, M.A.M.: Analysis of Rockfall Hazards Based on the Effect of Rock Size and Shape. International Journal of Civil Engineering 17, 1919–1929 (2019) 19. Pfeiffer, T.J., Bowen, T.D.: Computer Simulation of Rockfalls. Bulletin of the Association of Engineering Geologists XXVI, 135–146 (1989 ) 20. OFAT, OFEE, FEFP (eds.): Recommandations 1997 –Prise en compte des dangers dus aux mouvements de terrain dans le cadre des activit´es de l’am´enagement du territoire (1997)

A Review of Soil Improvement Methods for Tunneling Projects in Urban Areas and Their Application at the Hochiminh Metroline No. 1,Vietnam Minh Ngan Vu1 , Phuc Lam Dao2(B) , Vu Nam Chien Nguyen3 , and Duc Thinh Ta1 1 Hanoi University of Mining and Geology, Hanoi, Vietnam

{vuminhngan,taducthinh}@humg.edu.vn 2 University of Transport Technology, Hanoi, Vietnam

[email protected] 3 Vietnam Japan University, Hanoi, Vietnam

[email protected]

Abstract. When tunneling in cities with soft soil conditions, it can lead to unexpected impacts on existing buildings on the surface. Buildings locating on influence zones induced by tunneling might be damaged in the case of no mitigating method applied. The scope of the influenced zone when tunneling is represented in this paper in association with a review on recent soil improvement methods for reducing the effects of tunneling. Possible mitigating solutions applied in tunneling design and construction processes are discussed, including improving surrounding soil and/or compensating settlements, as well as protecting existing building solutions. A soil improvement application with jet grouting technique in a recent tunneling project in Hochiminh city for reducing effects of tunneling on surround buildings and the tunneling process is also analyzed. Keywords: Tunneling · Influence zones · Mitigating methods · Ho chi minh city · Vietnam

1 Introduction In recent decades, transportation space has been in high pressure due to economic development and population growth. Since the surface space becomes more and more expensive and restricted, underground space development has seemed like a vital solution in cities. In the underground construction, tunneling with TBMs (tunnel boring machines) has been popular in urban areas with advantages of safe and rapid construction and reduction of influences on existing buildings [21]. Especially when tunneling in soft soil conditions with shallow depths, it has to face difficulties of sensitive and complex foundations and urban utility systems. In cases that nearby buildings are forecast with large settlements, additional methods should be applied in order to ensure safety. Finding an appropriate solution in tunneling © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 250–269, 2021. https://doi.org/10.1007/978-3-030-60269-7_13

A Review of Soil Improvement Methods for Tunneling Projects

251

is actually necessary because this work affects safety when tunneling, construction cost, and the stability of surrounding areas. A summary of soil improvement methods applying in tunneling projects in this paper will help engineers to obtain an overview of mitigating methods and to have an ability to decide a sufficient solution in both technical and economic factors. Recent case studies of applying jet grouting technology for protecting a historical building and starting/arriving TBM areas in the tunneling process in a tunneling project in Hochiminh city, Vietnam are also introduced in this paper.

2 A Review of Soil Improvement Methods in Tunneling 2.1 The Scope of Influence Zones Induced by Tunnelling Assessing impacts of tunnel construction on existing structures is essential in the tunneling design. The stability of nearby buildings is often assessed by the ground movements around tunneling, the surface settlement trough, and the distance between the tunnel and the buildings [24]. Thus, the influence zone induced by tunneling should be determined in order to minimize the effects on the existing structures. The assessment of the impacts of tunneling excavations on existing buildings and the responses of buildings have been studied by authors all over the world including Rankin (1988) [20]; Boscardin and Cording (1989) [1]; Mair et al. (1996) [13]; Burland et al. (2001) [2]; Franzius (2004) [7]; Netzel (2009) [16] and Giardina (2013)[8]. In an investigation of ground movements induced by shallow tunneling, Vu et al. (2015) [21] proposed a model for predicting effects of tunneling on existing surface structures, as shown in Fig. 1. Based on the model, the minimum distance x from the building to tunnel axis is estimated by an allowable settlement umax and the volume loss VL as follows:

Fig. 1. Model of assessing the influence of tunnelling on existing buildings [21].

252

M. N. Vu et al.

Fig. 2. Scopes of influence zones induced by tunnelling with various tunnel diameters and case studies [24]

 x=



umax −2i2 ln Sv,max



   √  umax i4 2  2 = −2i ln √ VL D2 π

√ ωmax i3 2   x= 2 i4 √ 32ωmax 2 π VL D LambertW − π V 2 D4

(1)

(2)

L

where, the allowable settlement umax and allowable slope ωmax are indicated in technical standards [17] and the volume loss VL value depending on the tunnelling process can be obtained from [23]. Figure 2 shows that the scope of influence zones caused by tunnelling with various tunnel diameters. In this figure, the “safe” zone means that if the building location (x/D ratio) and the tunnel depth (C/D ratio) are in this zone, the building can be in safe when tunnelling. If the building location (x/D ratio) and the tunnel depth (C/D ratio) are in the “care” zone, the building can be protected without mitigating methods but the tunnelling process should be in a careful monitoring. The “additional” zone means that mitigating solution should be applied if the building location (x/D) and the tunnel depth (C/D) are estimated in this zone. On the basis of these analyses, engineers can predict and assess damage risks of existing buildings in the tunnelling process based on their locations and

A Review of Soil Improvement Methods for Tunneling Projects

253

(a)

(b)

(c)

Fig. 3. Effects of soil properties on the scope of influence zone [25] (a)Effects of the cohesion c; (b) Effects of friction angle ϕ; (c) Effects of the modulus of elasticity E.

254

M. N. Vu et al.

geo-conditions and decide to have a suitable monitoring and/or apply soil improvement methods. In a study on shallow tunneling, Vu (2016) [25] also investigated the effect of soil parameters on the scope of the influence zone. Figure 3 shows the effects of the cohesion c, the Angle of internal friction ϕ and the values of elasticity modulus E on the scope of the influence zone. It can be seen in the Figs. 3a and 3b that when increasing the cohesion c and the friction angle ϕ, the relative distance x/D reduces. This means that tunneling impacts on surrounding structures can be minimized by changing the cohesion c and the friction angle ϕ. Meanwhile, the scope of influence zone seems no change when increasing the modulus of elasticity E values. Based on these analyses, it shows that the influence scope of tunnelling can be controlled by soil improvement methods. 2.2 Settlement Compensation by Changing Soil Properties The analysis of the impact of soil parameters on the extent of zones influenced by tunneling with various relative distance x/D in Vu et al. (2017) [25] shows that with a given distance from the tunnel axis to an existing nearby building, the settlement can be achieved less than a given allowable settlement by changing parameters of the surrounding soil. On the basis of the result, following ground improvement methods with the aim of improving the soil properties can be applied in practice. Permeation grouting is the oldest grouting technique. The first application was in 1802 [26]. The principle of this method is filling voids in soil with an injection grout without changing the soil structure. The grout is pumped into a high permeable, granular soil to saturate and cement soil particles together in order to archive a stabilized soil zone for tunneling. In this technique, the grout can be pumped from the surface and/or from the tunnel section itself, ahead of the excavation face or from dedicated grouting/pilot galleries by sleeved pipes (tube à manchette, or TAM). In injection, the coarse injection grout should be used first, and then the fine injection grout. This technique with the TAM can inject different grouts in the same hole at different times (Fig. 4). The pressure used in this technique must not exceed the value h where h is the overburden pressure and αis an empirical factor with the value h = 0.3 ÷ 3 depending on soils ([11]). Permeation grouting technique is suitable for sands and gravels. In tunneling, permeation grouting has been applied in many projects, such as Turin Railway Interchange, Roma, and Napoli metro projects. Jet Grouting. In this technique, water or grout is injected with high pressures in order to disrupt the ground for improving [10]. The first application of this technique was in England in the 1950s, but the first real practical application was in Japan. In the early 1970s, rotating jet grouting developed in Japan in the case of various thickness and somewhat fragile strength. In the middle 1970s, jet grouting was introduced in Europe and has become popular [15]. Jet grouting can be used to reinforce almost all soil types, except for peat. In the procedure of jet grouting, firstly, a jet tube is injected into the soil by using a boring machine, then, the grout mix is injected with the sufficient pressure in order to erode and mix with the soil. There are three installation methods of jet grouting depending on geometry, as can be seen in Fig. 3: the single system injects only

A Review of Soil Improvement Methods for Tunneling Projects

255

grout, the double system injects grout combined with air, the triple system includes three components: grout, jetting water, and compressed air. Jet grouting has been used in many projects, for example, Galleria Valsesia Milan, Turin railway junction, Panel Grouting building pit Binnenrotte, Aechertunnel, Hochiminh Metro Line 1. Soil mixing measures are based on turning an auger into and out of the soil, while continuously adding injection fluid under pressure through the hollow core of the soil to make the soil-concrete mixture. There are three different techniques often applied: Soil Mixed Wall (SMW), Deep Soil Mixing (DSM), and Shallow Soil Mixing (SSM). This technique is normally applied for improving bearing capacity, decreasing settlement, and increasing stability for structures and embankment. For example, the railway and road embankment in Malaysia, Japan, and Sweden. This technique also applied for improving the bearing capacity of the foundation for high buildings, and highway-bridge in Poland, as well as excavation control in Japan. Ground freezing is the technique to make the soil impermeable and increase the stiffness of the soil by freezing the soil for stability. This technique can be applied for a wide range of soil types, especially fully saturated soils and in difficult ground conditions. The advantage of this technique is the ability to control the geometry of ground improvement zones by using flexible angles and length of freezing pipes. However, ground freezing requires refrigeration of massive soil volumes over a long time, so this technique is expensive comparing to other methods. When using liquid nitrogen in order to save time, the cost much increases. The other disadvantage is the expansion volume of frozen water, which can lead to unexpected heave. Therefore, it requires careful monitoring in the ground freezing process. In tunneling, ground freezing has been applied for some projects such as Copenhagen Metro, Denmark (Fig. 5).

Fig. 4. Principle of permeation grouting

256

M. N. Vu et al.

(a)

(b) Fig. 5. Jet grouting technique in tunnelling (a) Systems in jet grouting [15] and (b) Jet grouting arch umbrella in Aeschertunnel, Switzerland

2.3 Reducing Settlement Without Changing Soil Properties Compensation grouting or fracture grouting is often used for decreasing building settlements and distortions to allowed values, which are indicated in [22] or eliminating previous settlements of structures induced by tunnelling. In this method, a grout slurry is injected into the soil between building foundations and the tunnel lining by sleeve pipes (normally, TAMs), which are often installed with a drill dig (Fig. 6). In this method, the control of grouting operations works on ground and structure movements. When

A Review of Soil Improvement Methods for Tunneling Projects

257

fracturing under the foundation, the monitoring should be accurately controlled both for the settlements of the buildings and the injection performance. Based on hydraulic fracturing theory, this technique can be performed in any soil type.

Fig. 6. Compensation grouting in tunneling [12]

A large number of applications show that compensation grouting can be applied in challenging soil conditions such as soft and organic soils and peat. Compensation grouting design requires one to identify geotechnical conditions, expected heaves or settlements, and possible injection points. In tunneling, compensation grouting was used firstly in 1974 after the collapse of 23 m railway tunnel below a building in Canterbury in England in order to prevent further settlements. In the construction of the Bolton Subway Tunnel in Baltimore, the United States of America (USA), from 1977 to 1980, compensation grouting was used to prevent the settlement during tunneling. This method has been successfully applied in tunneling in Waterloo Stations, London [9], Antwerp Central Station [6], Jubilee Line Extension, London, and the North-South Line in Amsterdam. Compaction grouting is a technique that the soil is compressed by the grout around the injection point. The grout does not fill the soil pores but remains as a mass to compact the soil around (Fig. 7). In tunneling, the purpose of this technique is to compensate for previous settlement induced by tunneling by increasing the soil density and stress in the soils to heave the structures. Compaction grouting can be used for compensating the settlement of consolidation or relaxation induced by tunneling. In these cases, this technique is applied behind the TBM, from the analysis in Vu et al. (2016) [25]. Compaction grouting was firstly used in the early 1950s and then has become widely used in construction as an improvement technique in USA. In 1990, compaction grouting was exported to Japan and used extensively to repair structures that experienced settlement and tilting due to earthquakes. Although this technique has been used for more than 6 decades, there has been little research on the fundamental theory. Only some successful observation results have been published such as the drain tunnels in Phoenix and Bolton [4]. Micropiles. A micropile system is often used for transferring the structural load to competent bearing strata. Micropiles were introduced in Italy in the 1950s in the renovation of historic buildings that had been damaged during World War II. Then, this technique became popular in Europe, especially in the 1980s. Micropiles have been used with

258

M. N. Vu et al.

Fig. 7. Principle of compaction grouting ([12])

Fig. 8. Micropile construction

drilling rigs, grout mixing, and a pump for jetting the grout. The method of installing micropiles can be seen in Fig. 8. Firstly, a drilling rig drills a hole to designed depths. Then, reinforcements are placed in the hole. The grout is injected to the hole by pumping. The pile can be injected with further grout under high pressure to create a larger bearing capacity at the lower part of the pile. In tunneling, this technique can be used for reinforcing foundations above the tunnel. Cut–off wall technique uses a wall in the distance between the buildings and the tunnel in order to minimize the ground movement induced by tunneling, which leads to the settlement of nearby existing buildings, as can be seen in Fig. 9. The cut-off wall also reduces the change of groundwater when tunneling below the water table. The cut-off wall can be formed by steel sheet-piling, slurry trench walls, concrete diaphragm walls, bored pile walls, grout barriers, mix-in-place barriers or artificial ground freezing.

A Review of Soil Improvement Methods for Tunneling Projects

259

Fig. 9. Principle of the cut-off wall [3]

2.4 Mitigating Measure Selection for Tunneling Projects In tunnel design and construction, the choice of mitigating measures often depends on the cost of projects, the speed of carrying out the work, the reduction of uncertainties between design and construction, and the safety when tunneling. The selection of soil improvement methods in tunneling projects is a summary of the assessment of ground improvement on the flexibility, feasibility, durability, and the speed of carrying out of work [18]. In the case of tunneling in peat and soft clay, as investigated in [23], the volume loss of the tunneling face, along with the shield, at the tail and in consolidation might be very large. The ground improvement methods combined with reinforcement methods for the tunneling face are recommended to be applied in these cases. Careful control when tunnelling is also recommended in these cases. The mitigating measures for improving the soil properties are often applied before tunneling, and the injected grout quantity can be estimated in the laboratory in order to achieve the required soil parameters before actually being applied in projects. Meanwhile, the measures of compensating settlement without changing the soil properties are usually applied to compensate for the settlement induced by tunneling. The cavity expansion methods indicated in [27] and [23] can be used to estimate the quantity of required grout in these measures.

3 A Case Study at Hochiminh Metroline No. 1, Vietnam The Metro Line 1 in Hochiminh city is the state-of-the-art metro line built in Vietnam with 19.7 km length comprising 2.6 km underground under density areas of Ba Son shipyard, the Saigon Municipal Opera House and the Saigon river (Fig. 10). There are 14 stations along this metroline from the Ben Thanh station to the Long Binh deport.

260

M. N. Vu et al.

The Hochiminh Metro Line 1 project was started in 2012 and the underground work was finished in 2019. The tunnel is constructed under many historical buildings and density areas, thus allowable settlements and other effects on existing buildings on the surface are very strict. An Earth Pressure Balance Tunnel Boring Machine (EPB TBM) was used for tunneling from 11 to 30 m depths.

Fig. 10. Plan of Hochiminh Metro Line No1 in Vietnam

Table 1. Soil parameters at the Saigon Municipal Opera House, Hochiminh city, Vietnam Layer Thickness Unit N-value Cohesion Friction Young Poisson’s Permeability (m) weight γ c angle ϕ modulus ratio ν coefficient k (kN/m3 ) (kN/m2 ) degree αE0 kN/m2 (m/sec) Fill

1.1

18.0

1

0

28

2,500

0.35

1 × 10–6

Ac2 and Ac3

1.7

16.0

1

14

0

10,000

0.48

1 × 10–9

As1

13.9

19.5

6

0

31

16,000

0.33

5 × 10–5

As2

17.0

19.5

13

0

31

35,000

0.33

5 × 10–5

Dc

15.6

21.0

43

220

0

101,000

0.45

1 × 10–10

Ds

–

21.0

31

0

34

77,500

0.31

7 × 10–6

A Review of Soil Improvement Methods for Tunneling Projects

261

The soil-profile at the Saigon Municipal Opera House location is shown in Table 1. There are three soil materials including Fill layer at the top, Alluvium layers and Diluvium materials at lower depths. The Fill layer is of about 2 m. Alluvium layers are of about 30 m depth comprising Soft Clayey Silt (Ac2 and Ac3), Silty Fine Sand Layer 1 (As1) and Sand layer 2 (As2). Diluvium layers shown in Table 1 include Diluvium clayey silt (Dc) and silty sand layer (Ds).

the Saigon Opera House tunnels

(a)

the Saigon Opera House Tunnels

Jet grouting wall

(b) Fig. 11. Settlement analysis in Plaxis 2D for the Saigon Municipal Opera House in Hochiminh Metro Line 1 [19] (a) In the case of no protection solution (b) In the case of using jetgrouting wall

3.1 Jet-Grouting Wall for Protecting a Historical Building The Saigon Municipal Opera House is one of the most important buildings in Hochiminh city with French architectural shape like the Opera Garnier. The house was designed

262

M. N. Vu et al.

smaller than Hanoi Opera House with a specific characteristic of French Third Republic. All the design drawings, decorations and furnitures were sent from France. This house was built for a decade from 1901 to 1911 with 1800 seats. The opera house has a main seating floor with two seating levels. After some reparations, the opera house now has only a capacity of 500 seats. Since the tunneling works were carried out in a soft soil condition with a short distance between the house and the tunnel alignment, the house was predicted to have a significant influence if no protection method would be applied. Thus, finding protection methods of any damage to the house during the excavation process is necessary for the project design. Due to the historical role of the house, much effort had been carried out to study a sufficient method for protecting the house. Figure 13 shows the locations of twin tunnels at this location at levels of about −12.5 m and −24.5 m. Thus, tunnels are in the As1 and As2 layers, as can be seen in Table 1. A prediction of the house settlement carried out in Plaxis 2D was shown in Fig. 11a with a maximum settlement of 61.04 mm [19]. This means that the house would have large damage if no additional method would have been applied. In this project, a solution of using jet grouting technique was proposed. In this protection design, the tunnel alignment is mostly surrounded by jet grouting walls in both sides and above the tunnel crown, as can be seen in Fig. 13. The jet grouting wall was created by jet grouting technique in the distance between the house and the tunnel alignment for minimizing the ground displacement induced by tunneling. With this solution, the soil displacement is minimized not only in the house direction but also with the surface settlement. This wall also decreases groundwater variation under the house foundation. Therefore, this method can reduce potential damages to the building in the most safe way.

the Saigon Opera House

Jet grouting area

Fig. 12. Jet grouting plan at the Saigon Opera House

An analysis by Plaxis 2D was also carried out for this case, as can be seen in Fig. 11b. It was shown that settlements of the house are predicted from 10 to 12 mm when jet grouting wall applied with a depth of 2.7 m [19].

A Review of Soil Improvement Methods for Tunneling Projects

the Saigon Opera House

Fig. 13. Crosssection view of jet grouting wall at the Saigon Opera House location

Fig. 14. Inclinometer locations for monitoring at the Saigon Opera House

263

264

M. N. Vu et al.

Fig. 15. Observed data of IDT-03 inclinometer behind the jet grouting wall near the Saigon Opera House (in mm)

A layout of jet grouting work at the Opera House site is shown in Fig. 12. A total of 239 jet grouting holes comprising 66 jet grouting holes (1400 mm in diameter), 47 jet grouting holes (3000 mm in diameter), and 126 jet grouting holes (3500 mm in diameter) was constructed at the field. On-site, the grout was continuously mixed by a mixing machine with the combination per cubic meter comprising 760 kg PCB40 cement plus 750 water liters. This slurry combination was pumped with a pressure of about 40 MPa and 300 L/min discharge rate. The rotational speeds of the triple rod were 12 rpm for jet grouting holes with 3500 mm in diameter, and 14 rpm for jet grouting holes with 3000 mm in diameter. The pulled up speeds of the rod were 9 min/m for 3500 mm jet grouting hole and 7 min/m for 3000 mm jet grouting hole. The depth of the jet grouting wall is at the level of − 25 m with a width of 2.7 m (see Fig. 13). Four inclinometers were installed at the site for measuring ground displacements in the space between the jet grouting wall and the Saigon Municipal Opera House as shown in Fig. 14. The observed data of soil displacements at the location of IDT-03 is presented in Fig. 15. The role of the jet grouting wall is clearly shown in this case. The soil displacements have fluctuated with the maximum value of 15 mm is at the top of the wall, near the

A Review of Soil Improvement Methods for Tunneling Projects

265

surface in the “A” direction, where no jet grouting applied. The maximum displacement along the part of the wall without jet grouting is only about 6 mm. Meanwhile, at the jet grouting part, the maximum displacement is only less than 4 mm from the level − 6 m and becomes smaller with deeper depths. In the “B” direction, the data show that at the zone without jet grouting (above the level of –6 m), maximum soil displacement is only about 8 mm while only small deformation (less than 4 mm) was recorded in the jet grouting zone. This means that the jetgrouting wall, in this case, protected the important building effectively from the tunneling impact. The tunnel construction in Hochiminh Metro Line 1 has already been completed. The observation of soil displacements has been continued, and no damage has been recorded at this building. These small observed data show that the cut-off wall is a sufficient way for protecting existing buildings when tunneling in soft soils, in particular, the jet-grouting wall in this case. 3.2 Strengthening the Tunneling Process When the TBM starts/arrives shafts, there is a transformation from the concrete structure to the soil environment at the boundary of shaft areas. Practically, the soil in these areas is always improved in order to avoid possible damage that appeared on surrounding structures and the TBM due to large settlements. In the case of Hochiminh Metro Line

Fig. 16. Jet grouting plan at the starting area at Ba Son Station

266

M. N. Vu et al.

1, at shaft areas, the tunneling was carried out with shallow overburdens and soft soil conditions. Thus, if high supporting pressures had been applied at the tunneling face, blow-out and fracturing accidents would have also occurred in the tunneling process. On the basis of these situations, the jet grouting method was selected to improve the soil surrounding launching/arriving shafts in this project. Figure 16 presents a jet grouting plan at the starting shaft at Ba Son station. The starting areas at both East Bound and West Bound are designed with 23 jet grouting holes with 3500mm in diameter. The slurry components per cubic meter also include 760 kg PCB40 cement plus 750 water liters. A pressure of 40 MPa was used for pumping the slurry combination with the 300 L/min of discharge rate when jet grouting at the site. The triple type rod of the jetgrouting machine was hitched with a speed of 9 min/m and rotated with a speed of 12 rpm. The soil around the arrival shaft at the Opera House Station was improved by 27 jet grouting columns with 3500 mm in diameter and 7 jet grouting columns with 3000 mm in diameter (Fig. 17). The same mixing proportion, pumping pressure and discharge rate with the jet grouting work at the starting shaft at the Ba Son station was applied. For the jet grouting columns with 3500 mm in diameter, a rotational speed of 12 rpm and a pulling up speed of 9 min/m were applied. For the jet grouting columns with 3000 mm in diameter, slower rotational and pulling up speeds of 14 rpm and 7 min/m were used. At

Fig. 17. Layout of soil improvement at the arrival area by jet grouting technique at the Opera House Station

A Review of Soil Improvement Methods for Tunneling Projects

267

this location, there are 4 inclinometers (INC N01, INC NO2, INC NO3 and INC NO4) were installed at the site for monitoring soil displacements as can be seen in Fig. 17. Figure 18 shows sample test results, including unconfined compression tests of a sample from a jet grouting column after 7 days (Fig. 18a) and core samples from a jet grouting column after 14 days (Fig. 13b) of jet grouting column in this project. In detail, after 7 days, the unconfined compressive strength qu is measured of about 3000 kPa while the secant modulus at 50%q Es,50 is measured of around 870 MPa. With core samples from a jet grouting column after 14 days, the unconfined compressive strength qu is measured of about 2450 kPa and the secant modulus at 50%q Es,50 is 1758 MPa. These recorded values with jet grouting with large diameter technology used in the Hochiminh Metro Line 1 project are double than the values of normal jet grouting with smaller diameters. As above discussed, the scope of influence zones at these areas is minimized with highly improving soil properties.

Fig. 18. Results of sample tests for jet grouting columns in Metro Line No.1 Project in Hochiminh city: (a) Unconfined compressive test for a jet grouting sample after 7 days and (b) a core sample test for a jet grouting sample after 14 days

The underground construction in the Metro Line No 1 project in Hochiminh city has been finished and no problem at starting and arriving shafts of the project has been occurred. This success means that the jet grouting technique is a sufficient soil improvement method when tunneling in soft soil conditions, especially at sensitive zones of starting/arriving TBM.

4 Conclusions Population and infrastructure development in line with limited surface space lead to the high demand for underground construction. Tunneling in soft soils in urban areas has to face complex existing utility systems and geotechnical issues. The study shows that the scope of the influence zone induced by tunneling can be controlled by improving the soil

268

M. N. Vu et al.

properties. This paper reviews possible soil improvement methods applied in tunneling for ensuring the tunneling process in safe and protecting nearby buildings, including methods of changing soil properties and compensating tunneling effects. A case study of Hochiminh Metro Line 1 in this study shows that jet grouting technique applied at launching/arrival shafts for successful tunneling work and protecting the historical Saigon Municipal Opera House is a sufficient solution for future metro lines in Vietnam.

References 1. Boscardin, M.D., Cording, E.J.: Building response to excavation-induced settlement. J. Geotech. Eng. 115(1), 1–21 (1989) 2. Burland, J.: Assessment of risk of damage to buildings due to tunnelling and excavation. Imperial College of Science, Technology and Medicine (1995) 3. Burland, J.B., Standing, J.R., Jardine, F.M.: Building response to tunnelling: case studies from construction of the Jubilee Line Extension, London, vol. 200. Thomas Telford (2001) 4. Baker, W., Cording, E., MacPherson, H.: Compaction grouting to control ground movements during tunneling. Undergr. space 7(3), 205–212 (1983) 5. Chambon, P., Corté, J.F.: Shallow tunnels in cohesionless soil: stability of tunnel face. J. Geotech. Eng. 120, 1148–1165 (1994) 6. Chambosse, G., Otterbein, R.:. Central station Antwerp compensation grouting under high loaded foundations. In: Burland, Th., et al.(eds.). Proceedings of Conference on building response on tunnelling. Case studies from Construction of Jubilee Line Extension. Telford, London (2001) 7. Franzius, J.N.: Behaviour of buildings due to tunnel induced subsidence. PhD thesis, University of London (2004) 8. Giardina, G.: Modelling of settlement induced building damage. Ph.D. thesis, Delft Univ. of Technology, Delft, Netherlands (2013) 9. Harris, D., Mair, R., Love, J., Taylor, R., Henderson, T.: Observations of ground and structure movements for compensation grouting during tunnel construction at Waterloo station. Géotechnique 44(4), 691–713 (1994) 10. Huai-Na, Wu., et al.: Ground response to horizontal spoil discharge jet grouting with impacts on the existing tunnels. J. Geotech. Geoenviron. Eng. 146(7), 05020006 (2020) 11. Kolymbas, D.: Tunnelling and Tunnel Mechanics: A Rational Approach to Tunnelling. Springer, Heidelberg (2005) 12. Mair, R., Taylor, R.: Theme lecture: bored tunnelling in the urban environment. In: Fourteenth International Conference on Soil Mechanics and Foundation Engineering, Proceedings International Society for Soil Mechanics and Foundation Engineering, pp. 2353–2385 (1999) 13. Mair, R., Taylor, R., Burland, J.: Prediction of ground movements and assessment of risk of building damage due to bored tunnelling. In: Balkema, A.A. (ed.) Fourth International Symposium of International Conference of Geotechnical Aspects of on Underground Construction in Soft Ground, pp. 713–718 (1996) 14. Manassero, V., Di Salvo, G.: Two difficult tunnelling problems solved by using permeation grouting: the excavation of submerged large size tunnels for Roma and Napoli metro projects. In: 4th International Conference on Grouting and Deep Mixing (2012) 15. Moseley, M.P., Kirsch, K.: Ground improvement. CRC Press, Baco Raton (2004) 16. Netzel, H.D.: Building response due to ground movements. PhD thesis, Delft University of Technology (2009) 17. NEN-EN 1997–1, C.E.: Eurocode 7 Geotechnical design - Part 1: General rules. European Prestandard ENV, 1 (1997)

A Review of Soil Improvement Methods for Tunneling Projects

269

18. Pelizza, S., Peila, D.: Soil and rock reinforcements in tunnelling. Tunn. Undergr. Space Technol. 8(3), 357–372 (1993) 19. Liem, P.S., Hoang, N.B.: Using jet-grouting to reinforce soil mass surrounding the tunnel and protect the construction foundation in Metro Line 1 –Hochiminh city. J. Transp. Sci. Technol. 21 (2016) 20. Rankin, W.: Ground movements resulting from urban tunnelling: Predictions and effects. Geol. Soc. Lond. Eng. Geol. Spec. Publ. 5(1), 79–92 (1988) 21. Vu, M.N., Broere, W., Bosch, J.W.: Effects of cover depth on ground movements induced by shallow tunnelling. Tunn. Undergr. Space Technol. 50, 499–506 (2015a) 22. Vu, M.N., Broere, W., Bosch, J.W.: The impact of shallow cover on stability when tunnelling in soft soils. Tunn. Undergr. Space Technol. 50, 507–515 (2015b) 23. Vu, M.N., Broere, W., Bosch, J.W.: Volume loss in shallow tunnelling. Tunn. Undergr. Space Technol. 59, 77–90 (2016) 24. Vu, M.N., Broere, W., Bosch, J.W.: Investigation of influence zones induced by shallow tunnelling in soft soils. In: The 9th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground (2017) 25. Vu Minh, N.: Reducing the cover-to-diameter ratio for shallow tunnels in soft soils. PhD thesis. Delft University of Technology (2016) 26. Xanthakos, P.P., Abramson, L.W., Bruce, D.A.: Ground Control and Improvement. Wiley, Hoboken (1994) 27. Yu, H.S.: Cavity Expansion Methods in Geomechanics. Springer, Heidelberg (2013)

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture: A Case of X-well in Vietnam Hai Linh Luong(B) , Hung Van Nguyen, and Nhu Y. Ha Petrovietnam University, Ba Ria City, Vietnam [email protected]

Abstract. In order to predict the geometric development of fracture, we need to use geometry simulation models. The models commonly used in the petroleum industry are usually 2-dimensional (2D) or 3-dimensional (3D) models. The limitation of 2D models is the dependence of fracture height along fracture length. In contrast, the fracture height of the 3D models is not restricted by the thickness of the reservoir and the results of the 3D simulation are more accurate. Therefore, this topic will focus on solving the limitations of 2D model and simulating the fracture development for pseudo-3D models, applying to a case in Vietnam. The result shows that the calculation results of 2D PKN-C are the same compared with software results, so the model of the approach is reliable. Moreover, compared to the 3D software, the P3D results showed a difference, but it is minor because the fracture height in the equation is the maximum height that the fracture can achieve corresponding to the given stress value and rock properties. We also considered some factors affecting fracture height, such as stress contrast, plane strain modulus, fracture toughness, leak-off coefficient, slurry injection rate. Keywords: Hydraulic fracture · Pseudo-3D fracture geometry · Fracture height · Equilibrium fracture height model

1 Introduction Vietnam, with favorable geological structure features, has great oil and gas potential. However, more than 60% of the oil reserves remain in the reservoir due to exploitation challenges, and the oil recovery factor is low. Research by experts shows that the main cause of this phenomenon is due to the serious formation damage around the well (because of drilling and production with excessive flow rate), low conductivity (permeability), the product flowline’s consumption of a large amount of energy to flow into the well, resulting in a rapid depletion of the reservoir energy source, and consequently, the well cannot be further produced due to unprofitability. For such wells, there is a need to treat the near-wellbore area for the purpose of remove formation damage, maintain the natural permeability of the formation, increase the possibility of flow into the production well and increase the capacity of the injection © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 270–289, 2021. https://doi.org/10.1007/978-3-030-60269-7_14

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

271

well. Currently, there are many different treatment methods, with their own advantages and disadvantages, but the method of the near-wellbore treatment by hydraulic fracturing is probably used quite widely because of the short treatment time that can achieve high efficiency. However, to achieve the best result, this method requires an in-depth understanding of the connection between fractures and their geometric structure. In order to know the relationship of fracture in the formation, it is important to first know which direction the fracture develops, and what is the depth of penetration. The structure and propagation of fractures depend not only on the stress of the formation but also on the influence of many other factors in the propagating area. The presence of these factors makes the equation for calculating the fracture dimension become complicated and impossible to solve. Therefore, in the traditional design calculation methods (such as KGD, PKN models), it must be accepted to include some assumptions to simplify those equations. But these assumptions are unrealistic as we cannot find any place in the reservoir with “homogeneous and isotropic rock” or “ideal, frictionless flow” [13]. In order to predict geometric fracture development, we need to use fracture simulation models. The models commonly used in the petroleum industry are usually 2-dimensional (2D) or 3-dimensional (3D) models. The limitation of the 2D model is the dependence of fracture height on the fracture length. Notably, 2D PKN model is used when the fracture length (2xf ) is much larger compared to the fracture height (hf ); the 2D KGD model is suitable for calculating the fracture geometry where hf is much larger than 2xf . Opposed to the 2D model, the fracture height of the 3D model is not restricted by the reservoir thickness, and the results of the 3D simulation are more accurate [1, 8]. In Vietnam, there have been researches on 2D fracture development models, but there are still some unresolved matters such as inability to simulate the fracture development after well shut-in, especially propagating vertical fracture, changes in stress components around the fracture is not forecasted. Besides, there are very few researches on the 3D and pseudo-3D fracturing model, although there have been studies worldwide about these models [2, 3]. So this study will focus on: (1) building the fracture model according to the length, height, and width based on the specific characteristics, structure and permeability of Oligocene sandstone layer; (2) applying PKN-C hydraulic fracture simulation method of Perkins, Kern, Nordgren and Cater to design treatment schedules for specific objects, using software to simulate the fracturing process to compare and evaluate the study; and (3) evaluating the effects of the formation and fracturing process on fracture models in general and fracture height in particular.

2 Methodology This study uses the theory of 2D PKN-C simulation to determine the fracture size then uses the P-3D model to determine the actual fracture height, taking into account the real reservoir properties and heterogeneity. After that, it redesigns the pump schedule and other parameters of the 2D model. 2D models cannot simulate both vertical and horizontal propagation. Therefore, 3D pseudo-models were formed by eliminating the assumption of invariable and uniform fracture height. Height in pseudo-3D models is a function of position along the fracture and time. Assuming the fracture length is much greater than the fracture height and one

272

H. L. Luong et al.

significant difference between the pseudo-3D model and the 2D model is the addition of the vertical flow component. Pseudo-3D models have been applied to model fracture through many layers of rocks with different stresses and properties. These models are uncomplicated, quick, and comparably powerful [4, 6]. Similar to the PKN-C model, in cases when the crack height is small compared to its length, the P3D model can be applied. Nonetheless, in opposition to the PKN-C model, the fracture height of the P3D model is not restricted to the reservoir thickness (hf ) and grows vertically into the adjacent layer [5]. P-3D models are suggested to idealize crack development in multi-layered formations. Instead of considering the change of fracture geometry in 3-dimensional space, the pseudo-3D models modify the 2D PKN model by including fracture height variation (linear or parabolic) and its effect on the width [6]. There are 5 main factors affecting fracture height: geomechanics properties (Young’s modulus, tensile strength, Poisson ratio), layer interfaces, in-situ stress, fracture toughness and fluid leak-off coefficient [7, 9]. 2.1 Fracture Model in the Case of 3 Layers The equilibrium-height of a hydraulic fracture is the crack height when integrating net pressure inside the fracture along with the height, i.e., stress-intensity factor (Eq. 1), K I . The stress-intensity factors are equal to the fracture toughness of the materials, K IC , at the top and bottom layers. The equilibrium fracture height model (MFEH) fully describes the development of height between the layers and the different fluid properties (in-situ stress, reservoir thickness, fracture toughness, fluid density, etc.), and figure out the fracture equilibriumheight equation rapidly [17]. The equilibrium heights can then be applied to (1) give out the input data for the 2D model, (2) upgrade the 3D model, (3) specify the net pressure required to attain certain height growth, and (4) suggest maximum net pressure to ensure no fracture propagation into the aquifer or gas. Harold Liebowitz (1968) [18] summarized the theory and analysis of Griffith, Barenblatt, Irwin and the theory of linear elastic fracture mechanics (LEFM), drawing an expression to calculate the stress intensity coefficient when fracture develop and reach

Fig. 1. Parameters in calculating fracture height in dimensionless system [10]

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

273

the critical fracture toughness value, KIC (plane-strain fracture toughness), for cracks extending on the y axis as shown in Fig. 1. Figure 1 shows how the fracture develops in three layers. According to this figure, the middle layer usually has the smallest minimum principal stress (σ1 ). The two adjoining layers have larger minimum in-situ stress (σ2 , σ3 > σ1 ). The penetrations into the upper (hu) and lower (hd) layers go up as the pressure at the center of perforation rose. The requirement of equilibrium poses two constraints (stress intensities at both tips attain to fracture toughness), leading to it is possible to figure out two equations simultaneously:    1 hp 1+y ∗ dy. (1) pn (y) KI + = KICtop = π(yu − yd ) 1−y −1    1 hp 1−y KI − = KICbot = ∗ dy. (2) pn (y) π(yu − yd ) 1+y −1 where KI+ , KI− are the fracture toughness of the upper and lower layers, respectively, m/L0.5 t2 , psi-in0.5 yu yd are the dimensionless vertical positions of the top and bottom perforations, respectively. yu = yd =

−(

hp +hu +hd 2

− hu 2hu =1− . hp + hu + hd hp + hu + hd

hp +hu +hd 2

) + hd 2hd = −1 + . hp + hu + hd hp + hu + hd

(3)

(4)

where hu - fracture propagation into the upper formation, L, ft hd - fracture propagation into the lower formation, L, ft hp - thickness of the perforation interval, L, ft Finally, fracture height is calculated using the following equation: hf = hp + hu + hd

(5)

The improved model can detect and stop calculations if the peaks reach the upper or lower boundary (containing water or gas) or if the sensitivity of fracture growth is infinite with pressure. This new model can quickly and reliably calculate the theoretically maximum equilibrium fracture height in the different strata with different rock and fluid properties. 2.2 Fracture Model in the General Case of n Layers    a  a 1 a+y a+y 1 dy = √ dy Pn (y) (my + bi ) KI + = √ a−y a−y π a −a π a −a    a  a 1 a−y a+x 1 KI − = √ dy = √ dx Pn (y) (−my + bi ) a+y a−x π a −a π a −a

(6)

(7)

274

H. L. Luong et al.

Fig. 2. Hydraulic fracture in a multiple layer of rock formation [9] u where fracture half-height a = tipd −tip , tipu and tipd are the upper and lower tip depths. 2 [14–16] Hydraulic fracture in a multiple layer of rock formation is shown in Fig. 2. We solve the two nonlinear equations with the programming software (for diffu = 0 and diffd = 0) for the equilibrium tip depths tipu and tipd. The actual fracture height is then calculated:

hf = tipu + tipd.

(8)

4000000

Net pressure, N/m2

3500000 3000000 2500000 2000000 1500000 1000000 500000 0 50

60

70

80

90

100

110

120

Fracture height, m Fig. 3. Graph for calculating equilibrium height model (fracture height-dotted line, net pressuresolid line)

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

Fig. 4. Algorithm of the P3D model

275

276

H. L. Luong et al.

2.3 Combine P3D Equilibrium Fracture Height Model and PKN-C Model It is clearly that it would require the iterative solution of two Eqs. (1) and (2) (which are the function of net pressure distribution) to solve the equilibrium height equations. The graph for calculating equilibrium height model is shown in Fig. 3. The programming software is used to obtain fracture height. The net pressure can be obtained from the 2D PKN-C simulation, and when designing the fracture using this model, it is required that one of input parameters is fracture height. Therefore, the bisection algorithm is put into use to combine the fracture height calculation using equilibrium height into the fracture simulation. The algorithm of pseudo-3D model is shown in Fig. 4.

3 Result and Discussion Table 1 shows typical input data of X-well in Vietnam. In addition, we also used other detailed data, such as proppant parameters, fracture fluid, reservoir properties. Due to the requirement of increasing exploitation speed, the applied objects are tight and low permeability reservoirs. The object of this study is the lower Oligocene strata where the sign of the oil and gas is very good, the average porosity is 10–15%, the average permeability of the reservoir is poor, varying from 0.1–5 mD. Most of the cracks have very low conductivity, poor connection between fractures. The lower Oligocene-containing rocks consist mainly of medium-coarse-grained sandstone to conglomerate mixed with little siltstone and very thin layer of limestone. Research scope is X well which is located in Y field, block09-2/09, in Cuu Long Basin. 3.1 Compare Results from the 2D PKN-C Model and MFrac Software From the formulas developed for the PKN-C model combining the programming software (which calculates the fracture length and width), we calculated the final parameters of the model with the design pumping time of 115 min, determined the fracture development over time and calculated the pumping schedule. By comparing this result with the results produced from the software, we have the result Table 2 and Figures from 5, 6, 7 and 8. The fracture half-length is 232 m and the maximum fracture width is 0.01356 m. Based on the hydraulic fracture equilibrium height equation, with the pump flow rate (0.0477 m3 /s), fracture half-length, the leak-off coefficient in net pay (6.7E−04 m/s0,5 ) and the designed injection time (115 min), we calculate the parameters presented in Table 2, the calculation results are the same compared with software results, so the model of the approach is reliable. Thus we can customize input to optimize the fracture. Figures 6, 7, and 8 are graphs comparing the half-length, average width, maximum width and net pressure from the 2D PKN-C model and software. However, the 2D PKN-C model and MFrac software have limits as the fracture height is constant and equal to the open hole height, in fact this parameter increases during the fracturing process, reducing the value of the original simulation. With high rock strength and high stresses in Oligocene formation, the fracture height propagation is limited, which explains why operators often assume that the fracture height does not change in the fracturing process. However, studying fracture height is an important and necessary issue to control fracture propagation.

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

277

3.2 Compare Results from the P3D Model and MFrac Software To solve fracture equilibrium height equations, it requires repeat solutions of two Eqs. (1) and (2). In this study, programming software is mainly used to solve problems. As mentioned earlier, the range of fracture pressure (Pnet) for determining height by equilibrium height is limited. This range can be derived from the stress difference of the payzone and adjacent formations. The payzone has closure stress of 34.47 MPa and E’ of 37852.2 MPa. Both upper and lower limit layers have fracture toughness of 1000 psi-in0.5 . The Fig. 9 shows fracture height calculated by the equilibrium fracture height model. From the results of the 2D PKN-C model, we draw the value xf = 232 m and Pnmax = 10.81 MPa. Let hf = 2xf to find the lower limit of Pnet , we get Pnmin = 0.3944 MPa. Table 1. Input data for calculation of the 2D PKN-C and P3D fracture models True vertical depth, TVD, m

3499.6

Density of proppant SG (water = 1)

3.16

Proppant packed porosity, φp

0.35

Proppant packed permeability, kf md

320,500

Maximum proppant diameter, Dpmax , mm

0.66

Permeability, k md

2

Net pay thickness hn , m

23.77

Gross pay thickness (perforated interval), hp , m

23.77

Well radius, rw m

0.1

Drainage radius, re m

396.24

Skin before stimulation, s

0.0

Fracture height hf , m

23.77

Plane strain modulus E’, MPa

37852.2

Slurry rate (2 wing, liq+ prop), qi m3 /s Consistency Index, K’ (lbf/ft2) * sn

0.0477

Flow behavior index, n’

0.3289

Leak-off coefficient in net pay, CL m/min 0.5

6.7E−04

Spurt loss, Sp, m3 /m2

0.00000

0.1774

pc kg/m3

958.6

Closure stress, MPa

34.47

Fracture Toughness, KICtop = KICbot , MPa.m0.5 0.034 σ1, MPa 63.902 σ2 , MPa

68.83

σ3 , MPa

70.67

278

H. L. Luong et al. Table 2. Final results from the 2D PKN-C model and MFrac software Parameters

PKN-C MFrac Software

Fracture half length, xf , m

232

236.93

Average fracture width, wavg , m

0.0085

0.00864

Maximum fracture width, wmax , m

0.01356 0.011

Total pumping time, te , min

115

115

Pump efficiency, η, %

28.572

20.422

Net pressure, MPa

10.81

8.804

Fracture volume, m3

94.03

97.292

Total volume of fracturing fluid, m3 329.1 Pad pumping volume, m3 182.84

188.8

Pad pumping time, min

63.889

65.973

Open hole height, m

23,77

23.8

Upper Frac Height, m

11.89

11.9

Lower Frac Height, m

11.89

11.9

330.2

250

xf, m

200 150 2D-PKNC Software

100 50 0 0

20

40

60

80

100

120

140

time, min Fig. 5. Comparison of haft length calculated from the 2D PKN-C model and MFrac software

By using bisection solution for Pnet which varies from 0.345 MPa to 10.34 MPa, we obtained the change value table of hf according to Pnet (Table 3). By comparing the results produced from P3D model with the results produced from the MFrac software, we have the result Table 4 and Figures from 10, 11, 12 and 13. When considering the value of Ptreat and extending the thickness of the top and bottom layers to a sufficiently considerable value of the input data, the result is a full height map (Fig. 14). The calculation stopped because the top of the fracture was unstable due to increased pressure. This contributes to the limitation of fracture treatment pressure

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

279

wavg, m

0.01 0.008 0.006 2D-PKNC

0.004

Software 0.002 0 0

20

40

60

80

100

120

140

time, min

wmax, m

Fig. 6. Comparison of average width calculated from the 2D PKN-C model and MFrac software

0.016 0.014 0.012 0.01 0.008 0.006 0.004 0.002 0

2D-PKNC Software

0

20

40

60

80

time, min

100

120

140

Fig. 7. Comparison of maximum width calculated from the 2D PKN-C model and MFrac software

to control the growth of fracture height. In this case, maximum Ptreat should not exceed 69.637 MPa. Compared to the 3D software, the results showed a difference, but it is minor, because the fracture height in the equation is the maximum height that the fracture can achieve corresponding to the given stress value and rock properties. Therefore it predicts the limited perspectives of the fracture model, while the software models fully three-dimension fracture so the result will be lower than the P3D model. A full heightmap with the obtained upper and lower layer thickness shows that the final trend of the fracture will develop to infinity and offers the maximum treatment pressure limit is used. The model will find and stop the calculation, if the fractured top reaches the upper or lower boundary or if one end of the fracture begins to grow to infinite and instability. Finally, the calculation of height development is important in preventing it from reaching undesirable gas or water formation [9]. In summary, in the process of fracturing, it is necessary to study geology, characteristics of the fracturing area and the selection of the proppant, the fracturing fluid to

280

H. L. Luong et al. 12

Net pressure, MPa

10 8 6

2D-PKNC Software

4 2 0 0

20

40

60

80

100

time, min

120

140

Fig. 8. Comparison of net pressure calculated from the 2D PKN-C model and MFrac software

Table 3. Equilibrium height values according to net pressure Pnet_P3D

dhu

dhd

hf

xf

wavg

wmax

Pnet_PKN

MPa

m

m

m

m

m

m

MPa

0.4826

0.0372

0.0427

23.8545

231.2540

0.1021

0.1628

10.7648

1.0342

0.3950

0.2762

24.4459

225.8202

0.1021

0.1625

10.4755

2.0684

2.3226

1.3768

27.4744

201.6054

0.1006

0.1603

9.1986

2.7579

7.1495

3.8573

34.7815

160.3203

0.0978

0.1561

7.0749

3.4474

10.0637

5.2344

39.0728

143.1773

0.0966

0.1539

6.2155

4.1368

20.3953

9.6668

53.8366

104.8311

0.0930

0.1484

4.3499

4.8263

45.4618

19.5614

88.7982

64.4041

0.0881

0.1402

2.4913

5.1711

69.7095

30.9056

124.3898

46.3731

0.0847

0.1347

1.7113

5.8605

147.0568

82.3208

253.1523

23.1828

0.0780

0.1244

0.7751

6.2053

190.5633

76.5777

290.9159

20.2393

0.0768

0.1225

0.6638

6.5500

234.7961

71.4701

330.0408

17.8920

0.0759

0.1207

0.5767

ensure controlling the development of fractures, and setting limits on treatment pressure or the maximum proppant [11]. As previously mentioned, the equilibrium height model is only valid within the limited range of net pressure that can be obtained from stress contrast. Thus, the stress difference of target layers and adjacent formations is too high or too low, so that the equilibrium height cannot be calculated. Therefore, in case the height of the initial equilibrium height cannot be applied, it is necessary to provide assumptions to obtain fracture height values in all cases.

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

281

Table 4. Final results from the P3D model and software Parameters

P3D

MFrac Software

Fracture half length, xf, m

102.9191 132.3976

Average fracture width, wavg, m

0.0078

0.0074

Maximum fracture width, wmax, m

0.0123

0.0109

Total pumping time, te, min

115

115

Pump efficiency, η, %

26.607

19.22

Net pressure, Mpa

4.2593

5.5454

Fracture volume, m3

87.564

97.292

Total volume of fracturing fluid, mˆ3 329.1038 330.1826 Pad pumping volume, mˆ3

190.7810 205.8810

Pad pumping time, min

66.665

71.942

Open hole height, m

54.8647

51.4996

Upper Frac Height, m

32.7664

28.3309

Lower Frac Height, m

22.0983

23.1678

12000000

Pnet, N/m2

10000000 2D PKN-C

8000000

P3D Equivalent height

6000000 4000000 2000000 0 0

50

100

150

hf, m

200

250

300

350

Fig. 9. Fracture height calculated by the equilibrium fracture height model

3.3 Analysis Factors Affecting Fracture Height Effect of Stress Contrast (σu ), (σd ). We consider three cases with the change in a stress difference between layers and keep the remaining elements. Figure 15 shows the different intersections from the stress differences between layers. The dotted line characterizes the net pressure coming from a range of heights using the 2D PKN-C model. The lines with triangle, square and diamond shapes show fracture height at different net pressures by equilibrium height calculations. The trend of the equilibrium height curves of each stress difference between layers shows different changes. At the stress difference between larger layers, the tendency of the fracture height curve is steeper

282

H. L. Luong et al.

140 120

xf, m

100 80

P3D Software

60 40 20 0 0

20

40

60

80

time, min

100

120

140

Fig. 10. Comparison of haft length calculated from the P3D model and the MFrac software

0.009 0.008

wavg, m

0.007 0.006 0.005 0.004

P3D

0.003

Software

0.002 0.001 0 0

20

40

60

80

100

120

140

time, min Fig. 11. Comparison of average width calculated from the P3D model and MFrac software

0.014

wmax, m

0.012 0.01 0.008 0.006

P3D

0.004

Software

0.002 0 0

20

40

60

80

time, min

100

120

140

Fig. 12. Comparison of maximum width calculated from the P3D model and MFrac software

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

283

6

Net pressure, MPa

5 4 3

P3D

2

Software

1 0 0

20

40

60

80

100

120

140

time, min Fig. 13. Comparison of net pressure calculated from the P3D model and MFrac software 300 250

Tip_u Tip_d

200

hf, m

150 100 50 0 -50 -100 -150 64000000

65000000

66000000

67000000 68000000 Ptreatment, N/m2

69000000

70000000

71000000

Fig. 14. Full heightmap based on correlation Ptreat

because the fracture is harder to propagate. If the reduction in stress differentials causes a decrease in the value of the net pressure so the intersection or final solution for the height value will be larger. In Fig. 15, the fracture height varies from more than 39.624 m to about 57.912 m, corresponding to the stress contrast of σ1 and σ3 . We consider the upper formation stress σ2 = 68.83; 70.11 and 71.49 MPa to study the effect of in-situ stress in the upper layers (Fig. 16). As in-situ stress σ2 in the upper layers becomes smaller and smaller, the top of the fracture easily increases infinitely with the change of treatment pressure, and when σ2 is large enough, the stress limit will completely prevent the development of the fracture and the top of the fracture will remain in a small range regardless of the large treatment pressure. Similarly, we consider the lower formation stress σ3 = 70.67, 71.49, 72.52 MPa to study the effect of stress in the lower layers (Fig. 16). When the in-situ stress (σ3) in the lower layers increases gradually, the top of the lower fracture will be limited, even at large net pressure.

284

H. L. Luong et al.

12000000 P3D_u1=4928 N/m2, d1=6768 N/m2 from PKN-C P3D_u2=6205 N/m2, d2=7584 N/m2 P3D_u3=7584 N/m2, d3=8618 N/m2

Net pressure, N/m2

10000000 8000000 6000000 4000000 2000000 0 0

50

100

150

hf, m

200

250

300

350

70000000

71000000

Fig. 15. Effect of the stress contrast on the fracture height 300 Tip_u1 Tip_d1 Tip_u2 Tip_d2

hf, m

200 100 0 -100 -200 64000000

65000000

66000000

67000000 68000000 Ptreat, N/m2

69000000

Fig. 16. Map of entire height according to the processing pressure

Net pressure, N/m2

14,000,000 P3D_u=4928 N/m2, d=6768 N/m2 PKN-C, E1=37.8E+09 N/m2 PKN-C_E2=44.8E+09 N/m2 PKN-C_E3=51.7E+09 N/m2

12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 0 0.00

50.00

100.00

150.00 200.00 hf, m

250.00

Fig. 17. Effect of plane strain modulus on fracture height

300.00

350.00

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

285

Effect of Plane Strain Modulus E’ . We consider three cases with the change in plane strain modulus and keep the remaining elements. With the same stress difference, the range of equilibrium height difference is not much when the plane strain module changes and varies from 57.912 m to 67.056 m, respectively (Fig. 17). However, in some cases with different rock data, with the change of plane strain modulus, it will not be possible to get fracture height by the equilibrium height equation, so when the results of a mini-frac process are different compared to those of the original design and the new plane strain modulus leads to failing of calculation of fracture height, it will be necessary to adjust some other parameters, for instance, decreasing the injection rate and modifying the type of the fracture fluid. Otherwise, the fracture height must be determined by other approach (different assumptions) if other parameters stay the same. Effect of Fracture Toughness (KIC ). We consider three cases with the change in fracture toughness and keep the remaining elements.

Net pressure, N/m2

12000000 PKN-C P3D_KIC1 P3D_KIC2 P3D_KIC3

10000000 8000000 6000000 4000000 2000000 0 0

50

100

150

200

250

300

350

hf, m

Fig. 18. Effect of fracture toughness on fracture height

Figure 18 shows the results of fracture height calculation with changes in fracture toughness ranging from 1,000 psi-in0.5 to 2500 psi-in0.5 . It is obvious that the fracture toughness is not a vital variable in the fracture height calculation from a P3D design. We might conclude that it is useless to measure the value of fracture toughness in the stress field, and an assumed value (the common value is 1,000 psi-in0.5 ) would be adequate. Effect of Leak-off Coefficient (CL ). We consider four cases with the change in the leak-off coefficient and keep the remaining elements. Figure 19 shows different fracture heights for different leak-off coefficients. There is a trend of lower crack pressure with the higher leak-off coefficient. This means that fracture height from the P3D model will be small, and fracture height development will be limited when the leak-off coefficient of the injection fluid is higher. In addition, the P3D model also enable the monitoring of the fracture height growth. This supports ensure preventing the fracture from entering restricted layers to avoid

286

H. L. Luong et al.

Net pressure, N/m2

12000000 P3D PKN-C_CL1=0.0022 PKN-C_CL2=0.0035 PKN-C_CL3=0.0065 PKN-C_CL4=0.03

10000000 8000000 6000000 4000000 2000000 0 0

50

100

150

hf, m

200

250

300

350

Fig. 19. Effect of leak-off coefficient on fracture height 324

216 P3D CL1=0.0022 CL2=0.0035 CL3=0.0065 CL4=0.03 upward

274

hf, m

224 174 124

192 168 144 120 96 72 48

74

24

24 0

2,000,000

4,000,000 6,000,000 8,000,000 Net pressure, N/m2

0 10,000,000 12,000,000

Fig. 20. Effect of leak-off coefficient on fracture height-upward. Suppose that there is a gascontained formation 80 ft above the payzone

intrusion of undesirable fluid formations. This diminish the threat of gas and water extraction in oil wells or water in gas wells (Figs. 20 and 21). It is crucial to handle the fracture height so that it does not spread into those areas of gas and water [12]. Effect of Slurry Injection Rate (qi ). We consider four cases with the change in the slurry injection rate and keep the remaining elements. Figure 22 shows different fracture heights for different injection flows. There is a trend of lower net pressure with a lower slurry rate. This implies that the fracture height from the P3D model will grow as the injection rate increases. In addition, the P3D model also allows the monitoring of fracture height similar to the effect of the leak-off coefficient (Figs. 23 and 24).

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

105

324 P3D CL1=0.0022 CL2=0.0035 CL3=0.0065 CL4=0.03 dowrward

274 224 hf, m

287

174 124

84 63 42 21

74 24 0

2,000,000

4,000,000 6,000,000 8,000,000 Net pressure, N/m2

10,000,000

0 12,000,000

Pnet, N/m2

Fig. 21. Effect of the leak-off coefficient on fracture height-downward. Suppose that there is a water-contained formation 70 ft below the pay zone 16,000,000 14,000,000 12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 0

P3D PKN-C_qi1=18 PKN-C_qi2=25 PKN-C_qi3=35

0

50

100

150

hf, m

200

250

300

350

Fig. 22. Effect of the slurry flow rate on fracture height - downward 374 324

P3D qi1=018" qi2=25 qi3=35 CL4=0.03 upward

hf, m

274 224 174 124 74 24 0

5,000,000

10,000,000 Net pressure, N/m2

15,000,000

264 240 216 192 168 144 120 96 72 48 24 0 20,000,000

Fig. 23. Effect of the slurry flow rate on fracture height – upward. Suppose that there is a gascontained formation 80 ft above the payzone

288

H. L. Luong et al. 105

374 P3D qi1=018" qi2=25 qi3=35 CL4=0.03 Downward

324

hf, m

274 224 174 124

84 63 42 21

74 24 0

5,000,000

10,000,000 Net pressure, N/m2

15,000,000

0 20,000,000

Fig. 24. Effect of the slurry flow rate on fracture height - downward. Suppose that there is a water-contained formation 70 ft below the pay zone

4 Conclusions The study shows that the calculation results of 2D PKN-C are the same compared with the software results, so the model of the approach is reliable. Furthermore, compared to the 3D software model, the P3D results showed a difference, but it is minor because the fracture height in the equation is the maximum height that the fracture can achieve corresponding to the given stress value and rock properties. As viscosity increases, a small CW leads to a decrease of leak-off, besides, CW is conversely proportional to the fracture length, so the fracture length increases as the viscosity increases. Optimal selection of the fracturing fluid solution both limits the expansion of clay, while ensuring the viscosity contributes to the pressure increase inside the fracture, simultaneously improving the fracture conductivity, increasing the production rate after fracturing. Closure pressure directly affects the pressure in the fracture (net pressure). The larger closure pressure is, the higher the pump pressure is required to maintain necessary net pressure. If the net pressure (or effective pressure), is small, the fracture fluid cannot hold the fracture to open.

References 1. Jose, I.A.: An analysis of the classical Pseudo- 3D model for hydraulic fracture with equilibrium height growth across stress barriers, Int. J. Rock Mech. Min. Sci. 47(4), 625–639 (2010) 2. Yang, M.: Hydraulic Fracture Production Optimization with a Pseudo-3D Model in Multilayered Lithology. SPE (2012) 3. Michael, J.E.: Benefits of a p-3D Over a 2D Model for Unified Fracture Design. SPE (2008) 4. Economides, M.J., Daniel Hill, A., Christine, E.-E.: Petroleum Production Systems, pp. 486– 487. Prentice Hall, New Jersey (1994) 5. Zadeh, A.S.: Mathematical modeling of hydraulic fracturing in shale gas reservoir (2014) 6. Rahman, M.M., Rahman M.K.: A review of hydraulic fracture models and development of an improved pseudo-3D model for stimulating tight oil/gas sand (2015)

Development of a Pseudo-3D Fracture Geometry Model in Hydraulic Fracture

289

7. Jing, X.: A PKN hydraulic fracture model study and formation permeability determination (2012) 8. Usman, A.: Fracture height prediction, J. Pet. Technol. 40(07), 813–815 (1988) 9. Liu, S.: Hydraulic fracture height: modeling and evaluation using microseismic closure window (2017) 10. Pitakbunkate, T.: Incorporating rigorous height determination Into unified fracture design (2010) 11. Liu, S., Peter, P.V.: An Improved Model for Predicting Hydraulic-Fracture-Height Migration. SPE (2016) 12. Pitakbunkate Economides, T., Yang, M., Valkó, P.P.: Hydraulic fracture optimization with a p-3D model (2011) 13. Gandossi, L.: An Overview of Hydraulic Fracturing and Other Formation Stimulation Technologies for Shale Gas Production. Publications Office of the European Union, Luxembourg (2013) 14. Budennyy, S.: An Enhanced P3D Model of Hydraulic Fracture in Multi-layered Formation. SPE (2017) 15. Zhang, X., et al.: A New Pseudo-3D Model for Hydraulic Fracture Height Growth in Multilayered Rocks. American Rock Mechanics Association (2017) 16. Tang, J., et al.: A coupled three-dimensional hydraulic fracture propagation model considering multiple bedding layers. In: Unconventional Resources Technology Conference (2018) 17. Carpenter, C.: An Improved Model for Predicting Hydraulic-Fracture-Height Migration. SPE (2016) 18. Liebowitz, H., Rice, J., et al.: Fracture: An Advanced Treatise (Vol 2, Mathematical fundamentals), pp. 214–223. Academic Press (1968)

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution at Thi Vai River, Ho Chi Minh City, Vietnam Cuong Van Anh Le1,2(B) , Man Ba Duong3 , and Thong Duy Kieu4 1 University of Science, Ho Chi Minh City, Vietnam

[email protected] 2 Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam 3 Ho Chi Minh City Institute of Resources Geography, VAST, Ho Chi Minh City, Vietnam 4 Hanoi University of Mining and Geology, 18 Vien Street, Hanoi, Vietnam

Abstract. High–resolution seismic approach equipped with a sub–bottom profiler can be applied for researching river sedimentology and bathymetry. Ganh Rai bay in Ho Chi Minh City, Vietnam as a terminal or transit place for many rivers and sea routes helps to connect its host city to the outside world. Its important geography raises high need for understanding its geology such as seabed and fluvial sediment structures. For study of the marine shallow structures from Thi Vai river to Thanh An island, we have measured, analyzed, and interpreted datasets from nine 2D high–resolution seismic profiles and three drill holes. Conventionally 2D processed seismic data and its entropy texture attribute as chaos level of amplitudes can support interpretation of 2D seabed and shallow sediment horizons. 3D seabed topography and 3D Holocene sediment distribution in Ganh Rai bay are firstly modelled from the 2D interpreted horizons. Combination of the seismic results and the drill holes information can show the area shallow geology. It consists of four layers; (i) water, (ii) mud, mixture of muds and clay layers, (iii) plastic, grayish–blue clay from Holocene, and (iv) hard grey clay, sand powder from Pleistocene. Holocene sediments are distributed shallower on the south-western edge and deeper in the south–eastern edge. Keywords: High-resolution seismic · Holocene · Bathymetry

1 Introduction High–resolution seismic is applied for imaging formations of the river sediment, seabed, and low land [1–3]. Different geological units formed in various conditions can be characterized by acoustic impedance values. The seismic waves propagate into subsurface and reflect backward at the seismic boundaries. The fixed source-receiver distance diagram depicts environment seismic boundaries as well as different seismic patterns [1–3]. Then, processing/ analysing stages are applied to the raw seismic data for further interpretation [4]. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 290–304, 2021. https://doi.org/10.1007/978-3-030-60269-7_15

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

291

In this work, we have used Sub–bottom equipment for acquisition of 2D high– resolution seismic data profiles in the fork area of Thi Vai river to Thanh An island, Can Gio district, Ho Chi Minh City. Our objective is to investigate bathymetry and Holocene sediment distribution in Ganh Rai bay covering the fork area of Thi Vai river to Thanh An island, Can Gio district. 3D visualization of the processed seismic data is used to illuminate the underground geology structures below its seabed that are compatible with available drill holes.

2 Study Area The study area is the Ganh Rai bay at Can Gio district, Ho Chi Minh city, Vietnam (Fig. 1). Can Gio district, an isolated island bounded by rivers and seashore line, is the link between the city and waterway and plays an important role in the economic development of the city. Ganh Rai bay is not only a terminal of big rivers’ journeys

Fig. 1. Location of the survey area in Ganh Rai bay, Can Gio District, Ho Chi Minh City, Vietnam [5]

292

C. Van Anh Le et al.

such as Long Tau river, Go Gia river, Thi Vai river, and Cai Mep river but also the transshipment point for maritime routes. After propagation of the sea in mid-Holocene, shoreline retreats gradually to the Vietnam east sea. The Saigon–Dong Nai river basin ignites the formation of Can Gio around 7000 to 6000 cal BP [6, 7]. Specifically, Can Gio, as the mangrove area on the Dong Nai river delta, received a much smaller amount of sediment compared to the sediment Mekong river delta collected [7]. Moreover, sand bars in front of the estuaries and Can Gio foreshore can be formed by conditions of the weather (i.e., semi-diurnal regime tidal from the estuary and average water discharge from Soai Rap and Long Tau rivers on the rainy season) [8]. The Ganh Rai area, a part of Ho Chi Minh city, is adjacent to the Mekong river delta. Therefore, the geology of the Mekong Delta and Ho Chi Minh city could provide more information to the knowledge of its sediment formation [8–14]. The northern part of Ho Chi Minh city (i.e., Ba Mieu, Cu Chi and Thu Duc) with higher elevation is dominated by the pre–Holocene Cenozoic materials while its southern part (i.e., Nha Be and Can Gio) reveals the formation of Holocene [15, 16] in near-surface. Holocene sediments formations are connected to Saigon – Dong Nai rivers [15, 16]. The river sediment, seabed topography or low land can be imaged using high-resolution seismic method [1–3]. Different geological units can be interpreted through acoustics impedances or seismic amplitudes characteristics [4].

3 Methodology High–resolution seismic data using signal reflections from acoustic impedance contrasts between layers can provide highly detailed, precise image of subsurface. In Vietnam, high–resolution seismic is applied to investigate seabed as shallow geology formations [2]. Especially, it is used in investigating shallow sediment layers in rivers, seas, and bays having their water depth from several meters to hundreds of meters. The main object of the shallow geology is young sediment formation in which their existence connects with mineral deposit exploration or construction purposes. Moreover, the method plays a crucial role in exploring sediment process and erosion process in the water bottom and other environment risks [2, 17, 18]. 3.1 Data Collection We use the Sub-Bottom Profiler system with a fixed distance between transmitter and receiver, and then these devices are carried by a ship following the designed exploring routes (Fig. 2). The seismic wave from the transmitter propagates through the water layer. It continues to go down until its echoes reflects to the receiver. The Sub–Bottom Profiler named SB–216S [19], which is located around 1.5 m underwater, emits the wave pulses down to the sea environment with frequency ranging from 2 to 16 kHz over 20 ms length. The professional program, namely EdgeTech Discover [20], plays an essential role in data acquisition. GPS locator is used for defining the position of the measurement devices. The 2D seismic profiles’ lengths range from the minimum as

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

293

700 m for the profile T1C to the maximum as 4700 m for the profile T2C (Fig. 1). The measurement profiles were conducted in June 2016. The raw data is stored as a binary format with the extension *.jsf [20]. Then, the raw can be visualized and be processed in Reflexw software [21]. For 3D interpretation purpose, the processed data are imported to the OpendTect software [22].

Fig. 2. Deployment of the sub-bottom profiler (SBS-216S) [19, 23]. The strong seismic reflection seen as white lines or green lines can represent the boundary between two geology layers (i.e., water and mud layers)

3.2 Data Processing Data processing stage can transform the raw data into interpretable processed data. In our research, two steps as Subtract-DC-shift and Divergent (div.) Compensation Gain are applied to the nine seismic profiles [21]. The two filters work on a single trace independently as below: (i) Step 1: Subtract DC Shift uses the subtraction of a time-constant shift. In each trace, within its time range, the mean value is calculated and afterwards all the samples subtract the mean value. (ii) Step 2: the first-step processed data can be compensated for the spherical divergence losses in seismic propagation down to the sea [21]. One example of our processing results is illustrated in Fig. 3 for the profile T9P. There are significant amplitude gains from the raw data (the top image) to the first-step processed data (the middle image) and the second-step processed data (the bottom image)

294

C. Van Anh Le et al.

in the larger travel-time. Then, all the second-step processed data are setup in the 3D geometry shown by OpendTect software for key seismic reflection surface interpretation [22]. Moreover, tool of seismic attribute calculation in the software can support seismic interpretation [24]. In this work, we use seismic entropy texture attribute.

Fig. 3. Processing steps for seismic data in the profile T9P. Meaningful amplitude gains from the raw data (the top image) to the Step 1 processed data (the middle image) and the Step 2 processed data (the bottom image) are visible in the larger travel-time

Entropy texture analysis: Reflectors can be defined by tracing strong amplitudes of reflected waves [25]. For supporting interpretation of the key seismic reflection, entropy attribute [4, 26] showing chaos or out–of–order level of amplitudes is sided with the conventional seismic amplitude. The textural attributes (i.e., energy, entropy, contrast, and homogeneity) were developed to “pick out zones of common signal character” from 1970’s to the mid 2000’s [27]. The grey – level co – occurrence Matrix (GLCM) [28, 27] is applied for calculating each seismic textural attribute. In our research, entropy texture is considered for expressing measure of complexity/ disorder within the GLCM matrix. It is expressed by the Eq. (1) below: N−1    (1) Pi,j −ln Pi,j Entropy = i,j=0

Where Pi,j shows the ith row and jth column of the GLCM matrix P. The entropy texture analysis can be done by using the tool in the OpendTect software [22]. Practically, the textural entropy can be calculated in a sliding window that expands

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

295

through the full seismic section. For calculating every datapoint of entropy texture in each seismic section, its sliding window needs to be defined (Fig. 4). The sliding window as the dashed rectangular having the horizontal width and the time length (Fig. 4) is chosen to capture seismic information. The horizontal width includes number of seismic traces as equal as 2x Horizontal Stepout and the time length is defined as 2x time gate [22].

Fig. 4. 2D sliding window for computing seismic textural attributes [22]. The resulted data point of a seismic textural attribute is highlighted by the red dot. Its sliding window defined as a dashed rectangular has horizontal width and time length as 2x Horizontal Stepout, and 2x Time Gate, respectively

We can both use the conventional amplitude (i.e., the second step processed data) and its entropy attribute to specify 2D boundaries of each 2D seismic profile (See Fig. 5). Then 3D interpolated horizons are built up by using the Matlab built-in function, scatteredInterpolant.m [29]. In this field area, we have firstly imaged a 3D sea bottom horizon by the sub–bottom profiler data. Moreover, the top and bottom of Holocene horizons can form their Holocene layer supported with the interpretation of the three drill holes in the survey area. For two–way travel time (TWT) to depth conversion, we assumed sound velocities as 1500 m/s and 1550 m/s for sea water and subsurface sediments, respectively [2]. For checking the entropy texture section (See Fig. 5), we can see that within a layer, the seismic amplitude is quite similar leading to low value of randomness/chaos in entropy texture. However, strong reflection boundary shows the existence of possible geological layers’ boundaries resulting high entropy texture. For validating the seismic data, we apply the technique of checking the meeting point of two different seismic profiles in which their depths at their meeting point should be similar. In Fig. 6, blue arrows pointing at some meeting points of crossed seismic profiles can illustrate the similar depth of seabed shown as: (i) T3C and T4P, (ii) T9P and T3C, (iii) T9P and T2C, (iv) T9P and T1P, and (v) T3C and T2C.

296

C. Van Anh Le et al.

4 Results and Discussion In the area, the reflectors between different environments are presented through the distribution of strong reflected seismic waves. We have analyzed different wave shapes and their entropy values (Fig. 5) for imaging the reflectors. The water bulk is effectively interpreted from the surface down to the seabed by the first strong seismic boundary (see the white line in Fig. 5). Thin mud layers locate between the water bulk and the Holocene layer. The sediment of Holocene can be interpreted through each 2D seismic horizons (see top Holocene horizon as green line and bottom Holocene horizon as pink line), but its recognition is not continuous. That is, the Holocene sediment is eroded within

Fig. 5. Processed seismic data (upper) and its seismic attribute entropy (below). Green and pink lines are interpreted as top and bottom of Holocene layer, respectively

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

297

the trough of seabed in the middle of the research area from Thi Vai river to Thanh An island. Both the Holocene and Pleistocene sediments are found in the area. According to the drill holes, form of the sediment is composed of mainly coarse grains.

Fig. 6. Seismic representation for checking validity of the seismic data measurement. Two image examples (up and down) of meeting points from several seismic profiles where their depths at the meeting point share the similar values. Blue arrows show positions of meeting points. White lines are interpreted as seabed. The meeting depths (i.e., T3C and T4P, T3C and T9P, T3C and T2C) show the high quality of processed seismic data.

3D seabed is presented in the Fig. 7. The deepest area in the middle of the interest research area can reach to 37 m. Strong seismic reflection and rapid entropy variation in the boundary between marine material and the water layer can show the seabed quite well. The seabed depth in the side of Can Gio area is shallower than the one near the side of Ba Ria – Vung Tau province according to the analysis of all the nine seismic profiles when comparing their seabed depths close to their shorelines (i.e., see Fig. 7). Note that multiple noises resulted from the seabed can contaminate the real signal from other real signals from other deeper layers (See upper image in Fig. 5). Recognition of

298

C. Van Anh Le et al.

the multiples can be detected when they have high similarity in seismic pattern with the first arrival seismic wave along with the profile but double or triple in two–way travel time.

Fig. 7. 3D seabed made by 9 seismic profiles. The deepest area locates in the area middle. The seabed in the side of Can Gio area is shallower than the one near the side of Ba Ria – Vung Tau province

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

299

3D top Holocene surface and the Holocene thickness are display in Fig. 8. The key point shows that the deepest seabed in the middle area is also representing the lack of Holocene sediment material (Fig. 9). Two visible examples are in the profiles, T2C and T3C, where the seabed erodes the continuity of Holocene (See Figs. 7, 8 and 9). The Holocene thickness ranges from zero meter in the eroded seabed trough of the middle area to around 18 m. Especially, the thickest Holocene layer locates in the T1C profile near Can Gio shoreline (See Fig. 8d). The Holocene layer thickness in Fig. 8d interpreted from our research in Ganh Rai bay area (close to Can Gio and Ba Ria–Vung Tau) can provide an additional information to the Holocene thickness map made by Bui Viet, Stattegger, Unverricht, Phung Van and Nguyen Trung [2].

Fig. 8. 3D top and thickness of Holocene layer made by 9 seismic profiles. Holocene and Pleistocene layers are cut through by the deepest seabed in the area middle

We have investigated some distinct formations located in the Ganh Rai bay (Fig. 10). Reflected seismic data see the mud layers deposit in the seabed. Mud layers locating between the water layer and the Holocene layer can be interpreted in several 2D seismic profiles (Fig. 10). Two visible sediment layers (i.e., mud layers) are divided by a set of

300

C. Van Anh Le et al.

white arrows. For the seismic parts close to the shoreline of Can Gio, the diagonal stripe seismic pattern from the first mud layer is consistent in the profiles 1C, 3C, 4C, 4P, and 9P (Fig. 10).

Fig. 9. 3D top of Holocene layer made by 9 seismic profiles display with three drill holes. The thickest Holocene layer positions in the profile T1C (near Can Gio shoreline) compatible with information of the drill hole LK1. Meanwhile, the two drill holes LK2 and LK3 show the thinner Holocene layer near Thanh An island

From information of the three drill holes in Figs. 9, 10 and 11, three layers are interpreted as; (i)

the first layer contains mud, greyish or greyish–blue silk clay, organic matter from Present with depth range from 0 to 35 m, (ii) the second layer consisting of Holocene sediment matters has the shallowest depth, around 5 m (in the drill hole LK1) and the deepest, 37 m (in the drill LK3), (iii) and the third layer includes Pleistocene matters, seen from 33 m to 40 m (the maximum depth of the drill holes). For comparing results of the drill holes and seismic interpretation, the first two sediment layers in seismic can be matched with the first layer from drill holes. The Holocene pattern in the drill holes is compatible with the seismic layers bounded by two 3D interpreted seismic horizons (green and pink lines in Fig. 11). Interestingly,

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

301

the shallowest depth (5 m) of a Holocene layer in drill hole LK1 is matched with the shallowest depth (12 m) in the seismic profile T1C (see Figs. 7, 8, and 11). For the Pleistocene layer, its bottom layers are hardly interpreted because of lack of seismic signal in the deeper zone. Note that the drill hole LK1 is near the meeting point of Thi Vai and Go Gia rivers and two in Thanh An island (See Fig. 1).

Fig. 10. One 3D view of seismic sections with Holocene layer made by nine seismic profiles and interpretation of the first sediment layer. The white arrows represent the boundaries of different mud layers within the first sediment layer. The two visible mud layers are divided by a set of white arrows upper the Holocene top (see green lines). For the seismic parts close to the shoreline of Can Gio, the diagonal stripe seismic pattern from the first sediment layer is consistent in the profiles 1C, 3C, 4C, 4P, and 9P

302

C. Van Anh Le et al.

Fig. 11. Representations of three drill holes with their two nearby seismic data profiles, T1C and T4C

5 Conclusion Application of Sub–bottom profiler helps to image 3D seabed and sedimentation structures in Ganh Rai bay, Can Gio, Ho Chi Minh City. The high–resolution seismic imaging reflects contrast properties of different geology layers. Combination of the conventional processing techniques and seismic entropy texture calculation helps to effectively interpret 2D seismic boundaries which can be then interpolated into the 3D horizons. High consistency between the seismic result and lithology information from the drill holes can reveal the shallow geology structures such as 3D Holocene layer as well as existence of materials from Present and Pleistocene time. Knowledge of the distribution of the Holocene sediment and its lithological components is to help scientists and local authority in renovating the riverbed and canals for enhancing the efficiency of the waterway transport, in particular Can Gio District and Ho Chi Minh City in general. Acknowledgments. The authors would like to thank the Ho Chi Minh City Institute of Resources Geography project 2015 named “Research Holocene sediment distribution in fork area of Thi Vai river to Thanh An island, Can Gio District by using high–resolution seismic reflection” for accessing its data. We would like to thank Curtin University for its software support. We are also grateful to Dr. Long Quoc Nguyen from Hanoi University of Mining and Geology for his useful advices. Conflicts of Interest. The authors declare no conflict of interest.

High–Resolution Seismic Reflection Survey of Holocene Sediment Distribution

303

References 1. Ianniruberto, M., Campos, J.E., Araújo, V.: Application of shallow seismic profiling to study riverbed architectural facies: a case study of the Tocantins river (Pará-Brazil). Anais da Academia Brasileira de Ciências 84, 645–654 (2012) 2. Bui Viet, D., Stattegger, K., Unverricht, D., Van Phung, P., Nguyen Trung, T.: Late PleistoceneHolocene seismic stratigraphy of the Southeast Vietnam Shelf. Global Planet. Change 110, 156–169 (2013) 3. Novak, B., Björck, S.: Late Pleistocene–early Holocene fluvial facies and depositional processes in the Fehmarn Belt, between Germany and Denmark, revealed by high-resolution seismic and lithofacies analysis. Sedimentology 49, 451–465 (2002) 4. Le, C.V.A., Harris, B.D., Pethick, A.M.: New perspectives on solid earth geology from seismic texture to cooperative inversion. Sci. Rep. 9, 14737 (2019) 5. https://www.bandovn.vn/vi/page/mau-ban-do-hanh-chinh-nuoc-cong-hoa-xa-hoi-chunghia-viet-nam-181 6. David, F., Meziane, T., Tran-Thi, N.T., Van, V.T., Thanh-Nho, N., Taillardat, P., Marchand, C.: Carbon biogeochemistry and CO2 emissions in a human impacted and mangrove dominated tropical estuary (Can Gio, Vietnam). Biogeochemistry 138, 261–275 (2018) 7. Fujimoto, K., Umitsu, M., Nguyen, V.L., Ta, T.K.O., Kawase, K., Huynh, D.H., Nakamura, T.: Geomorphological evolution and mangrove habitat dynamics related to Holocene sea-level changes in the northern Mekong river delta and the Dong Nai river delta, southern Vietnam. In: Schmidt, P.E. (ed.) River Deltas: Types, Structures and Ecology. Nova Science Publishers, Inc. (2011) 8. Bui, V.T., Huynh, T.T., Le, T.N.D., Ly, H.M., Le, P.T.: Monitoring and predicting the shoreline change in Can Gio area in condition of the sea level rise. Sci. Technol. Dev. J. 17, 45–53 (2014) 9. Stattegger, K., Tjallingii, R., Saito, Y., Michelli, M., Thanh, N.T., Wetzel, A.: Mid to late Holocene sea-level reconstruction of Southeast Vietnam using beachrock and beach-ridge deposits. Global Planet. Change 110, 214–222 (2013) 10. Thoang, T., Giao, P.: Subsurface characterization and prediction of land subsidence for HCM City. Vietnam. Eng. Geol. 199, 107–124 (2015) 11. Ta, T.K.O., Nguyen, V.L., Tateishi, M., Kobayashi, I., Tanabe, S., Saito, Y.: Holocene delta evolution and sediment discharge of the Mekong River, southern Vietnam. Quat. Sci. Rev. 21, 1807–1819 (2002) 12. Ta, T.K.O., Nguyen, V.L., Tateishi, M., Kobayashi, I., Saito, Y.: Holocene delta evolution and depositional models of the Mekong River Delta, southern Vietnam. In: River Deltas–Concepts, Models, and Examples, vol. 83. SEPM (Society for Sedimentary Geology) (2005) 13. Ta, T.K.O., Nguyen, V.L., Tateishi, M., Kobayashi, I., Saito, Y.: Sedimentary facies, diatom and foraminifer assemblages in a late Pleistocene-Holocene incised-valley sequence from the Mekong River Delta, Bentre Province, Southern Vietnam: the BT2 core. J. Asian Earth Sci. 20, 83–94 (2001) 14. Nguyen, V.L., Ta, T.K.O., Saito, Y.: Early Holocene initiation of the Mekong River delta, Vietnam, and the response to Holocene sea-level changes detected from DT1 core analyses. Sed. Geol. 230, 146–155 (2010) 15. Kitazawa, T.: Pleistocene macrotidal tide-dominated estuary–delta succession, along the Dong Nai River, southern Vietnam. Sed. Geol. 194, 115–140 (2007) 16. Kitazawa, T., Nakagawa, T., Hashimoto, T., Tateishi, M.: Stratigraphy and optically stimulated luminescence (OSL) dating of a quaternary sequence along the Dong Nai River, southern Vietnam. J. Asian Earth Sci. 27, 788–804 (2006) 17. Yutsis, V., Krivosheya, K., Levchenko, O., Lowag, J., de León Gómez, H., Ponce, A.T.: Bottom topography, recent sedimentation and water volume of the Cerro Prieto Dam, NE Mexico. Geofísica internacional 53, 27–38 (2014)

304

C. Van Anh Le et al.

18. Aiello, G., Di Fiore, V., Marsella, E., Passaro, S.: High resolution seismic data coupled to Multibeam bathymetry of Stromboli island collected in the frame of the Stromboli geophysical experiment: implications with the marine geophysics and volcanology of the Aeolian Arc volcanic complex (Sicily, Southern Tyrrhenian sea, Italy). SpringerPlus 3, 1–27 (2014) 19. https://www.edgetech.com/edgetech-introduces-new-portable-sub-bottom-profiling-system/ 20. https://www.edgetech.com/resource-center/ 21. https://www.sandmeier-geo.de/reflexw.html 22. dGB Earth Sciences: OpendTect dGB Plugins User Documentation version 4.6 (2015) 23. VectorStock: Ship Image 24. Le, C.V.A., Harris, B.D., Pethick, A.M., Takam Takougang, E.M., Howe, B.: Semiautomatic and automatic cooperative inversion of seismic and magnetotelluric data. Surv. Geophys. 37, 845–896 (2016) 25. Yilmaz, O.: Seismic Data Analysis: Processing, Inversion, and Interpretation of Seismic Data. Society of Exploration Geophysicists, United States of America (2001) 26. Haralick, R.M., Shanmugam, K., Dinstein, I.: Textural features for image classification. IEEE Trans. Syst. Man Cybern. SMC-3, 610–621 (1973) 27. Chopra, S., Marfurt, K.J.: Seismic attributes for prospect identification and reservoir characteriztion. Tulsa, Okla. (8801 South Yale St., Tulsa OK 74137-3175). Society of Exploration Geophysicists United States of America (2007) 28. http://www.fp.ucalgary.ca/mhallbey/tutorial.htm 29. https://www.mathworks.com/help/matlab/ref/scatteredinterpolant.html

Initial Results of Using Biochar Derived from Spent Coffee Grounds to Remove Pollutants from Livestock Wastewater in Vietnam Tran Thi Thu Huong1 , Nguyen Van Hoang2 , Vu Ngoc Toan3 , Nguyen Xuan Tong4(B) , Tran Anh Quan1 , and Vu Kim Thu5 1 Faculty of Environment, Hanoi University of Mining and Geology, Hanoi, Vietnam 2 Institute of New Technology, Military Institute of Science and Technology, Hanoi, Vietnam 3 Institute for Chemistry and Materials, Military Institute of Science and Technology,

Hanoi, Vietnam 4 Institute of Environmental Science, Engineering, and Management, Industrial University of

Ho Chi Minh City, Ho Chi Minh City, Vietnam [email protected] 5 Faculty of Basic Sciences, Hanoi University of Mining and Geology, Hanoi, Vietnam

Abstract. Biochars derived from spent coffee grounds were pyrolysed at different temperatures and retention times, including CF1-CF4 samples (500 °C for 0.5, 1.5, 3, 6 h); CF5-CF8 samples (600 °C for 0.5, 1.5, 3, 6 h) and CF9-CF12 samples (700 °C for 0.5, 1.5, 3, 6 h). These biochars were examined to determine their ability to remove pollutants (COD, TSS, total N and total P) from livestock wastewater. The initial livestock wastewater was treated with 12 types of biochar with masses of 2, 4 and 6 g at reaction times from 1, 2, 4 to 8 h to assess the adsorption efficiency of the biochar. Adsorption efficiency for these pollutants increased with increasing reaction time and biochar mass. The combination of 8 h reaction time and 6 g biochar weight showed the highest adsorption efficiency. At an 8 h reaction time with 4 g biochar, only COD content was adsorbed by the CF4 biochar sample at a level meeting the output requirements according to the Vietnam standard QCVN 40:2011/MONRE national regulation for industrial wastewater; the remaining 11 treated wastewater samples retained pollutant concentrations that were 1.6 to 3.6 times higher than the acceptable values. The TSS content in all 12 samples met the standard limit value requirement. The total N content was 3.3 to 4.2 times higher (excepting the CF4 sample) and the total P content was 1.07 to 1.15 times higher (excepting the CF4, CF8, CF9 and CF11 samples) than the standard limit values. With 6 g biochar and 8 h reaction time, all four parameters adsorbed with 12 biochar samples were significantly reduced, producing water with concentrations lower than the required limit according to the QCVN 40:2011/MONRE regulation. The results showed that the biochar made from spent coffee grounds is a potential sorbent to remove pollutants from livestock waste water. Keywords: Adsorption capacity · Biochar · Coffee ground · Livestock wastewater · Pollutants · Vietnam © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 305–325, 2021. https://doi.org/10.1007/978-3-030-60269-7_16

306

T. T. T. Huong et al.

1 Introduction Organic pollution from livestock wastewater is a major problem in many countries because of its danger to the environment and human health [1]. Many technological solutions to neutralize pollutants have been developed to protect the environment. Currently, wastewater treatment technologies are available to remove pollutants at various scales and levels. However, none of them is optimal for all goals and economic conditions. Many treatment methods have been successfully applied to many types of wastewater, but they usually have high operating and maintenance costs, generate secondary toxins, or involve complicated operations [2]. One of the most concerning issues is the need to find alternative materials to support wastewater treatment technology. Biochar modified from agricultural wastes or other carbon-rich materials is potentially a low-cost environmental treatment material, particularly for organic pollution remediation [2]. Recently, biochar has received a lot of attention worldwide when applied in wastewater treatment thanks to its high adsorption efficiency and reusability. In addition, its production cost is low and it is safe for the environment [1]. Many researchers have used biochar adsorption for organic pollutants [3], such as sulfamethoxazole [4], phenanthrene and triazine pesticides [5], and heavy metals including mercury, lead and cadmium [6] from waste water. Popular processes often used to synthesize biochar include pyrolysis, gasification, and hydrothermal carbonization [1]. Coffee is a widely consumed drink with beans grown in more than 80 countries worldwide, and is considered one of the most important goods in these countries [7]. The total amount of coffee consumed worldwide exceeds 11 billion tons/year [7]. According to the International Coffee Organization (ICO), the global coffee yield has steadily increased during the last 150 years [8]. Coffee brew production is estimated at 151.6 million bags (60 kg/bag) for 2015–2016, and coffee grounds have filled up large landfills [7]. It is estimated that more than 4.4 million tons of brewed coffee waste are disposed of by this industry every year [9]. Coffee grounds contain high levels of substances known to be toxic to many life processes, such as acids, free phenols, caffeine, and tannins (polyphenols) [10]. Coffee waste thus constitutes a source of serious environmental problems in coffee-using countries [11]. Many authors have already reported that coffee sub-products and their wastes could be used in many ways [12]. Biochar has been used as a useful and economical material in many fields such as food additives, biogas, caffeine, pectin, feeds, protein, antioxidants, beverages, enzymes, compost, and the production of bioactive compounds [12]. According to some previous studies, protein content in coffee grounds accounts for about 10%, pectin content for 52.62–55.14%, and cellulose for 15.29–17.04% of the mass of the grounds, which also have high carbon content (above 50%) [13]. Coffee grounds are a lignocellulosic material that can separate heavy metals and dye in water based on their porous structure and cellulose composition. The surface of carbonaceous materials contains many phenolic hydroxyl and carboxyl groups. In cellulose materials, these groups play an important role in ion exchange, and studies of treatment effectiveness indicated that the adsorption of pollutants depends upon the surface polar groups on the carbonaceous materials [13]. Therefore, spent coffee grounds are an effective material in the synthesis of biochar in order to remove pollution from livestock wastewater.

Initial Results of Using Biochar Derived from Spent Coffee Grounds

307

The livestock industry accounts for about 40% of global agricultural output, creating jobs for more than 1.3 billion workers and supporting the livelihoods of more than 1 billion people in poor countries [1, 14]. Agricultural economic development associated with large-scale livestock production is an important means to help farmers increase their incomes [15]. Farming with large numbers of livestock will release large masses of waste; the aquatic environment is seriously threatened by pollutants such as phosphorus, ammonium and heavy metals. In Vietnam, the total amount of livestock waste produced is about 73 million tons/year, of which pig waste accounts for about 33.4% or 24.38 million tons/year, and 25–30 million cubic meters of liquid (liquid faeces, urine and rinse water). Of this, about 50% of solid waste (36.5 million tons) and 80% of liquid waste (20–24 million m3 ) are released directly into the environment or without treatment, which causes serious environmental pollution [16]. Biochar’s potential as an adsorption technology in wastewater treatment, especially for the low-risk handling of typical pollutants in livestock wastewater, is a current focus of efforts to make agricultural activity more environmentally sound. Many authors previously reported the result that the typical pollutants in livestock wastewater, such as organic pollutants [17], heavy metals [18], nitrogen, and phosphorus [19] could be adsorbed well by biochar. Therefore, biochar can be used as a slow-release fertilizer and is a material with agricultural environment-friendly characteristics [20]. According to the Ministry of Agriculture and Rural Development, the coffee-growing area in Vietnam in 2014 was estimated at 653,000 hectares, and the output reached 1.49 million tons. Vietnam is an agricultural country with the second-largest coffee export mass in the world (after Brazil). The total domestic consumption of coffee is 60,000 tons/year, of which instant coffee accounts for about 19,000 tons, roast/grind coffee accounts for 35,000 tons, and the remainder is unbranded roast/grind coffee [21]. Spent coffee grounds are a carbon-rich material with many advantageous characteristics [8, 9, 12]. Vietnamese scientists have carried out experiments on biochar mainly to improve soil characteristics for agriculture [22–24]. Studies of biochar for pollutant adsorption in Vietnam have included investigating the characteristics of biochar from rice husk [25], biochar modified H3 PO4 and NaOH for ammonium removal [26], the fixation of pesticides such as propoxur by biochar [27], using biochar modified from corn cobs to remove ammonium in mixture wastewater [28], etc. However, studies of biochar’s utility for removing pollutants/organic pollution or pathogenic microorganisms from livestock wastewater are very limited. From the above data, it can be seen that Vietnam’s annual yield of coffee grounds is huge; they are usually discarded, wasting a potential source of raw materials. To best utilize this strength of a third world coffeeexporting nation, we propose the synthesis of biochar materials by the slow pyrolysis method from spent coffee grounds to remove pollutants from livestock wastewater in an eco-friendly manner. In this study, laboratory experiments were performed to survey the removal potential of organic pollution from livestock wastewater by biochar pyrolysed from spent coffee grounds.

308

T. T. T. Huong et al.

2 Materials and Methods 2.1 Preparation of Carbon Material from Spent Coffee Grounds by Pyrolysis The coffee grounds used as raw materials were collected from a local company in Dak Lak province (An Thai Development and Investment Joint Stock Company, Lot B03–04 – Hoa Phu Industrial Pack – Buon Ma Thuot City – Dak Lak Province – Viet Nam) and a coffee shop in Ha Noi city (Titi Coffee Shop, No. 20 Vien Street, Duc Thang Ward, Bac Tu Liem District, Ha Noi city, Viet Nam). The steps to collect raw material are described in Fig. 1 and Fig. 2 as follows:

Fig. 1. Process of collecting raw material from the An Thai JSC

To remove the chemicals left in coffee grounds after brewing, the spent coffee grounds were washed multiple times with water by centrifugation methods and then filtered [39]. Next, the raw material was mixed together thoroughly and then dried at 105 °C in an oven for 10 h to evaporate the solvent completely. The pyrolysis of the material was then conducted at 500 °C, 600 °C and 700 °C in the presence of nitrogen. The heating rate of the furnace was set at 5 °C/min and temperature was maintained for 0.5 h, 1.5 h, 3 h and 6 h [30]. After the end of the pyrolysis phase, the material was manually ground to enhance the homogeneity in the whole material. 12 types of biochar material including CF1 (500 °C/0.5 h), CF2 (500 °C/1.5 h), CF3 (500 °C/3 h), CF4 (500 °C/6 h), CF5 (600 °C/0.5 h), CF6 (600 °C/1.5 h), CF7 (600 °C/3 h), CF8 (600 °C/6 h), CF9 (700 °C/0.5 h), CF10 (700 °C/1.5 h), CF11 (700 °C/3 h), CF12 (700 °C/6 h) and a control sample (raw spent coffee grounds) were prepared for further experiments. All chemicals and reagents used in the current research were ordered from Merck with A.C.S. certification.

Initial Results of Using Biochar Derived from Spent Coffee Grounds

309

Fig. 2. Process of collecting raw material from Titi Coffee Shop

The collected materials were evaluated for their ability to remove pollutants (COD, TSS, Total N and total P) based on three characteristics, including: (1) reaction time, (2) adsorbed biochar weight and (3) effective adsorption. 2.2 Identification of Material Characteristics The biochar material samples were characterized in the following manner: energydispersive X-ray spectroscopy (EDX) was used to determine the elemental composition of the material, and a scanning electron microscope (SEM) system was used to analyse the surface size and morphology. The Brunauer Emmett Teller (BET) technique was used to analyse the surface area and properties; the N2-BET surface area and properties of the biochar samples were measured by Tristar 3000. A Fourier Transform Infrared Spectrometer (FTIR, TESOR II- BRUCKER, USA) was also used to record and determine the surface functional groups on the sample. 2.3 Determination of the Adsorption Capacity of the Materials Sampling and Chemical Analysis The initial livestock wastewater was collected from a farm in Vinh Phu province and preserved before experiments in the laboratory. The wastewater samples were immediately analyzed for contamination factors to determine the initial pollution concentrations for COD, TSS, total N and total P. The analysis results in Table 1 showed that all four parameters exceed the permitted values according to the Viet Nam national regulation QCVN 40:2011/MONRE for industrial wastewater: COD concentrations exceeded their

310

T. T. T. Huong et al.

limit by a factor of 52, TSS by a factor of 2.7, Total N by a factor of 13.66 and total P by a factor of 3.66. Table 1. Analysis results of pollutants in the initial livestock wastewater sample Parameter

Unit

Concentration

Vietnam national standard 40:2011/MONRE

Exceedance factor

COD

mg/L

7800

150

52

TSS

mg/L

270

100

2.7

Total N

mg/L

546.32

40

13.66

Total P

mg/L

21.95

6

3.66

Experimental Setup To determine the adsorption capacity and efficiency of the material, the raw livestock wastewater was treated with 12 types of biochar samples at different reaction times of 1, 2, 4 and 8 h. After the livestock wastewater was filtered through filtering columns made from twelve types of biochar with various combinations of reaction time and biochar mass, the treated wastewater sample was collected and analyzed for indicators of COD, TSS, total N and total P. The adsorption capacity qe (mg/g) at equilibrium time is determined by the formula [31]: qe = (Co − Ce .V/m)

(1)

The adsorption efficiency H (%) at equilibrium time is determined by the formula [32]: H = (Co − Ce )/Co · 100(%)

(2)

where qe is the adsorption capacity at equilibrium (mg/g), Co is the initial concentration (mg/L), Ce is the concentration at equilibrium (mg/L), V is the volume of solution (L), and m is the mass of adsorbent material (g). To study the effect of the reaction time, the concentration of the initial adsorbent solution Co (mg/L), the adsorption volume (V = 100 mL), and the amount of the adsorbent (m = 4 g) were fixed, and the reaction time was varied with t = 1, 2, 4 and 8 h. For asssesing the effect of the adsorbent mass, the concentration of the initial adsorbent solution Co (mg/L), the adsorption volume (V = 100 mL), and the reaction time (t = 8 h) were fixed, and the amount of the adsorbent was changed with m = 2, 4 and 6 g. The pollution parameters were analyzed following the standard methods listed in Table 2.

Initial Results of Using Biochar Derived from Spent Coffee Grounds

311

Table 2. Standard methods of water analysis used in this study Parameter

Method

COD

ISO 6060:1989

TSS

ISO 11923:1997

Total nitrogen

ISO10048: 1991

Total Phosphorus ISO 6878:2004

The optimization experimental conditions are described in the chart as follows (Fig. 3):

Fig. 3. Summary of optimization experimental conditions

3 Results and Discussion 3.1 Characteristics of the Biochar Samples The selected properties of 12 biochar samples are listed in Table 3. The ash content of all these sorbents was relatively high (>25%), especially CF4 (31.25%). The CF4 sample also had the largest BET surface area of the 12 tested biochar samples. The BET surface area of the other samples was quite low, varying from 0.7917 to 1.2466 m2 /g. The results of elemental analysis showed high C content in all 12 biochars (over 80%). The surface element contents in Table 3 showed that the CF4 biochar sample had a C content of 90.61% on its surface, while the CF9 sample had the smallest surface C content (80.58%). However, the CF9 sample had the highest O content (13.46%), indicating that the CF9 sample may have the most functional oxygen groups.

312

T. T. T. Huong et al. Table 3. Biochar physical and chemical properties

Biochars

Pyrolysis temperature and time (°C)

Ash content (%)

BET surface area (m2 /g)

Surface elements C (%)

O (%)

CF1

500 °C/0.5 h

28.6

0.7917

84.55

9.98

3.91

CF2

500 °C/1.5 h

29.6

0.8564

84.61

9.06

4.82

CF3

500 °C/3 h

30.94

1.2466

84.65

9.53

4.06

CF4

500 °C/6 h

31.25

1.5016

90.61

6.86

1.41

CF5

600 °C/0.5 h

27.42

1.085

83.29

10.10

3.92

CF6

600 °C/1.5 h

27.57

0.9766

84.15

10.38

2.24

CF7

600 °C/3 h

27

0.7233

84.13

10.43

2.55

CF8

600 °C/6 h

26.55

0.8425

81.09

12.25

2.95

CF9

700 °C/0.5 h

25.11

0.956

80.58

13.46

3.29

CF10

700 °C/1.5 h

25.51

1.1071

81.76

12.73

2.62

CF11

700 °C/3 h

26,41

1.1602

81.63

12.79

2.94

CF12

700 °C/6 h

25.30

1.0569

80.97

11.26

3.76

N (%)

SEM images of the material structures were consistent with some previously published studies [17, 33–36]. The SEM micrographs (Fig. 5, 6 and 7) of the 12 biochar samples showed that their structures were homogeneous and differed little from each other. It was also observed that the biochar samples had pores in them, appearing in microporous form, which did not exist in the raw material (Fig. 4). The raw material has a dense material surface without porous structure and lacking in cellulose fibres. Most biochar samples produced from carbonized coffee grounds have many large-diameter pores on their rough surfaces. However, the CF4 biochar sample showed a significant difference in the surface structure compared to the remaining biochar samples: its porous surface structure was observed to be the most homogeneous among the 12 samples. CF4 is the sample with the highest C surface content (90.61%). The rough surface and porous structure with small channels of these biochars can be clearly observed and may have been either produced during the pyrolysis process or a characteristic of the raw material [34]. These characteristics will strongly affect their adsorption properties [33, 34]. It is also shown that biochar materials have better adsorption capacity than raw materials. The FTIR results of the 12 biochar samples are shown in Fig. 5, 6 and 7. The spectra of sample groups pyrolysed at different temperatures clearly differ from each other. In this study, the adsorption spectra of 12 biochar samples varied from 400 to 4000 cm−1 . All samples show spectral features associated with bonds such as C=O, O-H, C-C, C=C, etc. Due to high C contents of over 80%, the FTIR spectra of all 12 samples show peaks in the range from 3423 to 3434 cm−1 for adsorption groups of O=H. The C=C groups also appeared in most samples, with a peak at 1623 to 1628 cm−1 (CF2, CF4 and CF5-CF8); the CF1 and CF3 samples appeared to show C=C adsorption groups as a peak at 1574–1583 cm−1 . However, the biochar samples pyrolysed from coffee

Initial Results of Using Biochar Derived from Spent Coffee Grounds

313

Fig. 4. SEM images of raw material samples (spent coffee grounds)

Fig. 5. SEM images and FTIR spectra of biochar samples pyrolysed from spent coffee grounds at 500 °C

grounds at 500 °C also show the adsorption groups of a C-H stretch (peak at 2918– 2924 cm−1 ), C-H aromatics (out-of-plane bend, peak at 873–876 cm−1 ), C-Cl groups (peak at 561–568 cm−1 ) and N=O (peak at 1372–1423 cm−1 ). Moreover, adsorption groups in biochar samples pyrolysed from coffee grounds at 700 °C are observed that appear in the 1000–1400 cm−1 band for C-F bonds. Finally, of the biochar samples pyrolysed from coffee grounds at 600 °C, only the CF8 sample contains adsorption groups of C-F (1089 cm−1 ) and C-Cl (596 cm−1 ). These results indicated that pyrolysis time and temperature significantly affected the material characteristics. Some authors

314

T. T. T. Huong et al.

Fig. 6. SEM images and FTIR spectra of biochar samples pyrolysed from spent coffee grounds at 600 °C

Fig. 7. SEM images and FTIR spectra of biochar samples pyrolysed from spent coffee grounds at 700 °C

have reported that the combination of micropores with larger pores [35] will lead to an increase in porous structure and decrease in surface area; therefore adsorption capacity for pollutants increased accordingly [35]. 3.2 Effect of the Reaction Time The concentrations of COD, TSS, total N and total P before and after adsorption by the 12 biochar samples are shown in Fig. 8 and Table 4. The differing results of COD, TSS, total N and total P were analysed in 12 different biochar samples, showing that the biochar had complex compositions. These results are consistent with the FTIR results obtained for the material characteristics. After different reaction times (from 1 to 8 h), the COD contents of almost all samples were significantly reduced (Fig. 7). The fastest reduction efficiency was recorded after 1 h of treatment; the reduction speed tended to slow down after 2 to 4 h and reached the highest efficiency after 8 h. The amount of COD adsorbed by the CF4 biochar sample was highest (98.08%) at the reaction time of

Initial Results of Using Biochar Derived from Spent Coffee Grounds

315

8 h and lowest (47.18%) after 1 h adsorbed with CF12 biochar. However, only the COD parameter in the sample treated with the CF4 biochar sample for 8 h met the output requirements according to the Viet Nam standard QCVN 40:2011/MONRE - national regulation on industrial wastewater. The other samples remained 1.6 to 3.6 times higher than the standard limit values. Figure 8 and Table 4 shows the TSS analysis results after adsorption onto the biochar samples. The results showed that when the reaction time was altered from 1, 2, or 4 h to 8 h, the capacity of the 12 biochar samples to adsorb the TSS increased, removing colour and clear contaminants in wastewater. As seen for the COD parameter, the adsorption efficiency of TSS reached its highest value after 8 h, the fastest adsorption efficiency was obtained at the reaction time of 1 h. Between 2 h and 4 h, the adsorption speed decreased. The TSS content of wastewater treated with all 12 biochar samples after adsorption met the output requirements according to the QCVN 40:2011/MONRE standard. Of these, the CF4 biochar sample was observed to have the highest adsorption efficiency (95.56%) at 8 h while the lowest value was observed for the CF12 sample (63.7%) at 1 h. The remaining samples showed reductions in TSS content varying from 67.78% (CF5) to 93.7% (CF6) after 1 and 8 h adsorption, respectively. Compared to pollutant parameters in the initial wastewater, the total N concentrations of all water samples after adsorption onto biochar were reduced significantly at a reaction time of 1 h. Between 2 h and 4 h, the reaction speed slowed. The total N value of the CF4 biochar water sample was reduced with the highest adsorption efficiency (93.16%) at a reaction time 8 h, while the lowest adsorption efficiency (25.56%) was observed in the CF3 sample at a reaction time of 1 h. Only the total N value obtained from the CF4 biochar sample after 8 h met the output requirements according to the Viet Nam standard QCVN 40:2011/MONRE, similar to the COD parameter results. The other 11 water samples had total N values 3.3 to 4.2 times higher than the standard limit values. Results of the experiments to assess the effect of reaction time on total P are shown in Fig. 7. As with COD, TSS and total N, total P decreases significantly after a reaction time of 1 h. At 2 h, a further decrease is only observed in some samples (CF6, 7, 8 and CF10). At 4 h reaction time, the adsorption speed also decreased. Adsorption efficiency of total P increased significantly and reached its maximum (77.04%) after 8 h in the CF4 sample. The lowest result (15.44%) was obtained from the CF12 sample at 1 h reaction time. However, the total P parameter in biochar samples differed significantly from other parameters such as COD, TSS and total N. The total P parameter concentrations in the CF4, CF8, CF9 and CF11 samples was lower than the QCVN 40:2011/MONRE limit values, and the other samples (CF1 to CF3, CF5 to CF7, CF10, CF12) were 1.07 to 1.15 times higher than the standard limit values. The reaction time in a batch sorption process is an important parameter in determining the capacity of a sorbent. The longer the adsorption time, the higher the processing efficiency [33]. The adsorption capacity (qe ) of the pollutants changed with the reaction time as shown in Table 5. It can be clearly seen that when increasing the reaction time, the adsorption capacity for these pollutants increased as well. A similar result was found for adsorption efficiency. The result in Table 5 shows that the adsorption capacity of the COD parameter reached a maximum of 191.2 mg/g for the CF4 sample at 8 h, and a minimum of 92 mg/g for the CF12 sample at 1 h. Similarly, the adsorption capacity of

316

T. T. T. Huong et al.

Fig. 8. Changes in concentration in wastewater of (a) COD, (b) TSS, (c) total P and (d) total N parameters adsorbed with biochar (mg/L) at the reaction times of 1, 2, 4 and 8 h

TSS also reached 6.45 mg/g (CF4 sample) at 8 h and 4.3 mg/g (CF12 sample) at 1 h. Total N adsorption capacity reached as high as 12.725 mg/g when adsorbed with the CF4 biochar sample for 8 h, and as low as 3.492 mg/g with the CF3 biochar sample at 1 h. The adsorption capacity of total P is also similar to the above three parameters: the highest adsorption capacity value is 0.423 mg/g for the CF4 sample at 8 h and the lowest value observed is 0.085 mg/g for the CF12 sample at 1 h. Different biochar types exhibit different adsorption capacities for pollutants with different reaction times. However, all biochar samples in this study had an adsorption capacity which allowed removal of pollutants in the livestock wastewater sample. In the three groups of biochar materials, the group of materials pyrolysed from spent coffee grounds at 500 °C has the highest adsorption efficiency, followed by the 600 °C group and finally the 700 °C group. Moreover, this result is consistent with results obtained

Initial Results of Using Biochar Derived from Spent Coffee Grounds

317

Table 4. The adsorption efficiency (H%) of COD, TSS, total N (N) and total P (P) by 12 biochar samples H (%) CF1

CF2

CF3

CF4

CF5

CF6

CF7

CF8

CF9

CF10 CF11 CF12

COD 55.90 62.56 55.38 80.76 56.41 64.62 58.46 67.69 55.38 61.54 65.64 47.18 1h COD 85.38 84.62 86.15 88.01 80.00 86.15 85.38 80.00 85.64 87.69 85.38 62.18 2h COD 94.36 93.08 91.79 94.36 85.38 92.05 88.72 88.72 90.77 92.56 93.08 85.38 4h COD 94.36 95.38 95.38 98.08 93.08 94.87 93.08 94.87 93.85 96.92 97.69 93.08 8h TSS 1h

72.96 74.81 70.37 83.70 67.78 71.11 77.78 79.26 72.22 75.19 75.56 63.70

TSS 2h

89.63 87.04 86.30 89.63 78.52 86.30 90.37 82.96 86.30 85.93 83.33 86.67

TSS 4h

90.37 88.89 86.67 91.48 87.04 92.22 91.11 84.07 88.89 90.74 86.67 89.63

TSS 8h

90.74 93.33 93.33 95.56 91.85 93.70 91.85 92.59 90.00 92.59 92.59 90.37

N 1h

27.56 33.42 25.57 49.50 30.64 47.01 33.21 36.54 29.65 28.70 47.00 25.60

N 2h

52.82 42.06 47.70 64.04 44.01 61.86 65.32 60.50 48.83 62.35 60.50 43.88

N 4h

69.85 69.83 67.33 82.48 70.57 65.87 70.57 63.84 65.32 65.29 69.84 64.16

N 8h

71.32 73.30 74.63 93.17 75.30 71.10 74.63 71.36 71.69 69.82 82.29 69.18

P 1h

19.95 23.14 17.81 51.98 17.90 23.74 15.90 23.28 28.11 24.19 47.79 15.44

P 2h

28.75 37.86 28.75 58.31 23.69 56.54 56.40 32.67 52.35 55.08 52.48 28.29

P 4h

62.60 62.00 62.51 61.37 61.69 61.14 60.82 57.95 61.50 60.96 62.87 56.08

P 8h

75.13 73.35 65.97 77.04 68.34 76.49 68.43 72.67 73.99 76.54 72.85 68.43

109

166.5

184

184

6.05

4.925

6.1

6.125

3.764

7.214

9.540

9.741

0.110

0.158

0.344

0.412

COD 1 h

COD 2 h

COD 4 h

COD 8 h

TSS 1 h

TSS 2 h

TSS 4 h

TSS 8 h

N 1 h

N 2 h

N 4 h

N 8 h

P 1 h

P 2 h

P 4 h

P 8 h

CF1

qe (mg/g)

0.403

0.340

0.208

0.127

10.012

9.537

5.745

4.564

6.3

5.875

5.05

5.875

186

181.5

165

122

CF2

0.362

0.343

0.158

0.098

10.193

9.196

6.515

3.492

6.4

5.825

4.75

5.85

186

179

168

108

CF3

0.423

0.337

0.320

0.285

12.725

11.266

8.747

6.761

6.25

5.675

5.35

5.6

191.2

184

171.62

157.5

CF4

0.375

0.339

0.130

0.098

10.284

9.638

6.011

4.185

6.4

6.15

5.3

4.575

181.5

166.5

156

110

CF5

0.420

0.336

0.310

0.130

9.712

8.996

8.449

6.421

6.325

6.225

4.8

5.825

185

179.5

168

126

CF6

0.376

0.334

0.310

0.087

10.193

9.638

8.922

4.537

6.2

6.15

6.1

5.25

181.5

173

166.5

114

CF7

0.399

0.318

0.179

0.128

9.747

8.720

8.264

4.990

6.175

6

6.05

5.95

185

173

156

132

CF8

0.406

0.338

0.287

0.154

9.791

8.922

6.669

4.049

6

6.075

5.825

4.875

183

177

167

108

CF9

0.420

0.335

0.302

0.133

9.537

8.917

8.516

3.920

6.375

6.05

6.325

5.075

189

180.5

171

120

CF10

Table 5. The adsorption capacity (qe ) of COD, TSS, total N (N) and total P (P) by 12 biochar samples

0.400

0.345

0.288

0.262

11.239

9.539

8.264

6.419

6.45

5.975

6.25

6.1

190.5

181.5

166.5

128

CF11

0.376

0.308

0.155

0.085

9.449

8.764

5.994

3.497

6.1

6.05

5.85

4.3

181.5

166.5

121.25

92

CF12

318 T. T. T. Huong et al.

Initial Results of Using Biochar Derived from Spent Coffee Grounds

319

for the adsorption capacity of biochar by Cui et al. [33] and Hirata et al. (2002) [13, 33]. According to Cui et al. [33], the surface functional groups such as O-H, C-H, C=O, C-C, etc. on the biochar material have a strong influence on ion adsorption capacity and therefore enhance their adsorption efficiency [33]. Hirata et al. (2002) indicated that the surface area contains many functional groups such as O-H and C-O, and a specific surface area of less than 1 m2 /g will easily remove pollutants [13] due to the polarized surface and the large base consumption level of these materials. The fact that the C content is over 80% in all biochar samples shows that the adsorption mechanisms are based on the amounts of organic pollutants adsorbed and the properties of the carbonaceous material. 3.3 Effect of Adsorbent Material Mass The effect of reaction time on pollutant removal was investigated for four time values from 1 h to 8 h. The results showed that the adsorption capacity of material increased with longer reaction times from 1 h to 8 h. At a reaction time of 8 h, the highest efficiency for pollutant removal was achieved. Thus, 8 h was chosen as the optimum reaction time and used in the subsequent experiments. The initial livestock wastewater was adsorbed

Fig. 9. Changes in adsorption capacity of (a) COD, (b) TSS, (c) total N and (d) total P parameters adsorbed with biochar at the adsorbent material masses of 2, 4 and 6 g.

320

T. T. T. Huong et al.

Table 6. The adsorption efficiency (H%) of COD, TSS, total N (N) and total P (P) by 12 biochar samples H (%) CF1

CF2

CF3

CF4

CF5

CF6

CF7

CF8

CF9

CF10 CF11 CF12

COD 94.36 95.38 95.38 94.36 96.41 93.85 90.00 94.36 90.77 95.38 97.69 93.85 2g COD 98.21 97.95 97.69 97.13 97.95 94.87 94.36 94.87 94.62 96.92 97.95 97.18 4g COD 98.36 98.36 98.36 98.77 98.72 98.36 98.46 98.18 98.59 98.67 98.56 98.36 6g TSS 2g

90.74 89.63 87.41 92.59 90.74 90.00 91.11 91.48 88.89 90.74 91.85 90.37

TSS 4g

96.67 95.19 96.67 97.04 95.93 96.30 95.19 96.30 95.56 96.30 96.67 96.30

TSS 6g

96.67 95.19 96.67 97.04 95.93 96.30 95.19 96.30 95.56 96.30 96.67 96.30

N 2g

69.35 71.17 74.63 71.36 74.48 65.87 65.14 69.84 65.32 69.82 74.79 69.18

N 4g

73.30 73.30 74.97 81.92 76.03 68.06 69.13 73.30 70.71 71.11 76.66 76.05

N 6g

74.95 76.40 76.19 93.84 76.26 76.43 70.91 76.19 74.26 76.28 77.54 79.78

P 2g

62.60 64.24 65.97 68.34 59.23 62.82 56.40 61.37 54.26 62.87 63.37 63.87

P 4g

77.72 81.69 82.19 81.59 81.09 70.80 68.11 71.98 72.76 74.17 76.49 77.36

P 6g

83.19 82.82 83.92 84.65 82.28 82.92 81.82 82.78 81.50 80.59 81.05 82.23

onto biochar masses of 2, 4 or 6 g over 8 h. The experimental results are shown in Fig. 9 and in Tables 6 and 7. As with reaction time, the solid/solution ratio in a batch sorption process is an important parameter determining the capacity of adsorption of pollutants. The results of this study showed that when the adsorbent material weight was increased from 2 g to 6 g (weight/volume) with a reaction time of 8 h, the adsorption capacity/efficiency for pollutants (COD, TSS, total N and total P) increased accordingly. With different adsorbent masses (from 2 g to 6 g) and after a reaction time of 8 h, the adsorption capacity with all four parameters of the 12 biochar samples were significantly reduced (Fig. 9). The maximum reduction efficiency was found for the adsorbent mass of 6 g, with lower reduction efficiency for 4 g and lowest for 2 g. Unlike the results for different reaction times, adsorption with 6 g of material for all 12 biochar samples produced water with concentrations lower than the wastewater limit values for all four

Initial Results of Using Biochar Derived from Spent Coffee Grounds

321

parameters. The highest COD adsorption capacity was found for 2 g of the CF11 biochar sample (381 mg/g), and the lowest for 6 g of the CF8 sample (127.63 mg/g). As seen for the COD parameter, the adsorption capacity of TSS content was highest for 2 g of CF4 material (12.5 mg/g) and lowest for 6 g of adsorbent material (4.333 mg/g). The total N and P parameters showed maximum adsorption capacities of 20.429 mg/g (CF11) and 0.75 mg/g (CF4) on 2 g of biochar, and minimum adsorption capacities of 6.456 mg/g (CF7) and 0.295 mg/g (CF10) on 6 g of biochar, respectively. The analysis results showed that the removal of these pollutants increased as the weight of the adsorbent increased. This result is consistent with some previous results obtained for the adsorption capacity of biochar with different substrates, such as pesticides, heavy metal, textile dye, ions (NH4 + , PO4 3− ), etc. [4, 18, 20, 35]. Zheng et al. (2010) assessed the absorption ability for two pesticides (triazine–atrazine and simazine) by unmodified biochar. The analysis parameters included contact time, solution pH, particle size and solid/solution ratio [4]. The authors showed that adsorption affinity for the two pesticides increased from 451 to 1158 mg/kg and from 243 to 1066 mg/kg, respectively, when the solid/solution ratio decreased from 1:50 to 1:1000 (g/mL) [4]. The results obtained by Yao et al. [20] showed that nanoscale MgO (periclase) particles provided the main adsorption sites on biochar surfaces for phosphate in solution [20]. Vu et al. (2017) used a modified biochar material synthesized from corn cobs to remove ammonium from synthetic water in which the ammonium concentration varied from 10 to 100 mg/L. The research results showed that when the weight/volume ratio decreased from 1:1 to 1:5, the adsorption capacity for ammonium increased accordingly [35]. The sorption capacity of Hg(II) in water by two biochar samples synthesized from bagasse and hickory chips was assessed by Xu et al. [18]. The results showed that the sorption of Hg(II) metal can be conducted by functional group complexation on the biochar surface, ion exchange adsorption or co-precipitation with mineral [18]. Furthermore, the results of SEM images and FTIR spectra in this study also indicated that carbonaceous materials have many carboxyl and phenolic hydroxyl groups on the surface area combined with a porous structure. This property leads to a high organic pollutant adsorption capacity, influenced by the interaction between pollutants and the surface or pores of carbon materials or by interactions with functional groups [4]. Therefore, the different sorption mechanisms occurred on carbonized and non-carbonized phases of biochar [37, 38]. These results in currently study showed that the CF4 biochar sample is the highest adsorption capacity with all 4 pollutant parameters. These results of the current study showed that the CF4 biochar sample has the highest adsorption capacity for all four pollutant parameters. There is a high potential that biochar material pyrolysed from coffee grounds could become a suitable activated carbon for removal of organic pollution from livestock wastewater. However, it is necessary to fully analyze other characteristics of this material such as TEM (Transmission Electron Microscopy), sorption isotherms, the effect of particle size on sorption kinetics, etc. to find out the optimal pyrolysis conditions to produce biochar material. In addition, reaction time and the biochar mass also need to be fully studied to find out the optimal experimental conditions in which the adsorption efficiency is maximized.

368

191.5

127.87

12.25

6.425

4.35

18.945

10.012

6.824

0.687

0.427

0.304

COD 2 g

COD 4 g

COD 6 g

TSS 2 g

TSS 4 g

TSS 6 g

N 2 g

N 4 g

N 6 g

P 2 g

P 4 g

P 6 g

CF1

qe (mg/g)

0.303

0.448

0.705

6.956

10.012

19.442

4.283

6.325

12.1

127.87

191

372

CF2

0.307

0.451

0.724

6.937

10.239

20.386

4.35

6.375

11.8

127.87

190.5

372

CF3

0.310

0.448

0.750

8.544

11.188

19.493

4.367

6.4

12.5

128.4

189.4

368

CF4

0.301

0.445

0.650

6.944

10.384

20.345

4.317

6.3

12.25

128.33

191

376

CF5

0.303

0.389

0.690

6.959

9.295

17.993

4.333

6.325

12.15

127.87

185

366

CF6

0.299

0.374

0.619

6.456

9.442

17.793

4.283

6.225

12.3

128

184

351

CF7

0.303

0.395

0.674

6.938

10.011

19.079

4.333

6.25

12.35

127.63

185

368

CF8

0.298

0.399

0.596

6.761

9.657

17.843

4.3

6.225

12

128.2

184.5

354

CF9

0.295

0.407

0.690

6.945

9.712

19.073

4.333

6.375

12.25

128.27

189

372

CF10

Table 7. The adsorption capacity (qe ) of COD, TSS, total N (N) and total P (P) by 12 biochar samples

0.297

0.420

0.696

7.061

10.470

20.429

4.35

6.45

12.4

128.13

191

381

CF11

0.301

0.425

0.701

7.264

10.387

18.898

4.333

6.375

12.2

127.87

189.5

366

CF12

322 T. T. T. Huong et al.

Initial Results of Using Biochar Derived from Spent Coffee Grounds

323

4 Conclusion The results of this study showed that when increasing the reaction time and adsorbent mass, the adsorption efficiency for COD, TSS, total N and total P increased accordingly. The greatest removal of pollutants was achieved at a reaction time 8 h with adsorbed biochar weight of 6 g. It can be clearly seen that the CF4 sample performed the best in removing contaminated substances from wastewater. At a reaction time of 8 h with a biochar mass of 4 g, only the COD parameter as adsorbed by the CF4 biochar sample met the output requirements following the Vietnam national regulation QCVN 40:2011/MONRE on industrial wastewater; the other 11 wastewater samples had concentrations 1.6 to 3.6 times higher than the standard limit values. Three remaining parameters including TSS, total N and total P exhibited different biochar adsorption capacities at different reaction times: the TSS parameter met the standard limit requirements in all samples, total N was 3.3 to 4.2 times higher than its limit (except for the CF4-treated sample) and total P was 1.07 to 1.15 times higher than the standard limit values (except for samples treated with the CF4, CF8, CF9 and CF11 biochars). When an adsorbent mass of 6 g and a reaction time of 8 h were used, all four parameters adsorbed with 12 biochar samples were significantly reduced in concentration, and the output values were all lower than the standard limits. Therefore, this application can be extended to treating other types of wastewater. The results showed that the biochar from spent coffee grounds is a potential sorbent to remove pollutants from livestock wastewater. However, further studies on the optimal temperature conditions to produce biochar material, desorption/adsorption, etc., need to be fully conducted before application.

References 1. Deng, Y., Zhang, T.,Wang, Q.: Biochar adsorption treatment for typical pollutants removal in livestock wastewater: a review. In: Engineering Applications of Biochar (2017) 2. Bhatnagar, A., Sillanpää, M., Witek-Krowiak, A.: Agricultural waste peels as versatile biomass for water purification – a review. Chem. Eng. J. 270, 244–271 (2015) 3. Kong, H., He, J., Gao, Y., Wu, H., Zhu, X.: Cosorption of phenanthrene and mercury(II) from aqueous solution by soybean stalk-based biochar. J. Agric. Food. Chem. 59, 12116–12123 (2011) 4. Zheng, W., Guo, M., Chow, T., Bennett, D.N., Rajagopalan, N.: Sorption properties of greenwaste biochar for two triazine pesticides. J. Hazard. Mater. 181, 121–126 (2010) 5. Guo, Y., Tang, H., Li, G., Xie, D.: Effects of cow dung biochar amendment on adsorption and leaching of nutrient from an acid yellow soil irrigated with biogas slurry. Water Air Soil Pollut. 225, 1–13 (2014) 6. Park, J.H., Choppala, G., Lee, S.J., Bolan, N., Chung, J.W., Edraki, M.: Comparative sorption of Pb and Cd by biochars and its implication for metal immobilization in soils. Water Air Soil Pollut. 224, 1–12 (2013) 7. Murthy, P.S., Naidu, M.: Sustainable management of coffee industry by-products and value addition-a review. Resour. Conserv. Recycl. 66, 45–58 (2012) 8. Daglia, M., Papetti, A., Gregotti, C., Bertè, F., Gazzani, G.: In vitro antioxidant and ex vivo protective activities of green and roasted coffee. J. Agric. Food Chem. 48, 1449–1454 (2000) 9. Clarke, R.J., Clarke, R.J.: Coffee. Springer, Dordrecht (1987)

324

T. T. T. Huong et al.

10. Leifa, F., Pandey, A., Soccol, C.R.: Production of Flammulina velutipes on coffee husk and coffee spent-ground. Braz. Arch. Biol. Technol. 44, 205–212 (2001) 11. Figueroa, G.A., Homann, T., Rawel, H.M.: Coffee production wastes: potentials and perspectives. Austin Food Sci. 1, 1014–1018 (2016) 12. Jagdale, P., Ziegler, D., Rovere, M., Tulliani, J.M.: Waste coffee ground biochar: a material for humidity sensors. Sensors. 9, 801 (2019) 13. Hirata, M., Kawasaki, N., Nakamura, T., Matsumoto, K., Kabayama, M., Tamura, T., Tanada, S.: Adsorption of dyes onto carbonaceous materials produced from coffee grounds by microwave treatment. J. Colloid Interface Sci. 254, 17–22 (2002) 14. Tung, D.X.: An Overview of Agricultural Pollution in Vietnam: The Livestock Sector. World Bank Regional Agricultural Pollution Study Report (2017) 15. Gale, F., Marti, D.,Hu, D.: China’s Volatile Pork Industry: A Report from the Economic Research Service (2012) 16. Sang, N.: Research on the use of animal waste water by biological methods combined with membrane filtration. Hanoi National University - Hanoi University of Science (2016) 17. Chen, B., Chen, Z., Lv, S.: A novel magnetic biochar efficiently sorbs organic pollutants and phosphate. Bioresour. Technol. 102, 716–723 (2011) 18. Xu, X., Schierz, A., Xu, N., Cao, X.: Comparison of the characteristics and mechanisms of Hg(II) sorption by biochars and activated carbon. J. Colloid Interface Sci. 463, 55–60 (2016) 19. Fang, C., Zhang, T., Li, P., Jiang, R., Wang, Y.: Application of magnesium modified corn biochar for phosphorus removal and recovery from swine wastewater. Int. J. Environ. Res. Public Health 11, 9217–9237 (2014) 20. Yao, Y., Gao, B., Inyang, M., Zimmerman, A.R., Cao, X., Pullammanappallil, P., Yang, L.: Removal of phosphate from aqueous solution by biochar derived from anaerobically digested sugar beet tailings. J. Hazard. Mater. 190, 501–507 (2011) 21. Hong, T.: Report of “Area and output of coffee in the world”. Western Highlands Agriculture and Forestry Science Institute (2018) 22. Doan, T.T., Henry-des-Tureaux, T., Rumpel, C., Janeau, J.L., Jouquet, P.: Impact of compost, vermicompost and biochar on soil fertility, maize yield and soil erosion in Northern Vietnam: A three year mesocosm experiment. Sci. Total Environ. 514, 147–154 (2015) 23. Mohammadi, A., Cowie, A.L., Anh Mai, T.L., Brandão, M., de la Rosa, R.A., Kristiansen, P., Joseph, S.: Climate-change and health effects of using rice husk for biochar-compost: comparing three pyrolysis systems. J. Clean. Prod. 162, 260–272 (2017) 24. Ngo, P.T., Rumpel, C., Ngo, Q.A., Alexis, M., Vargas, G.V., de la Luz Mora Gil, M.: Biological and chemical reactivity and phosphorus forms of buffalo manure compost, vermicompost and their mixture with biochar. Bioresour. Technol. 148, 401–407 (2013) 25. Tu, T.T.: Physical and chemical characterization of biochar derived from rice husk. Hue Univ. J. Sci. 120, 233–247 (2016) 26. Mai, V.T., Tuyen, T.V.: Research on ammonium removal from aqueous solution using modified corncob-biochar by H3PO4 and NaOH. VNU J. Sci. 32, 274–281 (2016) 27. Nghia, N.K., Oanh, N.T.K., Sang, D.H., Lang, L.T.: Immobilization capacity of biochar for the pesticide Propoxur degrading bacteria, Paracoccus sp. P23-7. Sci. J. Can Tho Univ. (Vietnam) 38, 88–94 (2015) 28. Vu, T.M., Trinh, V.T., Doan, D.P., Van, H.T., Nguyen, T.V.: Removing ammonium from water using modified corncobbiochar. Sci. Total Environ. 579, 612–619 (2017) 29. Kemp, K.C., Baek, S.B., Lee, W.G., Meyyappan, M., Kim, K.S.: Activated carbon derived from waste coffee grounds for stable methane storage. Nanotechnology 26, 385602 (2015) 30. Jagdale, P., Sharon, M., Kalita, G., Maldar, N.N.: Carbon nano material synthesis from polyethylene by chemical vapor deposition. Adv. Mater. Phys. Chem. 2, 1–10 (2015) 31. Abdelkreem, M.: Adsorption of phenol from industrial wastewater using olive mill waste. APCBEE Procedia 5, 349–357 (2013)

Initial Results of Using Biochar Derived from Spent Coffee Grounds

325

32. Figueiredo, M.K.K., Caldas, K.N.C., Nascimento, B.P., Schroeder, P., Romeiro, G.A.: Use of biochar obtained from coffee grounds as adsorbent of dyes and solid biofuel. Revista Eletrônica em Gestão, Educação e Tecnologia Ambiental Santa Maria 21, 92–103 (2017) (Special Edition) 33. Cui, J., Yang, Y., Hu, Y., Li, F.: Rice husk based porous carbon loaded with silver nanoparticles by a simple and cost-effective approach and their antibacterial activity. J. Colloid Interface Sci. 455, 117–124 (2015) 34. Liu, N., Charrua, A.B., Weng, C.H., Yuan, X., Ding, F.: Characterization of biochars derived from agriculture wastes and their adsorptive removal of atrazine from aqueous solution: a comparative study. Biores. Technol. 198, 55–62 (2015) 35. Vu, T.M., Trinh, V.T., Doan, D.P., Van, H.T., Nguyen, T.V., Vigneswaran, S., Ngo, H.H.: Removing ammonium from water using modified corncob-biochar. Sci. Total Environ. 579, 612–619 (2017) 36. Liu, S.H., Huang, Y.Y.: Valorization of coffee grounds to biochar-derived adsorbents for CO2 adsorption. J. Clean. Prod. 175, 354–360 (2018) 37. Chen, B., Zhou, D., Zhu, L.: Transitional adsorption and partition of nonpolar and polar aromatic contaminants by biochars of pine needles with different pyrolytic temperatures. Environ. Sci. Technol. 42, 5137–5143 (2008) 38. Cao, X., Ma, L., Gao, B., Harris, W.: Dairy-Manure derived biochar effectively sorbs lead and atrazine. Environ. Sci. Technol. 43, 3285–3291 (2009)

Assessment the Impact of Climate Change and Sea Level Rise on the Unconfined Aquifer at the Red-River Delta of Vietnam: A Case Study at Thai Binh Province Tran Thi Thanh Thuy1(B) , Pham Khanh Huy1 , Dao Duc Bang1 , and Pham Hoang Anh2 1 Hanoi University of Mining and Geology, Hanoi, Vietnam {tranthithanhthuy,phamkhanhhuy,daoducbang}@humg.edu.vn 2 Vietnam Association of Hydrogeology, Hanoi, Vietnam [email protected]

Abstract. Thai Binh is a coastal province in the Red River Delta in Vietnam, which is bounded by river systems and coastline, causing complicated hydrogeological characteristics of aquifers including the Holocene aquifer. Groundwater plays a crucial role to water supply system in Thai Binh province. The Holocene aquifer is a heterogeneous aquifer interspersed between saltwater and freshwater zones. The total dissolved solids (TDS) of its groundwater ranges from 0.2 to 4.8 g/l. Compare to the research in 1996, the distribution area of the freshwater zone is increased 180 km2 especially in the area that the upper soil layer has good permeability. However, groundwater resources in this province are highly vulnerable to human activities, climate variation and sea-level rise. Insights into impacts of climate variation and sea-level rise is an essential task to sustainable groundwater management in this area. The present study aims to investigate potential impacts of climate change and sea-level rise on the groundwater by using a three-dimensional transient density-driven groundwater flow model (the SEAWAT package) based on the impacts of the groundwater exploitation activity, rainfall, river and sea-level rise. The results obtained revealed that the groundwater table in Holocene aquifer in 2100 was decreased from 0.5 to 0.8 m depending on different locations in the study area. The most affected area ranges within 3.0 km from the shoreline. The areas of the Holocene aquifer with saline water, especially coastal areas such as Tien Hai and Thai Thuy, will increase as seawater intrusion intensifies. The increase in seawater intrusion will vary according to each climate change and sea-level rise scenario. The predicted increase is 79.9 km2 for Scenario B1, 94.1 km2 for Scenario B2, and 109.7 km2 for Scenario A2, whereas the remaining freshwater reserves in the Holocene aquifer for each scenario are 513.243, 505.282, and 492.443 million m3 , respectively. This paper concludes with proposed solutions for sustainable groundwater usage in this region. It is our hope that these results will contribute to Vietnam’s sustainable development by providing necessary information for resource managers to control groundwater usage, reduce pollution, limit saline intrusion, and save natural resources. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 326–348, 2021. https://doi.org/10.1007/978-3-030-60269-7_17

Assessment the Impact of Climate Change and Sea Level Rise

327

Keywords: Aquifer · Climate change · Sea-level rise · Intrusion · Vietnam

1 Introduction In recent years, climate change and sea-level rise are widely recognized as the most severe issues causing many negative impacts on the environment, economy and society. Along the shoreline impacts include beach erosion and marine inundation resulting from the increased frequency and magnitude of waves. For inland areas, include reduced infiltration, drainage, and saltwater intrusion [1]. The field of water resources was concerned as one of the important research issues. Especially in areas with severe conditions, such as coastal areas where freshwater sources play an essential role in life and society. While climate change affects surface water resources directly through changes in the major long-term climate variables such as air temperature, precipitation, and evapotranspiration, the relationship between the changing climate variables and groundwater is more complicated and poorly understood. The greater variability in rainfall could mean more frequent and prolonged periods of high or low groundwater levels, and saline intrusion in coastal aquifers due to sea-level rise and resource reduction. Groundwater resources are related to climate change through the direct interaction with surface water resources, such as lakes, rivers, and sea and indirectly through the recharge process. The literature reviews showed that some studies evaluated the effect of climate change and sea level rise to groundwater at coastal area [2–11]. Climate change researchers have assessed the restoration of groundwater levels and the intrusion of seawater in groundwater sources, including the unconfined Holocene Aquifer. The studies show that the changing quality and quantity of groundwater depends on geological, topographic, hydrogeological, and climate conditions, the geological evolution and socio-economic history of each region, and sea-level rise [12–14]. Because groundwater in aquifers is recharged mainly by precipitation or interactions with surface water bodies, the direct influence of climate change on precipitation and surface water will ultimately affect groundwater systems [15]. The changing of salt of groundwater was influenced by seawater and rainwater [2, 16, 17]. Assessing the impact of climate change and sea level rise on groundwater requires a combination of advanced knowledge and innovative methods. Methods applied to studies of sea-level rise and seawater intrusion include numerical simulation and analytical approaches [5, 18]. Monitoring wells along the coast are used to analyze the radius of sea-level rise to determine its impact on saltwater intrusion [2, 4–6, 12]. Evaluating the impact of climate change and sea-level rise on groundwater resources is generally based on simulations of groundwater recharge, groundwater levels, surface-groundwater interactions, and seawater intrusion [5, 6]. Besides, changes in the quality and volume of groundwater were also simulated using a 3D numerical model based on climate change scenarios, using key factors such as rainfall, sea level, groundwater level and exploitation data. Most of the researches use mathematical models and the salinity - light shift mechanism. Groundwater flow is adapted to take into account density differences between fresh, brackish and saline groundwater. Previous studies have also used this code to simulate and quantify future salinization processes in the subsurface, calculate

328

T. T. T. Thuy et al.

the effects of future climactic and anthropogenic scenarios, and devise measures to combat salinization [7–9, 16–24]. Base on the Intergovernmental Panel on Climate Change’s assessment, Vietnam is one of the countries that will be affected seriously by climate change and sea-level rise. Climate change impacts on Vietnam are serious and threaten sustainable development goals, such as poverty reduction and the realization of millennium goals. Sea level rise in the coastal areas of Vietnam will be higher than the average global sea level rise [25]. In Vietnam, some research was carried out to assess the impact of climate change and sea-level rise on the groundwater in coastal regions. Vu Thanh Tam et. al (2016) studied and assessed the impact of climate change on a quaternary groundwater system on the narrow sloping coastal plain of Central Vietnam [15]. Nguyen Thi Ha (2016) assessed the saltwater intrusion into the river and groundwater by climate change in Hau Loc, Thanh Hoa province, and Tan Thanh, Ba Ria – Vung Tau province [26]. It is obvious that climate change and sea-level rise have highly potential impacts on groundwater resources in coastal areas. However, the impacts of these phenomena are still high uncertainty depending not only on regional climate system but also specific geographical and geological conditions as well human activities of a certain area. In the Red River Delta plain, Thai Binh is one province that will be heavily influenced by sea-level rise. Challenges, in particular, include water shortage, scarcity of freshwater, and saltwater intrusion. Therefore, studying their impact on the Holocene aquifer is is necessary to anticipate, prevent, minimize, or mitigate adverse effects.. This research emphasized the need to development management models that simulate seawater intrusion in the Holocene aquifer, assess future compound groundwater challenges, and provide solutions, protective measures, and sustainable exploitation methods.

2 Study Area and Data 2.1 Description of the Study Area Thai Binh is a coastal provinces in the Red River Delta plain with an area of 1,542.24 km2 . Thai Binh experiences two seasons with a mean annual rainfall ranging from 1,600 to 1,800 mm. The rainy season lasts from April to September, and the dry season lasts from October to March. Annual evaporation is about 600 mm. According to statistics, the average temperature has increased slightly from 22.8 to 23.6 °C since 1996. This has additionally contributed to an increase in the number of storms. The mean annual maximum temperature is 40.2 °C, while the minimum is 7 °C. Thai Binh has 52 km of coastlines, which is directly affected by seawater and tides, leading to the possibility of saltwater intrusion inland. The unconfined aquifer, known as the Holocene Aquifer, is distributed throughout the study area and is the main water source for production and other domestic uses. It also plays a vital role in maintaining the ecological value of the area (Fig. 1). 2.2 Data Used Topography of Thai Binh is low plain, flat. The average elevation is 1–2.5 m above the sea level and lower in the East. The studying area has two main aquifers that are Holocene

Assessment the Impact of Climate Change and Sea Level Rise

329

Fig. 1. Map of Thai Binh province

(qh) and Pleistocene (qp) aquifers. These aquifers were separated by the clay layer of Hai Hung and Vinh Phuc formation. The Holocene aquifer is distributed and exposed to most areas of the province. The distribution of aquifers and aquitards is shown in Fig. 2. Upper Holocene unconfined aquifer (qh2 ) is the groundwater in Quaternary sediments, has the sources from the sea, sea - wind, and river sea origin. This aquifer distributes throughout the province is about 1,200 km2 .

Fig. 2. Hydrogeological cross section in Thai Binh province [27]

It has mainly composed of sand with little silt, capable of absorbing and circulating water good. The thickness increases from the North – North West to the South – South East, with a maximum is 25 m. The groundwater level is from 1 to 2 m depending on the seasons. According to the pumping and hydrogeological test shown that well capacity is from 0.1 to 0.7 l/s, average transmissibility coefficient is 1.8 m2 /day, hydraulic conductivity is 1.49 m/day, and storage coefficient is 0.18. To assess the recharge capacity of rainwater and surface water to Holocene aquifer, the parameters of surface coating and hydrogeological characteristics were investigated and summarized in Table 1.

330

T. T. T. Thuy et al. Table. 1. The parameters of surface coating

Location

Area, km2

K, m day−1

P, mm year−1

Tv, mm year−1

Cv

I, mm year−1

Tan Hoa, Hung Ha

80

0.005

1,680

845

0.62

134.4

Minh Tan, Dong Hung

30

0.004

1,695

840

0.63

135.6

Bach Thuan, Vu Thu

40

0.003

1,740

830

0.65

139.2

Song An, Vu Thu

30

0.070

1,740

830

0.25

104.4

Quang Trung, Kien Xuong

25

0.150

1,750

830

0.26

140.0

Tay Phong, Kien Xuong

30

0.300

1,750

845

0.25

140.0

Nam Hung, Tien Hai

32

0.220

1,770

855

0.22

141.6

Thuy Luong, Thai Thuy

35

0.007

1,780

860

0.24

142.4

An Thai, Quynh Phu

84

0.009

1,700

840

0.45

136.0

Thuy Truong, Thai Thuy

46

0.015

1,780

860

0.43

142.4

Bach Thuan, Vu Thu

150

0.025

1,740

830

0.3

139.2

Tay An, Kien Xuong

330

0.030

1,750

850

0.3

140.0

Dong Huy, Dong Hung

130

0.020

1,695

840

0.28

135.6

Nam Hung, Tien Hai

120

0.028

1,770

855

0.22

141.6

Hung Dung, Hung Ha

280

0.020

1,680

845

0.47

134.4

Dong Tien, Quynh Phu

100

0.050

1,700

840

0.48

136.0

The quality of Holocene aquifer in this region is not uniform: it is interspersed between fresh and saline groundwater zones. The thickness have affected the complementary roles of rain, rivers, and sea, changing the salt - freshwater boundary of groundwater. To assess the relationship of the qh aquifer with climate and hydrological factors, we have analyzed concentrations of Na+ , Cl− and TDS. Total dissolved

Assessment the Impact of Climate Change and Sea Level Rise

331

solids is considered as another indicator of the amount of contaminant present in the groundwater which is directly proportional to EC [28]. According to the standard QCVN 01:2009/BYT, TDS between 0 g/l to 1 g/l is freshwater. If TDS more than 1,000 mg/l is poor for drinking use and unacceptable if it is greater than 2,000 mg/l. In whole province, TDS changed from 0.3 g/l to 18.3 g/l. The concentration of Cl− changed from 72.0 mg/l to 1,569 mg/l. The concentration of Na+ in the water varied widely, and many places had a concentration higher than the standard. The area has medium permeability coefficient, high flow rate of groundwater and the component is mainly composed of fine to medium coarse sand, gravel that is salinely cleaned by rainwater and surface water. The freshwater area has TDS that is 0.3 to 0.8 g/l and mainly distributed in Hung Ha district, Vu Thu district, a part of Thai Binh city, and the riverside of Tien Hai district, Thai Thuy district. The chemical composition of qh2 aquifer in fresh areas has Mg - Na – Ka – Cl type and is shown in the following formula [27]: M0.5

3 Cl60 HCO37 pH8.5 Mg39 (Na + K)35 Ca26

(1)

In the area which has low permeability coefficient, a low flow rate of groundwater and the component of soil is mainly composed of silt and clay, the groundwater flow rate is less related to river water so that the groundwater is not salinely cleaned. We can divide this aquifer into two representative areas saline based on the TDS value that is Quynh Phu – Dong Hung area and Hong river – Tra Ly river with one part of Vu Thu district. The chemical composition of qh2 aquifer in saline areas has Na – Ka – Cl type and is shown in the following formula [27]: M11.15

Cl92 pH8.1 (Na + K)88 Mg10

(2)

Surveying results of the TDS content of groundwater in Holocene aquifers are shown in Fig. 3, showing the current status of the distribution of salty and freshwater zones of this aquifer. In particular, the freshwater area is about 521.1 km2 and mainly distributed in the central and the southeast of the study area. These freshwater zones were located in Hung Ha, Dong Hung Quynh Phu, and Kien Xuong districts. Compared with the previous research results in 1996 by Lai Duc Hung et al., the distribution area of these freshwater bodies has changed quite a lot. The area of the saltwater bodies is narrowed 180 km2 , mainly in the coastal area of Tien Hai and Thai Thuy districts. The over exploitation through excessive abstraction of groundwater resource exceeding its replenishment capacity is the cause of lower water levels and saline intrusion. So the study has surveyed and assessed the status of using groundwater of local people. The exploitation quantity of groundwater in this area is nearly 300,000 m3 /day. According to the statistic data of Thai Binh Department of Natural resources and Environment and field survey result, the whole province had 216, 926 wells funded by UNICEF with capacity about 200,000 m3 /day, 100 industrial exploitation wells with exploitation capacity about 20,852 m3 /day, and 6 groundwater treatment plants. Aside from, there are many well that is self – exploitation and treatment by households in this area (Table 2) [29].

332

T. T. T. Thuy et al.

Fig. 3. Distribution salt-freshwater boundary of Holocene aquifer in 1996 and 2017

Table. 2. The currently status of exploitation groundwater and the population using clean groundwater in Thai Binh province No

District

Population

Private well

Dug well

Water supply system, m3 .day−1

The population using clean groundwater

%

1

Thai Binh

66,937

13,250

10,145

30,000

53,395

80

2

Kien Xuong

237,410

38,460

37,895

8,001

84,356

35

3

Thai Thuy

259,629

53,975

47,675

9,996

111,646

43

4

Dong Hung

252,220

76,000

34,015

6,003

116,018

46

5

Hung Ha

247,349

72,150

40,315

4,007

116,472

47

6

Quynh Phu

240,013

74,200

48,655

6,240

129,095

54

7

Tien Hai

206,874

27,040

34,930

5,172

67,142

32

8

Vu Thu

228,726

36,850

43,960

4,003

84,813

35

Assessment the Impact of Climate Change and Sea Level Rise

333

The exploitation process is caused making the changes of groundwater, which has been making groundwater level of this aquifer go down, and changing the distribution of salty and freshwater zones of this aquifer. Hence, this study will be the basis for forecasting the changes in the quantity and quality of Holocene aquifer under the impact of climate change and sea-level rise on the future. To evaluate the impact of climate change, sea-level rise on quality changes and groundwater reserves in the study area and forecast the impact of climate change and sealevel rise on groundwater in the future need to assess the relationship between groundwater and seawater, rainwater. Most of the groundwater resources are mainly renewed (recharged) directly from rainwater i.e., precipitation through infiltration into the saturated zone, thereby maintaining the recharge potential of an aquifer seems essential for the sustainability of that aquifer. Therewith, the climate change and sea-level rise scenarios for Thai Binh province is essential to assess the seawater intrusion on groundwater [25]. The simulation input data is based on the result of predicted and built scenarios with climate change and sea-level rise by the Ministry of Natural Resources and Environment, Vietnam. Climate change and sea-level rise scenarios are assumed belong to the increase of greenhouse gas emissions that leading to an increase in temperature, humidity, precipitation and sea level in the future. Climate change and sea-level rise scenarios for Vietnam are formulated based on different assumptions of greenhouse gas emissions: low emission (B1), medium emission (B2), and high emission (A2) [25]. In particular, the annual average temperature in Thai Binh province under the A2 emission scenario will increase about 1.3 °C in 2050 and 3.1 ºC in 2100 compared to the year (1980–1999) [30]. Table. 3. Scenarios of increasing rainfall and sea-level rise compared to the year (1980–1999) in Thai Binh province [30] No

Year

Rainfall, %

Sea level rise, cm

B1

B2

A2

B1

B2

A2

1

2030

2.0

2.1

2.3

10 ÷ 12

11 ÷ 12

11 ÷ 13

2

2040

2.8

3.0

3.3

14 ÷ 17

15 ÷ 17

16 ÷ 18

3

2050

3.6

3.9

4.1

19 ÷ 22

20 ÷ 24

22 ÷ 26

4

2060

4.2

4.7

4.9

23 ÷ 29

25 ÷ 31

29 ÷ 35

5

2070

4.5

5.5

5.8

28 ÷ 36

31 ÷ 38

38 ÷ 46

6

2080

4.8

6.2

6.9

33 ÷ 43

36 ÷ 47

47 ÷ 58

7

2090

4.9

6.8

8.1

38 ÷ 50

42 ÷ 55

56 ÷ 71

8

2100

4.9

7.4

9.4

42 ÷ 57

49 ÷ 64

66 ÷ 85

In addition, three sea-level rise scenarios will be set up, including the sea-level will rise up to 50 cm, 70 cm and 85 cm. Then forecasting the areas that is risk by climate change and sea-level rise. The report of the Ministry of Natural Resources and Environment, Vietnam is shown the risk of flood in coastal areas in Thai Binh province [30]. So that,

334

T. T. T. Thuy et al.

this research estimated the flood area under three types of sea-level rise scenarios: a sea-level rise of 0.5 m per century, 0.7 m per century and 0.85 m per century. The impact of sea-level rise has changed the land use area by the increase of flood. Wetland areas in different regions under the climate change scenarios are shown in Table 4. Table. 4. Scenarios of Flood area in Thai Binh province [30] District

Area (km2)

Scenario of Flood area (km2)

Percent Flood area (%)

50 cm

70 cm

85 cm

50 cm

70 cm

85 cm

Tien Hai

195.97

58.79

117.58

156.78

30

60

80

Thai Thuy

258.77

38.82

90.57

168.20

15

35

65

Kien Xuong

215.51

32.33

43.10

86.20

15

20

40

Quynh Phu

208.40

10.42

20.84

52.10

5

10

25

Ðong Hung

205.83

6.18

14.41

20.580

3

7

10

Hung Ha

212.93

4.26

10.65

17.04

2

5

8

Vu Thu + Thai Binh city

244.84

19.59

29.38

36.73

8

12

15

Province

1,542.24

170.37

326.53

537.62

11.05

21.17

34.86

In the scenario of a sea-level rise of 0.85 m by 2100, the area of the flooded land will account for 34.86% of the total area of the province, and the coastal areas are the places that will be most affected. This will push saltwater going far into the land and penetrate the aquifer.

3 Background of the Methods Used 3.1 Monitoring the Water Level Monitoring the groundwater level, seawater level, and TDS of groundwater at coastal areas where seawater intrusion was report or a seawater intrusion risk was high. The groundwater monitoring well was located at 1.5 km to 3.5 km from the coastal line and close to the tide gauge station that to investigate seawater intrusion for this study. The observed data were used for this study to assess the relationship of groundwater and river water, seawater. After that, the recharge of river water and seawater to dissolution and diffusion of the saltwater into the aquifer is forecasted.

Assessment the Impact of Climate Change and Sea Level Rise

335

3.2 Physically Based Model A based model of a groundwater system under possible climate change based on available data is very important to prevent the deterioration of regional water-resource problems in the future. Although uncertainties are inevitable, new response strategies in water resource management based on the model may be useful. Based on the physical and chemical characteristics are researched and the features of the models, using Visual Modflow model and Seawat code to predict the effect of climate change and sea-level rise to the qh aquifer. The modeling approach was validated by solving Henry’s steady-state solution. Model describes the groundwater flow system with the river and coastal boundaries, topographic condition, hydrogeology parameters (hydraulic conductivity, specific yield, groundwater level), exploited condition, land use and precipitation, evaporation, sea-level rise. The model is established based on hydrogeological characteristics of the study area and the hydrogeological structure of the whole Northern Delta. Structure model consists of 04 layers: (1) Surface layer having low permeability; (2) Holocene aquifer (qh1, qh2), (3) Weak permeability layer separate the Holocene aquifer and the Pleistocene aquifer, and (4) Pleistocene aquifer (qp). The area of the simulation model of the study area is 1,554 km2 consisted of a rectangular grid of 98 rows and 108 columns with grid cells size 500 m x 500 m. Topographic data were obtained from the digital elevation model (DEM). Hydrogeological parameters (hydraulic conductivity, specific yield, groundwater level) were based on the results of previous researches and survey results. Amount of water recharged for the aquifer was determined by rainfall which was given in Table 5 and Fig. 6. The three types of boundary conditions used in the model are river boundary, general head boundary (GHB), and sea boundary (H = const). After the simulation model is adjusted according to the actual current situation, it will be used to predict the changes in groundwater dynamics of the Holocene aquifer under 3 climate change scenarios A2, B1 and B2 were shown in Tables 3 and 4. Coastal and river boundary conditions will be changed by the effect of climate change and sea-level rise both value and position on nodes. In which, with sea boundary the position and water level on the boundary position were determined based on the flood map and the change of sea level over the years in each climate change scenario. For river boundary, there is only change in water level at the nodes. The amount of water recharged for groundwater was determined based on the rainfall for each scenario. Thence, the model was calibrated to observations of groundwater altitude and contrasted in salinity detected in boreholes by using data collected as part of our ongoing investigation. A three-dimensional transient density-driven groundwater flow model, considering open boundary conditions for coasts and a sharp interface between freshwater and saltwater, was applied to the aquifer under steady-state conditions for freshwater surplus and deficits at the coastline. When recharges of saltwater occur at the coastline, essentially of freshwater deficits, a hypothesis of mixing for the freshwater-salt water transition zone allows the model to calculate the resulting seawater intrusion in the aquifer. Hence, adequate treatment and interpretation of the hydrogeological data, which are available for the coastal aquifer, were of main concern in satisfactorily applying the proposed numerical model. The results of the steady-state simulations showed

336

T. T. T. Thuy et al.

reasonable calculations of the water table levels and the freshwater and salt-water thicknesses, as well as the extent of the interface and seawater intrusion into the aquifer for the total discharges or recharges in the delta and along the coastline. From the modeling application on Thai Binh province, it appears that, on an annual basis, most tested scenarios predict a decrease in groundwater levels and reserves in relation to variations in climatic conditions. So that, the map of the groundwater level of the aquifer and the map of the distribution of salt and freshwater zones in Holocene aquifer are established in the model. The model results are proper groundwork that helps to calculate the quantity of this aquifer and show the impact of climate change and sea-level rise to the Holocene aquifer in this study area.

4 Proposed Methodology for the Assessment the Impact of Climate Change and Sea Level Rise on the Unconfined Aquifer in This Research Step 1: Data collection and processing To research and forecast the saltwater intrusion of Holocene aquifer under the climate change and sea level rise, the combinatorial hydrogeology methods including outdoor and indoor were used. Firstly, collecting the data from the previous research by Lai Duc Hung in 1996 about the distribution salt - fresh water boundary of Holocene aquifer, the hydrogeology and geology characteristics of the aquifers, distribution of aquifers and division of hydrological geological structure [27]. Thence, the study designed a survey roadmap perpendicular to the saline boundary that is surveyed in 1996 to re-correct the current saline boundary (Fig. 3). Besides that, a field survey was conducted to collect data about the currently status of exploitation, using groundwater for people in the districts and direction of using groundwater in the future. Especially, to evaluate the impact of climate change, sea-level rise on quality changes and groundwater reserves in the study area and forecast the impact of climate change and sea-level rise on groundwater in the future need to assess the relationship between groundwater and seawater, rainwater. Most of the groundwater resources are mainly renewed (recharged) directly from rainwater i.e., precipitation through infiltration into the saturated zone, thereby maintaining the recharge potential of an aquifer seems essential for the sustainability of that aquifer. Thence, the data of precipitation, river water level and seawater level were collected from the Vietnam Center of Hydro - Meteorological Data and the groundwater level data of monitoring wells (Q155, Q156, Q158, Q159) was collected from NAWAPI. Our observed results were combined with other observation results of NAWAPI to assess the relationship of the tide, river water level and groundwater level. Similar, the data about the status of land use and direction of using land in the future was collected to assess the recharge of rainwater and surface water to the aquifers. Final, collecting the climate change and sea-level rise scenarios by the Ministry of Natural Resources and Environment, Vietnam is necessary to assess the impact climate change and sea-level rise on groundwater in this province. Step 2: Field survey and hydrogeological test First of all, field survey to study the geological structure to clarify the formation of sediments, lithological composition, and hydrogeology characteristics of Holocene

Assessment the Impact of Climate Change and Sea Level Rise

337

aquifer in this study area. The permeability coefficient of the topsoil layer was determined by double ring infiltrometer test in the excavation hole and combined with petrographic analysis of the core sample in the laboratory. The location of the field test points and the drilling positions for lithological sampling were arranged evenly and alternatively together. However, the double ring infiltrometer test is arranged more often in riverside areas to evaluate the replenishment capacity of surface water for aquifers or vice versa. The total number of in situ testing and lithological sampling was 55 points. Step 3: Sampling and in situ measurement To assess the impact of climate change and sea level rise in qh aquifer, the changing of saltwater quality over time had to study. In this research, the authors carrying out survey, observation, rapid measurement method and sampling to evaluate the water quality at the 98 boreholes in the rainy and dry seasons over the studying area. These survey points located in the area that is around the salt-light boundary defined in 1996 by Lai Duc Hung and in the riverside and coastal areas to correct the salt-light boundary at the time of the study. This data is used as the basis for forecasting the shift boundaries under the impact of climate change and sea-level rise in the future. Locations of observation points are shown in Fig. 4. Step 4: Monitoring the groundwater level, river water and sea level over time in the study area to assessment the relationship each other. Collecting the data of monitoring the groundwater level from the NAWAPI to study the fluctuation of water level in some boreholes in whole province to assess the complementary of rainwater and surface water to Holocene aquifer over time. Besides that, carry out monitoring of water level fluctuations in the 15 boreholes at the coastal areas (Thai Thuy, Tien Hai) to assess the relationship and effect of tide on the Holocene aquifer (Fig. 4). Step 5: Groundwater sample analysis in laboratory After that, using the analysis sample method in laboratory to assess the quality of groundwater. The number of samples was analyzed is about 30 samples. The quality of groundwater results and the distribution of TDS are used to build the salt - fresh water boundary in the study area (Fig. 3). It is the basis for forecasting the movement of this boundary under the impact of climate change and sea level rise. Step 6: Summary and calculation Calculation and prediction of the recharge of rainwater, surface water for qh aquifer based on the results of field surveys, observation, sampling and analysis data in this study. The groundwater recharge is calculated as [17]: R = P − Sv − Tv − I

(3)

where P is the precipitation [L], I is the interception [L], Sv is the surface runoff [L], Tv is the actual transpiration [L] and R is the groundwater recharge [L] Sv = Cv (P − I)

(4)

where Cv is the surface flow coefficients depend on soil composition, vegetation cover and topographic slope.

338

T. T. T. Thuy et al.

Then, using a three-dimensional transient density-driven groundwater flow model by the SEAWAT package to simulate saltwater intrusion in a coastal aquifer for three types of sea-level rise scenarios: a sea-level rise of 0.5 m per century, 0.7 m per century, and 0.85 m per century to assess the impact of sea-level rise on groundwater. Finally, the quantity of fresh groundwater will be estimated under the scenarios in the future.

Fig. 4. Map showing the sampling locations for groundwater of Holocene aquifer, in situ measurements and monitoring the groundwater level in coastal area

5 Results and Discussion 5.1 Groundwater Flow Dynamics The results of monitoring the water level in the Holocene aquifer, seawater level, and rainfall in Fig. 5 shows that the trend of the groundwater level is rise up and similar to the change of the rainfall in the study area. This shows that rainwater has a close relationship with groundwater and has an influence on the formation of water storage of this aquifer. The groundwater level in Holocene aquifer is increased over time. From the above results, we can clearly see that the rainwater directly affects the quantity of groundwater and implement this aquifer. In this study, the effects of climate

Assessment the Impact of Climate Change and Sea Level Rise

339

Fig. 5. The changing of rainfall, groundwater level and sea level over time in the study area

change on the groundwater system of the catchment are assessed based on changes in the recharge and baseflow. Studies on the effects of rainfall intensity on groundwater recharge are the basis for calculating flow simulations and assessing changes in groundwater quality [31]. Therewith, the quality of water in the Holocene aquifer is changing. Rainwater and surface runoff have accelerated the process of saline washing for this aquifer, especially in a coastal area where the upper layer has good permeability. This result is similar to the previous research (Mzila et al., 2003) that studies the unconfined aquifers in coastal areas of Singapore. Saline intrusion intensity is inversely proportional to the rainfall [19]. The rising of the groundwater table will reduce the saline intrusion process of the saltwater mass. Based on Eqs. (3) and (4), the amount of rainwater for groundwater has been calculated as shown in Table 5 [32]. According to the calculation results, the amount of recharge for aquifers varies depending on the characteristics of each area. The amount of water recharge to the Holocene aquifer is an average total about of 26% of the total precipitation [33]. The total amount of water replenished is about 345,460 m3 /day. According to the results of calculating the amount of rainwater recharges to groundwater, the map of the distribution recharge area of rainwater to groundwater was built. This is an essential condition to put into the Seawat model to predict the recharge of rainfall to groundwater in the future under the climate change and sea-level rise scenarios. The study of the impact of sea-level rise on groundwater is important to study the saline intrusion at the coastal areas [10, 34]. Based on the results of the field surveys, groundwater level monitoring at the coastal area of Tien Hai, Thai Thuy, the study has built a hydraulic relationship between seawater level and groundwater level of Holocene aquifer at different distances of 1.5 km, 3.0 km and 3.5 km which is shown in Fig. 7.

340

T. T. T. Thuy et al. Table. 5. The amount rainwater recharges to Holocene aquifer

Location

Area, km2

Sv, mm year−1

R, mm year−1

Ratio of concreted areas

Tan Hoa, Hung Ha

80

517.7

182.9

0.5

5.5

Minh Tan, Dong 30 Hung

538.65

180.75

0.52

5.1

Bach Thuan, Vu 40 Thu

591.5

179.3

0.45

5.7

Song An, Vu Thu

30

227.5

578.1

0.45

24.6

Quang Trung, Kien Xuong

25

239.2

540.8

0.4

24.7

Tay Phong, Kien Xuong

30

226.25

538.75

0.4

24.3

Nam Hung, Tien Hai

32

201.3

572.1

0.3

25.5

Thuy Luong, Thai Thuy

35

220.8

556.8

0.3

25.0

An Thai, Quynh 84 Phu

382.5

332.3

0.4

14.8

Thuy Truong, Thai Thuy

352.6

333

0.2

15.9

Bach Thuan, Vu 150 Thu

273.0

497.8

0.45

20.0

Tay An, Kien Xuong

330

270.0

490

0.4

20.2

Dong Huy, Dong Hung

130

239.4

480

0.5

19.8

Nam Hung, Tien Hai

120

201.3

572.1

0.38

20.0

Hung Dung, Hung Ha

280

392.45

308.15

0.45

10.6

Dong Tien, Quynh Phu

100

412.8

311.2

0.45

10.1

46

Percent supplemented by rainwater, %

Assessment the Impact of Climate Change and Sea Level Rise

341

Fig. 6. The distribution recharge area of rainwater to groundwater

Fig. 7. The changing of the groundwater level at the 1.5 to 3.5 km distance far from the sea

As assessed about the effect of seawater to Holocene aquifer at different distances, the graph of the groundwater table by different distances from the coast and the sea level shows the fluctuation of the underground water level in sync with the sea level by the tide. At the distance of 1.5 km from the sea, the groundwater level in the observation borehole

342

T. T. T. Thuy et al.

155 is rising up and fluctuating the same with the seawater level. And in borehole QT 1–5 at 3.0 km is the similar to Q155. This result is the same as the research of Hoover, D.J et al., 2017 which assess the effect of the tide on the groundwater in coastal at Malibu, California. The research built a relationship between the groundwater levels and observed tide at Santa Monica. Note that well and tide data are plotted with different scales to facilitate visualization of tidal response in wells are 60 m, 65 m and 115 m from the ocean, respectively [11]. More recent data show that the significant damping of the tidal happened with the respective distance from shore. The further inland, the less impact of the tide to groundwater. Moreover, this also shows that the groundwater of Holocene aquifer has a hydraulic relationship with the sea. The change of the water table under the fluctuation of the tide decreases when going far inland. The area where the Holocene aquifer is most affected by tides is within 3.0 km of the coastline. The relationship between sea level by tide and groundwater is the linear relationship shown in Fig. 8 and is shown by Eq. 3. y = 0, 3254 x + 0, 00946

(5)

Fig. 8. The relationship between seawater level and groundwater level of Holocene aquifer

5.2 Simulation of Changes in Holocene Aquifer According to Climate Change Scenario The principal focus of climate change research with regard to groundwater has been on quantifying the direct impacts of changing precipitation, temperature and sea level rise. It is generally known that climate change may have impacts on sea-level rise and precipitation, which may be associated with freshwater recharge rates and that seawater intrusion is influenced by not only sea level but also freshwater recharge rates. While a higher freshwater recharge rate could lower salinity in groundwater, a higher sea level

Assessment the Impact of Climate Change and Sea Level Rise

343

may increase seawater intrusion [23]. Such studies have used a range of modeling techniques such as soil water balance models, empirical models, conceptual models and more complex distributed models. Changes to the groundwater system in response to sea-level rise at this site may be of concern to those of continued coastal evolution and shoreline change considering the periodic climate change overwash that currently affects groundwater flow in this area using Visual Modflow model to describe the groundwater flow system in Thai Binh province with the river and coastal boundaries, topographic condition, hydrogeology parameters. The new steady state condition for the salt wedge simulated by the model after raising the sea level. After that, the map of the groundwater level of the qh aquifer and the map of the distribution of the supplement area are established in the model. According to climate change and sea-level rise scenarios for Thai Binh province, this model can be simulated the groundwater flow and forecast the change of groundwater level over time (Fig. 9).

(a)

(b)

Fig. 9. The groundwater level of Holocene aquifer over time, (a) - The groundwater level in 2017, (b) - The groundwater level in 2100 with A2 scenario

According to the simulation results in Fig. 9, the water level of the Holocene aquifer until 2100 has changed compared to 2017. Results obtained revealed that the groundwater table in Holocene aquifer in 2100 was decreased from 0.5 to 0.8 m depending on different locations in the study area. The water level at the seaside boundary will abruptly increased from 1 to 2 m to simulate an instantaneous sea-level rise in the future. After that, the map of saline intrusion into river system, and the map of the distribution of the supplement area are established in the SEAWAT model. (Fig. 10). Later, the simulations transient effects of the sea-level rise on saltwater wedge in Holocen aquifer with vertical sea-land interface by this model. This method has been used by many studies in the world. Besides the raising of sea level based on the scenario, another condition of the model is that the rainfall in the studying area will be increased over time. As the results of the TDS content of Holocene aquifer, the area of freshwater zones along the Tra Ly River in Dong Hung, Kien Xuong, and Tien Hai districts and in

344

T. T. T. Thuy et al.

the center of studying area are expanded. In the study about the relationship of seawater with the qh aquifer, it shows that the influence area of seawater is within 3.0 km from the coast. Due to the climate change and sea level rise, the qh aquifer is affected by the saline intrusion from the sea to inland.

Fig. 10. Simulation of saline intrusion into river system in Thai Binh province by SEAWAT model with A2 scenario

Modeling result showed that qh aquifer is affected by seawater intrusion, especially in coastal areas such as Tien Hai and Thai Thuy districts. The area of saltwater in this area will be expanded by 20% compared to 2017. Furthermore, it also gives out the boundaries between saline and fresh water in the future will be changed so much. The salt-and-fresh area of qh aquifer tends to be narrowed down but not significantly. Forecast to 2100, the area of saltwater in Holocene aquifer will be increased respectively with scenarios B1, B2, A2 is 79.9; 94.1 and 109.7 square kilometres. The mixing zone between fresh and saline water in Holocene aquifer will be moved into inland 1,700 m respectively with sea-level rise is 0.85 m per century. Figure 11 is shown the distribution of saltwater boundary of Holocene aquifer with A2 scenario. The research result is similar to some studies in the world. In which, the research of Oude Essink et al., 2010 assessed the effect of climate change to Holocene aquifer at the coast of Netherlands. The results of this study showed that the sea level rise about 0.5 m per century will increase the salinity in

Assessment the Impact of Climate Change and Sea Level Rise

345

the Holocene aquifer within 10 km from the shoreline [6]. According to the research of Sherif et al., 1999, the sea level in the Mediterranean Sea increases by 50 cm, the saline boundary in the aquifer Holocen in the Nil Delta (Egypt) moves further inland than 9 km. As for the sea level in the Bay of Bengal (India) will be increased by 50 cm, the saline boundary of the aquifer only moved deep into the continent 0.4 km [20]. Moreover, by the research of Carretero et al., 2013, with a 1m rise in sea level, an increase in saltwater intrusion length by more than 200 m into the coastal aquifer, Partido de La Costa, Argentina will severely degrade the aquifer. Punta Méda- nos would be the most affected by the sea-level rise; this is something to be considered because the are presents the main reservoir for Partido de La Costa [35].

Fig. 11. Distribution of saline and fresh groundwater boundary of Holocene aquifer with A2 scenario

From the forecast result about the area of saltwater of Holocene aquifer, the research will be estimated the quantity of fresh groundwater under the scenarios in the future. The calculation results of the potential of the groundwater storage are given in Table 6. The distribution of this freshwater mass is uneven throughout the area. Freshwater zones with large storage were distributed mainly in Thai Thuy, Hung Ha, Vu Thu districts and Thai Binh city. In 2100, the groundwater storage will be reduced by 263.663 Mio. m3 compared to 2017 with the A2 scenario.

346

T. T. T. Thuy et al.

Table. 6. The quantity of fresh groundwater in Holocene aquifer with the three scenarios District

The quantity of fresh groundwater (million m3 ) 2017

Dong Hung Hung Ha

B1 scenario

B2 scenario

A2 scenario

2060

2060

2060

2100

2100

2100

50.750

45.006

39.377

44.838

39.207

44.450

39.037

121.298

110.347

98.343

110.299

98.295

110.200

98.185

Kien Xuong

64.115

58.012

49.901

57.492

49.439

56.842

48.955

Quynh Phu

84.945

77.677

70.198

77.623

69.382

77.569

69.382

Thai Thuy 206.371

156.146

116.553

150.161

113.696

141.078

107.254

Tien Hai

101.447

62.579

35.531

58.480

31.667

52.717

26.436

Vu Thu and Thai Binh city

127.180

114.795

103.341

114.763

103.595

114.717

103.194

Total

756.106

624.562

513.243

613.655

505.282

597.573

492.443

6 Conclusion While climate change affects surface water resources directly through changes in the major long-term climate variables such as air temperature, precipitation, and evapotranspiration, the relationship between the changing climate variables and groundwater is more complicated and poorly understood. In Thai Binh province, there have been many studies relating to the effect of climate changes on surface water. However, this study is the first research about the potential effects of climate change and sea-level rise on groundwater in this area. The research was carried out based on the results of field surveys, in situ tests, analysis in laboratory and simulation results, from which to predict saline intrusion and changes in water quality of Holocene aquifers under the impact of climate change and sea level rise in the future. The study show that the Holocene aquifer has a close relationship with runoff, and groundwater is recharged by surface water and rainy water. The groundwater recharge and discharge conditions are a reflection of the precipitation regime, climatic variables, landscape characteristics, and human impacts such as agricultural drainage. In it, the fresh, brackish water and saltwater areas are intertwined. In 2017, the area of saltwater zones is smaller than that of 1996 by 180 km2 . The amount of rainfall supplying for Holocene aquifer is 345,460 m3 /day. The Holocene aquifer is most affected by tides is within 3.0 km of the coastline. By using the Visual Modflow model and Seawat code, groundwater flow, and the movement of salt and freshwater boundaries in Holocene aquifer under the impact of climate change and sea-level rise with scenarios were simulated. And the area of the saltwater zone of the Holocene aquifer will be expanded maximum with A2 scenario, more 109.7 km2 by 2100. By assessing the status of the

Assessment the Impact of Climate Change and Sea Level Rise

347

distribution of saline and freshwater zones in the Holocene aquifer in Thai Binh province and the movement of this boundary in the future, estimate the quantity of fresh groundwater under the scenarios in the future. This is the main aquifer to be used for domestic water supply in this area. The research results will be the basis that helps the managers give out protective solutions and sustainable using methods for this natural resource. Hence, we need to relocate reasonably the exploiting wells far away from the saline boundary to limit the movement of this boundary by the exploitation process. Reducing the number of private wells and replacing them with water plants to protect the aquifer from pollution sources and to control exploitation activities. Enhance the role of management, reasonable using and exploitation for sustainable development.

References 1. Rotzoll, K., Fletcher, C.H.: Assessment of groundwater inundation as a consequence of sealevel rise. Nature Clim. Change 3(5), 477–481 (2013) 2. Hiscock, K., Tanaka, Y.: Potential impacts of climate change on groundwater resources: from the High Plains of the US to the flatlands of the UK. In: Proceedings of the National Hydrology, Seminar: Water Resources in Ireland and Climate Change, pp. 19–26 (2006) 3. Ranjan, P., Kazama, S., Sawamoto, M.: Effects of climate change on coastal fresh groundwater resources. Global 16(4), 388–399 (2006) 4. Giambastiani, B.M.S., Antonellini, M., Oude, E.G., H. P., and Stuurman, R. J. : Saltwater intrusion in the unconfined coastal aquifer of Ravenna (Italy): a numerical model. J. Hydrol. 340(1–2), 91–104 (2007) 5. Werner, A.D., Simmons, C.T.: Impact of sea-level rise on sea water intrusion in coastal aquifers. Ground Water 47(2), 197–204 (2009) 6. Oude Essink, G.H.P., Van Baaren, E.S., Louw, P.G.B.: Effects of climate change on coastal groundwater systems: a modeling study in the Netherlands. Water Resour. Res. 46(10), 1–16 (2010) 7. Chang, S.W., Clement, T.P., Simpson, M.J., Lee, K.K.: Does sea-level rise have an impact on saltwater intrusion. Adv. Water Resour. 34(10), 1283–1291 (2011) 8. Holger, T., Martin-Bordes, J., Gurdak, J.: Climate change effects on groundwater resources. Published by CRC Press/Balkema, A global synthesis of findings and recommendations (2012) 9. Rasmussen, P., Sonnenborg, T.O., Goncear, G., Hinsby, K.: Assessing impacts of climate change, sea level rise, and drainage canals on saltwater intrusion to coastal aquifer. Hydrol. Earth Syst. Sci. 17(1), 421–443 (2013) 10. Luoma, S., Okkonen, J.: Impacts of future climate change and Baltic Sea level rise on groundwater recharge, groundwater levels, and surface leakage in the Hanko aquifer in southern Finland. Water 6(12), 3671–3700 (2014) 11. Hoover, D.J., Odigie, K.O., Swarzenski, P.W., Barnard, P.: Sea-level rise and coastal groundwater inundation and shoaling at select sites in California, USA. J. Hydrol. Reg. Stud. 11, 234–249 (2017) 12. Ferguson, G., Gleeson.: Vulnerability of coastal aquifers to groundwater use and climate change. Nature Clim. Change 2(5), 342–345 (2012) 13. Lemieux, J.M., Hassaoui, J., Molson, J., Therrien, R., Therrien, P., Chouteau, M., Ouellet, M.: Simulating the impact of climate change on the groundwater resources of the Magdalen Islands, Québec. Canada. J. Hydrol. Reg. Stud. 3, 400–423 (2015) 14. Sreekesh, S., Sreerama Naik, S.R., Rani, S.: Effect of sea level changes on the groundwater quality along the Coast of Ernakulam District. Kerala. J. Clim. Change 4(2), 51–65 (2018)

348

T. T. T. Thuy et al.

15. Tam, V.T., Batelaan, O., Beyen, I.: Impact assessment of climate change on a coastal groundwater system. Central Vietnam. Environ. Earth Sci. 75(10), 1–15 (2016) 16. Dausman, A., Langevin, C.D., U. S. G. Survey: Movement of the Saltwater Interface in the Surficial Aquifer System in Response to Hydrologic Stresses and Water Management Practices, Broward County, Florida. Scientific Investigations Report 2004–5256, p. 73 (2005) 17. Batelaan, O., Smedt, F.D.: GIS-based recharge estimation by coupling surface-subsurface water balances. J. Hydrol. 337(3–4), 337–355 (2007) 18. Shi, W., Lu, C., Werner, A.: D: Assessment of the impact of sea-level rise on seawater intrusion in sloping confined coastal aquifers. J. Hydrol. 586, 124872 (2020) 19. Mzila, N., Shuy, E. B.: Studies on groundwater salinity distribution in a coastal reclaimed land in Singapore. In: International Conference on Estuaries and coasts, pp. 590–596 (2003) 20. Sherif, M.M., Singh, V.P.: Effect of climate change on sea water intrusion in coastal aquifers. Hydrol. Process. 13(8), 1277–1287 (1999) 21. Pauw, P., De Louw, P.G.B., Oude Essink, G.H.P.: Groundwater salinisation in the Wadden Sea area of the Netherlands: Quantifying the effects of climate change, sea-level rise and anthropogenic interferences. Netherlands J. Geosci. 91(3), 373–383 (2012) 22. Hussain, M.S., Javadi, A.A.: Assessing impacts of sea level rise on seawater intrusion in a coastal aquifer with sloped shoreline boundary. J. Hydro-Environ. Res. 11, 29–41 (2016) 23. Knott, J.F., Jacobs, J.M., Daniel, J.S., Kirshen, P.: Modeling groundwater rise caused by sea-level rise in coastal new hampshire. J. Coastal Res. 35(1), 143–157 (2019) 24. Chun, J.A., Lim, C., Kim, D., Kim, J.S.: Assessing impacts of climate change and sea-level rise on seawater intrusion in a coastal aquifer. Water 10(4), 1–11 (2018) 25. Ministry of Natural Resources and Environment, Climate change and sea level rise scenarios for Vietnam (2016) 26. Ha, N.T.: Develop and implement the climate change adaptation solution at the coastal region of Vietnam (Vietadapt II), NAWAPI (2016). (in Vietnamese) 27. Hung, L.D.. et al.:Report on engineering geological and hydrogeological mapping scale 1/50000 for Thai Binh area. Ministry of Natural Resources and Environment (1996). (in Vietnamese) 28. Kolb, C., Pozzi, M., Samaras, C., VanBriesen, J.M.: Climate change impacts on bromide, trihalomethane formation, and health risks at coastal groundwater utilities. ASCE-ASME J. Risk Uncertainty Eng. Syst. 3(3), 1–11 (2017) 29. Thuy, T.T.T., Lam, N.V., On, D.H.: Distribution of saline and freshwater in groundwater in Thai Binh province and solution for reasonable exploitation. J. VietNamese Environ. 6, 120–125 (2014) 30. Ministry of Natural Resources and Environment, Climate change and sea level rise scenarios for Vietnam (2013). (in Vietnamese) 31. Wang, H., Gao, J. E., Zhang, M. Jie., Li, X. hua., Zhang, S. long., Jia, L.Z.: Effects of rainfall intensity on groundwater recharge based on simulated rainfall experiments and a groundwater flow model. Catena 127, 80–91 (2015) 32. Healy, R.W., Cook, P.G.: Using groundwater levels to estimate recharge. J. Hydrol. 10(1), 91–109 (2002) 33. Jan, C.D., Chen, T.H., Lo, W.C.: Effect of rainfall intensity and distribution on groundwater level fluctuations. J. Hydrol. 332(3–4), 348–360 (2007) 34. Melloul, A., Collin, M.: Hydrogeological changes in coastal aquifers due to sea level rise. Ocean Coast. Manag. 49(5–6), 281–297 (2006) 35. Carretero, S., Rapaglia, J., Bokuniewicz, H., Kruse, E.: Impact of sea-level rise on saltwater intrusion length into the coastal aquifer, Partido de La Costa. Argentina. Cont. Shelf Res. 61–62, 62–70 (2013)

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo of France Minh Tuan Vu1(B) , Yves Lacroix2 , and Quoc Hung Vu1 1 National University of Civil Engineering, Hanoi, Vietnam

[email protected] 2 Seatech, University of Toulon, 83162 La Valette du Var, France

Abstract. Giens double tombolo linking Giens island to the mainland is a unique geomorphological formation in the world. However, its existence has been threatened by coastal erosion, especially in the eastern part of this tombolo. The investigation of historical shoreline changes along the eastern Giens tombolo were carried out applying the integration of satellite remote sensing and geographic information system (GIS) techniques. Additionally, the combination of the Digital Shoreline Analysis System (DSAS) and linear regression method was used to predict the location of future shorelines. The results obtained from the analysis of shoreline position showed that the average annual change rate along the eastern Giens tombolo varied around +0.18 m/yr during the duration from 1973 to 2015, revealing a general progradation trend. Even though accretion is dominant, there are some local areas undergoing severe erosion. The most severely vulnerable areas were Les Cabanes du Gapeau, the south of Ceinturon, Pesquiers, and the north of La Capte with the maximum change rates of −1.05 m/yr, −0.77 m/yr, − 0.44 m/yr, and −0.29 m/yr, respectively. The change analysis of shorelines in 2020 and 2050 also reveals these severely eroded areas. On the other hand, this work demonstrates that both natural factors and human activities are the main causes of the shoreline changes in the eastern Giens tombolo. Keywords: Giens tombolo · Erosion and accretion · Satellite images · GIS · Shoreline prediction

1 Introduction The shoreline is the boundary between coastal land and the water body, where its shape and position changes continuously occur due to dynamic natural conditions and anthropogenic interventions [1]. The shoreline change normally induces coastal deposition or erosion, which is decided by the dominant processes acting on the shoreline. Both accretion and erosion issues greatly influence human lives, agriculture and aquaculture, natural resources, and waterway transport activities along the coastal zone. Therefore, shoreline mapping and change detection are essential tasks [2]. Over the years, several approaches have been developed to detect and monitor the shoreline evolution. They can be classified into four types. Firstly, the ground field surveying was used as the © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 349–372, 2021. https://doi.org/10.1007/978-3-030-60269-7_18

350

M. T. Vu et al.

only method for generating the shoreline maps. This method can obtain high accuracy of measurement, but is costly and time-consuming [3]. Next, the historical maps were used to provide a historical record, which cannot be found in any other data sources. However, it also contains many potential errors related to historical coastal chart or maps. As the aviation industry evolved, aerial photographs have been applied to provide sufficient pictorial information. The major disadvantages of this approach are the low frequency of data acquisition, limited temporal coverage, and costly as well as timeconsuming in the photogrammetric procedure. Additionally, the minimal spectral range of these photographs can induce errors in shoreline extraction [4]. From 1972, Landsat remote sensing satellite images combined with GIS techniques provide another highly efficient solution for shoreline mapping. The main advantages of this method are not time-consuming, timely and large ground coverage, inexpensive implemented cost, and have frequent data updates. These are also the main reasons why this technology is increasingly used in investigating and monitoring shoreline changes [1]. The double tombolo of Giens lies on the coast of South East of France, between the Gulf of Giens and Hyères bay (Fig. 1). Its shoreline has suffered the drastic changes due to the impact of both natural factors and human actions. Based on the study on the coastal and submarine sedimentology, Blanc [5] uncovered that the sediment transport was done mostly under the east wind, from east to west coast and from north to south along the eastern tombolo. Jeudy De Grissac [6] conducted a series of the field granulometric experiments in combination with the field wind and wave analyses to establish the map of current in Giens and Hyères bay. In the bay of Hyères, the waves generate the longshore current first directed from east to west and from north to south. This drift and notable fluvial Gapeau formed the eastern tombolo of Giens. Besides, by using aerial photographs taken from 1955 to 1972, he concluded that the eastern branch of Giens tombolo was threatened in some places and more particularly between the mouth of Gapeau and La Capte river. This erosive phenomenon could be caused by the implementation of transverse structures (port, jetty, and groynes) stopping the longshore drifts from east to west and from north to south. GEOMER [7] conducted a diachronic study of the coastline by photo-interpretation at the right of Les Cabanes du Gapeau from 1954 to1993 and also noted an almost continuous decline of the coastline from the mouth of Gapeau river to approximately 800 m south with the change rate of −1.5 m/yr, especially a remarkable decline of −40 m between 1969 and 1975. Capanni [8] investigated the shoreline changes of the eastern branch by using aerial photographs as well as some results of field surveys. The results of his study showed that the southern part of Gapeau river mouth, Ceinturon beach and Pesquiers beach experienced the retreat with an average change rate of − 0.68 m/yr, −0.35 m/yr, and −0.1 m/yr, respectively; whereas the remainders along the eastern tombolo were accreted with an average change rate of +0.26 m/yr between 1972 and 2003. Recently, Vu [9] reported that the regression of Posidonia might increase in wave height, current speed, and shoreline change rate along the coast of Giens tombolo. In this paper, we present a methodology to quantify the shoreline evolution from 1973 to 2015 and anticipate the position of the future shoreline, using the Landsat satellite images and GIS techniques. This technique shows to be of a comparable accuracy to the existing ones, while the data it relies on is freely available and has a regular and frequent update, also, is available worldwide.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

351

Fig. 1. The map of the study area and annual swell rose.

2 Study Area The eastern branch of Giens tombolo directly faces Hyères bay. It is bounded on the north by Gapeau river mouth and on the south by La Badine beach, approximately about 10 km [10]. For analysis purposes, the eastern branch is divided into five zones, the presence of artificial structures, and river mouths, created different morphological characteristics for each zone (Fig. 1). The first zone lies between the Gapeau river and Roubaud river extending about 1.95 km. It is accreted by the largest sediment volume from the Gapeau river, especially after the upstream jetty was constructed at the Roubaud river mouth from 1955 to 1960 [11]. The second zone of 1.925 km length is limited by the Roubaud river and the Hyeres port, adjacent to the road of DR 42. Its shoreline evolution is being dominated by anthropogenic interventions such as groynes, rock-fill revetments and breakwaters. Like the second zone, the shoreline of the third zone from Hyeres port to La Capte port is interrupted by some groynes which were installed to protect Pesquiers beach. This is also the shortest zone with only 1.4 km length. The fourth zone extends the entire La Capte beach with a total length of about 1.625 km. In the northern part of this zone, the concrete seawall was implemented to prevent erosion due to waves. The last zone, 1.7 km long, covers all Bergerie beach and La Badine beach. Three main wave directions frequently influence the Giens tombolo. The most frequent direction (36.9%) is western. The wave height of these western waves varies from 0.5 to 2.5 m in 75% of observation cases. The second frequent direction is southwest with frequency of 28.8%. Although these low energy waves usually occur with heights

352

M. T. Vu et al.

of less than 1.25 m and periods of about 6 s in 77% of observation cases, they mainly affect the beach evolution in the western tombolo. Conversely, the southeastern waves with frequency of 19.1% of total regime play an essential role in the shoreline evolution of the eastern branch. The weight and period of these waves approaching the study area are about 2 m in 31% of cases and more than 6 s over 25% of cases, respectively. Because of the low tidal variation of less than 0.3 m, waves play a decisive role in the shoreline evolution of of Giens tombolo [12]. Gapeau river and Pansard-Maravenne river, away from Giens tombolo about 6 km north and 11 km north-east, respectively, are two primary sources supplying the sediment for the study area. The longshore currents induced by the oblique waves have the great impact on redistributing sediment on the coast of Giens tombolo. Nevertheless, a large amount of the longshore sediment drifts is blocked by the transverse structures, viz. Port, jetty, and groynes, and part of them is also caught by the seagrass meadows of Posidonia [13]. Most of Hyères bay bed is covered with coarse sand derived mainly from organic products, particularly important the seagrass of Posidonia [6]. A decrease trend of sediment size is found and coincided with the longshore current direction [11]. From the mouth of Gapeau to the port of Hyères, the sediments are refined into southbound (an average grain of 0.65 to 0.25 mm). Between the mouth of Roubaud and the port of Hyères, the size refinement is accompanied by a marked reduction of the pebble percentage (from 34% to 0%) as well as improved sorting (from 1.2 to 0.8 ϕ with ϕ is the Krumbein phi scale) in agreement with the dominant longshore drift. From the port of Hyères to la Capte, the sediments are coarser (medium grain often greater than 0.5 mm), moderately sorted (0.8 to 1.2 ϕ), with a high concentration of pebbles (from 23–54%), which could here also come from inherited supplies or the dismantling of works [8]. From la Capte to la Badine, sediments are finer (0.25 mm), better sorted (0.35 to 0.4 ϕ) and the percentage of pebbles is reduced (40% at la Capte to 0% at la Badine).

3 Methodology 3.1 Data Acquisition and Preprocessing In this study, Landsat 1 MSS (Multispectral Scanner), Landsat 4 TM (Thematic Mapper), Landsat 7 ETM + (Enhanced ThematicMapper), and Landsat 8 OLI (Operational Land Imager) satellite images between 1973 and 2015 were acquired for extracting the shorelines. The image selection was based on some important criteria. Firstly, all the high-quality satellite images has been selected at the same time during the summer with the aim of eliminating the effects of storm surge and sea-level rise due to waves; only the images with cloud cover less than 10% have been selected [14]. The details regarding satellite data are presented in Table 1. After downloading, the Landsat satellite images normally many defects, viz. Radiometric and geometric distortion, wedge-shaped gaps, existence of noise, etc., caused by the altitude and attitude variations or velocity of the sensor platform [15]. Therefore, the preprocessing procedures including radiometric calibration, atmospheric correction, gap filling, pan-sharpening, and geometric rectification, need to be conducted to enhance the quality of image before being used as map bases.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

353

Table 1. Details of Landsat satellite images used for this study. Local time (GMT + 1)

No

Satellite/sensor

Acquired date (dd/mm/yyyy)

Resolution (m)

1

Landsat 1 (MSS)

1/3/1973

9:51

79

2

Landsat 4 (TM)

27/08/1988

9:47

30

3

Landsat 7 (ETM+)

28/08/2000

10:08

15/30

4

Landsat 7 (ETM+ )

18/08/2008

10:06

15/30

5

Landsat 8 (OLI)

30/08/2015

10:17

15/30

Firstly, all Landsat images were radiometrically calibrated and then converted to reflectance values. The 6S model were used to correct the reflectance values of each date atmospherically [16]. The atmospheric conditions, target and sensor altitude, band definitions, concentration and aerosol model, azimuth as well as zenith angles of the sun and sensor are the input parameters of this model [17]. Moreover, a failure of the Scan Line Corrector (SLC) created wedge-shaped gaps on both sides of all Landsat 7 ETM+ images. Thus, a gap filling preprocessing step was done by using the NASA Landsat gapfill tool linked to ENVI software program. The spectral data across gaps in SLC off images was interpolated from a gap free image [2]. After gap filling, the satellite images were increased the resolution through pansharpening step, but still maintain the spectral imagery quality. The Landsat TM and Landsat MSS images were resampled by using nearest neighbor and 1st order polynomial transformation into 15 m resolution. Moreover, the Landsat 7 ETM+ and Landsat 8 OLI multispectral images were sharpened to match the panchromatic band (Band 8) with the highest resolution of 15 m and obtained new images with the resolution of 15 m. The Gram-Schmidt algorithm in ENVI software based on principal component analysis was used to sharpen these satellite images. In the last step of preprocessing, all satellite images were geo-referenced to WGS_1984_UTM_Zone_49N map projection system in ENVI software. At least 6 Ground Control Points (GCP) was distributed and positioned throughout the spatial area of satellite image. The image rectification accuracy is assessed using the Root mean square error (RMSE) of less than 0.8 pixels. 3.2 Shoreline Extraction Up to now, some approaches have been created and developed to extract the shoreline from optical imagery, such as a single band method, the histogram thresholding method, the band ratio method, or a combination of these methods. However, the main difficulties of these methods are time-consuming, and the shoreline tends to move towards water in some coastal areas [4]. Recently, a new technique for extracting the shoreline, which is relatively simpler to implement than others, is the automatic technology of edge detection [18]. The most outstanding advantage of this method is that supplies the clear boundary between the land and the water in a short period of time. In this study, the exact shoreline

354

M. T. Vu et al.

was obtained by using the Matlab code programmed based on the Canny edge detector algorithm. This algorithm allows to localize well-optimizing detection localization [19]. Normally, the edge of the satellite image corresponds to the discontinuity of the image grey value. Therefore, the pixels in the land-water boundary will be determined by using the Canny edge detector algorithm if their grey values have relatively large changes [20]. An appropriate color composite which distinguishes clearly the boundary between water, soil and vegetated land can be used to extract the shoreline from the satellite image. The previous works showed that the best color composites include RGB (Red Green Blue) 5-6-7 for Landsat 1 MSS images, 5-4-3 for Landsat 4 TM and 7 ETM+ images, and 6-5-2 for Landsat 8 OLI images. These color composites not only nicely enhance the objects but also they are easily digitized. The files of digitized shorelines are in shape format for further analysis in DSAS version 4.3 which is an ArcGIS extension. 3.3 Shoreline Change Analysis For quantifying the shoreline evolution along the coast of the eastern Giens tombolo, several methods were available in DSAS. However, only the End Point Rate (EPR) and Linear Regression Rate-of-change (LRR) methods were used in this study. Two different approaches are utilized to compute the coast change, viz. EPR analyzing the short term changes in the period of 1973–1988; 1988–2000; 2000–2008; and 2008–2015, while LRR evaluating the long term changes between 1973 and 2015 as well as predicting future shoreline movements of 2015, 2020, and 2050. In this case, the onshore baseline was generated at a location approximately 100 m back from the shoreline of 2015. From the initial settings, a total of 347 transects were created along the eastern tombolo and perpendicular to the baseline. Each transect is about 200 m long and spaced about 25 m evenly. From there, DSAS may estimate the coordinate of the intersection points between shorelines and transect lines as well as other statistical results. Consequently, the data calculated from each transect were used to evaluate the rate of shoreline changes (m/yr). The intersection points between transects and multi-temporal shorelines were computed by DSAS to input into the linear regression Eq. (1) to predict the position of shoreline in the future. Using the linear regression method to determine shoreline position change rate will eliminates not only potential random error but also short term variability [21]. This method is based on the assumption that the observed periodical rate of shoreline change is the best estimate for the prediction of shoreline position in the future. Nevertheless, it does not consider wave interference or the sediment transport [22] because the cumulative effect of all the underlying processes are assumed to be captured in the historic position of shoreline [23]. y = a.x + b Where: y: The distance from baseline, x: Date of the shoreline, a: The rate of shoreline change calculated as follows:    n ni=1 xi .yi − ni=1 xi . ni=1 yi a=   n ni=1 xi2 − ( ni=1 xi )2

(1)

(2)

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

yi : The distance from shoreline at date of x i to baseline, n: The number of shorelines, b: Constant computed as follows:  n 1 n yi − a. xi b= i=1 i=1 n

355

(3)

Cross-validation of the predicted historical shoreline locations is used to determine the predictability and the quality of the model. Specifically, the accuracy of the predicted shoreline position of the zone 2 in 2015 was proved by validating to the extracted shoreline from the satellite image of 2015 (Fig. 2). Root Mean Square Error (RMSE) was calculated to assess the accuracy of the predictions. The overall RMSE error calculated for the whole shoreline of zone 2 was found approximately 6.57 m. Furthermore, the regression coefficient of R-squared is estimated about at 0.981. It is seen that the value of errors shows better agreement with measurement result. Therefore, the linear regression method could be used to estimate the position of shoreline in the future.

Fig. 2. Comparison between the shoreline extracted from the satellite image and the shoreline predicted by Matlab code in 2015.

Before conducting any analysis, some main sources of error which can greatly affect the precision of shoreline position and consequently rate of shoreline changes should be calculated. They are classified into two categories including positional and measurement uncertainties. Two positional uncertainties, viz. Seasonal error E s , and tidal fluctuation error E td , are associated with the features and phenomenon deceasing the precision and accuracy of defining a shoreline position in a given year; whereas three measurement uncertainties, viz. Digitizing error E d , rectification error E r and pixel error E p , are related

356

M. T. Vu et al.

to the approach and skill [24]. These errors are random and uncorrelated [24], hence the total errors, U t , is defined by the square root of the sum of the squares of the different errors:  2 + E2 + E2 + E2 Ut = ± Es2 + Etd (4) p r d Where seasonal error, E s , was determined as the shoreline position differences between the spring and fall and ranged around ±5 m [25]. The tidal fluctuation error, E td , could be omitted because a variation of tide in the study area is quite small and less than 0.3 m [26]. Regarding the digitizing error E d , it was estimated at ±12 m for 1973, ±6 m for 1988 and ±3 m for remainders based on the results of the pan-sharpening process [27]. Next, the rectification error, E r , was calculated from the geo-referencing process. Finally, the pixel error E p , was considered the same value for all satellite images and about ±5 m [11]. The annualized error of rate of shoreline change at any given transect was estimated as follows [28]:  2 + U2 + U2 + U2 + U2 Ut1 t2 t3 t4 t5 (5) Ua = ± T 2 , U 2 , .. U 2 are the total error of shoreline position for the different year Where Ut1 t2 t5 and T is the total year of analysis. The maximum annualized error estimated for individual transects is approximately ±0.72 m/yr (Table 2).

Table 2. Estimated potential errors for historical shoreline position in the period from 1973 to 2015. Type of errors

1973 1988 2000 2008 2015

Seasonal error (E s )

5

Tidal fluctuation (E td )

0

0

Digitizing error (E d )

12

6

Rectification error (E r ) 12

9.9

Pixel error (E p )

5

5

Total error (U t )

18.39 13.57 10.67 13.25 10.23

5

5

5

5

0

0

0

3

3

3

7.35

10.8

6.75

5

5

5

Annualized error (U a ) 0.72 m/yr

4 Results and Discussion 4.1 Historical Shoreline Change During the Period from 1973 to 2015 For zone 1, the rate of change was investigated over 79 transects (No. 1-79) and observed both accretion and erosion, but most of transects exhibit accretion except the period from

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

357

2008 to 2015 (Fig. 3). The accreted realms are fed by the Gapeau river. In the period from 1973 to 1988, the shoreline advanced with the change rate of +1.02 m/yr, but the erosion pattern was dominant between 1988 and 2000. In this period, the northern shoreline (Transects 1–46) was declined with the maximum change rate of about − 3.81 m/yr (Table 3) perhaps attributed to the action of southeastern waves. In order to reduce the decline of shoreline, the 500 m rockfill revetment was implemented from the Gapeau river mouth to downstream in the period from 1995 to 2000. Meanwhile, the southern part (Transect 47–79) was accumulated because the longshore sediment transport was blocked by the jetty of the Roubaud river. The presence of revetment maintained the stable condition of shoreline for accreting during the period from 2000 to 2008 period except for few transects. The maximum progradation rate of about + 4.36 m/yr appeared nearly Transect 5. Nevertheless, the positive trend was changed entirely to the negative trend between 2008 and 2015. All shoreline suffered erosion at a mean retreat rate of −1.35 m/yr. The main reason for this phenomenon may be due to the shortage of sediment from the Gapeau river. On the other hand, the overall shoreline changes from 1973 to 2015 are depicted in Fig. 4. It is showed that erosion is reported from Transects 4 to 24 (immediately downstream of revetment) with the highest change rate of −1.05 m/yr. In contrast, accretion is observed from Transects 25 to 79 at the highest accumulation rate of +1.35 m/yr. More than 72% of transects in this zone exhibit moderate accretion.

Fig. 3. Location of historical shorelines and rate of shoreline change of Zone 1 (Transects 01–79) using the EPR method in over the period of 1973–2015.

In zone 2, the result of short-term analyses reveals that both accretion and erosion are observed in many places. The shoreline changes during each period were estimated for 78 transects (No. 80-157) and shown in Fig. 5 as well as Table 3. Between 1973 and 1978, the northern shoreline of this zone, viz. Ceinturon beach, was subject to severe erosion with the highest change rate of −2.27 m/yr. This shoreline decline has been mainly attributed to the sediment deficit due to the main longshore sediment transport stopped by the jetty of Roubaud port and the strong impact of the Southeast waves. In order to trap of sediment and limit erosion, four groynes were implemented in this area during period from 1978 to 1982. Conversely, the southern shoreline from Transect 128 to 157 continuously advanced seaward from 1973 to 2008, probably due to the upstream breakwater of

358

M. T. Vu et al.

Fig. 4. Rate of shoreline change of Zone 1 (Transects 01–79) using the LRR method in over the period of 1973–2015.

Hyeres port which stopped the north-south sediment transport. The presence of the groynes and breakwaters in this realm modified the position of shoreline in the positive trend. The percentage of accretion transects was increased from 46% in the period from 1973 to 1988 to 84% in the period from 1988 to 2000. The groynes have induced both negative and positive effects. They could lead to localized beach accretion, but severe erosion occurred at several places in downstream. In particular, between 1988 and 2000, the shoreline advanced seaward from transects 80 to 107, but the retreat of the shoreline was observed from transect 107 to 125. After that, 92% of transects manifest accretion in 2000–2008 because of annual beach nourishment [8]. However, the positive trend was entirely changed to the recession mode in the period of 2008–2015. The shoreline retreated landward at the mean change rate of approximately −0.6 m/yr. In parallel with the mentioned-above short-term analyses, the shoreline changes of this zone were also carried out throughout the long-term analyses of 1973–2015. It is noted that both erosion and accretion occurred along the shoreline (Fig. 6). The maximum retreat and deposition rates are −0.77 m/yr and +2.08 m/yr, respectively. Erosion occurred at transects 105-135 in the south of the groynes, whilst accretion is reported in the remainders. The increase of erosion in this zone directly threatens the existence of the RD 42 road. This decay is triggered by the combination of action from southwest waves as well as the deficit of sediment due to the upstream groynes and jetty. For zone 3, from Transect 158 to 213, the results of short-term analysis indicate that the shoreline of this zone was both eroded and accreted, but accretion is dominant (Fig. 7). Furthermore, the shoreline evolution is very complex due to the disturbance of the groynes distributed along the coast of this zone. In the period from 1973 to 1988, the erosion predominantly dominates with more than 92% of transects with the highest change rate of −2.14 m/yr. Subsequently, this negative trend was turned to positive in the

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

359

Fig. 5. Location of historical shorelines and rate of shoreline change of Zone 2 (Transects 80–157) using the EPR method in over the period of 1973–2015.

Fig. 6. Rate of shoreline change of Zone 2 (Transects 80–157) using the LRR method in over the period of 1973–2015.

periods from 1988 to 2000 and from 2000 to 2008 with mean change rates of +0.24 m/yr and +0.91 m/yr, respectively. The progradation area of this zone was slightly decreased in the period from 2008 to 2015. In this duration, solely 62.5% of transects exhibit accretion. Nevertheless, the long-term analysis demonstrates that erosion is dominant from 1973 to 2015 (Fig. 8). The negative trend results in not only the shortage of sediment due to the presence of the breakwaters in Hyeres port but also to the wave action. This is an area of convergence of orthogonal west wave and correspondingly contributes to an increase of its energy, particularly near the vicinity of Hyeres port. Around 73% of eroded transects are reported in the northern part of this zone, while about 27% of deposited transects concentrated in the southern part. The highest retreat rate of about − 0.44 m/yr often occurs round Transect 158 and the highest deposition rate of +0.46 m/yr is recorded near Transect 210–213 (Table 3).

360

M. T. Vu et al.

Fig. 7. Location of historical shorelines and rate of shoreline change of Zone 3 (Transects 158– 213) using the EPR method in over the period of 1973–2015.

Fig. 8. Rate of shoreline change of Zone 3 (Transects 158–213) using the LRR method in over the period of 1973–2015.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

361

There are 65 transects, namely from transect 214 to 278 in zone 4. The results of the short-term analysis elucidates that the coast has undergone both accretion and erosion (Fig. 9). After completing the construction of 80-m wing perpendicular to the jetty of La Capte port [8], all shoreline of zone 4 underwent a serious decline with the mean change rate of approximately −1.33 m/yr in the period from 1973 to 1988. The maximum erosion rate of −2.71 m/yr was recorded at the Transect 215 in the north of the La Capte beach. This negative trend sharply decreased in the period from 1988 to 2000. Erosion was only reported in the northern part of zone 4 at the highest change rate of −0.74 m/yr, whilst the southern one exhibits accretion with the highest change rate of +1.25 m/yr. In order to stabilize the shoreline as well as protect the onshore properties, two geotube submerged breakwaters were constructed in the south of La Capte port in 2007. Along with the annual beach nourishment with the large sediment volume, accretion was seen in most of the transects (about 90%) in the period from 2000 to 2008. The highest deposition rate is estimated about +4.81 m/yr at transect 216. However, the alongshore accretion pattern in the north part of this zone was completely transformed into erosion pattern during the period from 2008 to 2015. The highest retreat rate of -1.91 m/yr is observed at transect 216. The main cause of this trend change may be attributed to the decline of the geotube breakwater height due to the geotextile bag stretched by hydrodynamic effects or torn by boat anchors and mechanical forces [10]. Furthermore, Fig. 10 describes the

Fig. 9. Location of historical shorelines and rate of shoreline change of Zone 4 (Transects 214– 278) using the EPR method in over the period of 1973–2015.

362

M. T. Vu et al.

shoreline change rate in the long-term period of 1973–2015. It is also noted that the southern part is accreted with the highest deposition rate of +0.19 m/yr, whereas the northern part has retreated with the highest retreat rate of −0.29 m/yr (Table 3). The numbers of accreted and eroded transects are 36 and 29 corresponding to 55.4% and 44.6%.

Fig. 10. Rate of shoreline change of Zone 4 (Transects 214–278) using the LRR method in over the period of 1973–2015.

In the last region, zone 5, the shoreline is located between transect 279 and transect 347. The shoreline changes over the short-term periods (1973–1988, 1988–2000, 2000– 2008, and 2008–2015) were estimated and shown in Fig. 11. It is clearly seen that during the periods from 1973 to 1988 and from 1988 to 2000, the coast experienced dominant erosion with the mean retreat rate of −0.5 m/yr and −0.24 m/yr, respectively (Table 3). Nonetheless, from 2000, the longshore erosion pattern totally changed to accretion pattern except few eroded transects. Over the period from 2000 to 2008, the shoreline advanced seaward at the mean deposition rate of +0.89 m/yr. At that time, more than 97% of transects showed accretion. The positive trend kept maintaining in the period of 2008–2015 with the mean accretion rate of +0.71 m/yr. On the other hand, the long-term analysis expresses that the shoreline experienced very little change from 1973 to 2015 (Fig. 12). Consequently, there is certain stability or even a slight deposition at the mean progradation rate of +0.02 m/yr. This accretion is due to its position sheltered by the cape of Esterel. Notably, the divergence of orthogonal east and southeast waves in this area reduces its energy, so prompting to the accumulation of the fluvial sediment contribution from Gapeau and Roubaud river which is transported by southward longshore drift.

4

3

2

2.12 4.36 0.47 1.35

−3.81 −1.18 −3.43 −1.05

1988–2000

2000–2008

2008–2015

1973–2015

2.86 0.82 2.08 0.29 1.25 2.55 1.36 0.46 −0.1 1.25

−0.68 −1.64 −0.77 −2.14 −1.13 −0.75 −1.7 −0.44 −2.71 −0.74

2000–2008

2008–2015

1973–2015

1988–2000

2000–2008

2008–2015

1973–2015

1988–2000

1973–1988

1973–1988

65

1625

1400

3.43

56

2.62

−0.44

1925

−2.27

1973–1988

1988–2000

78

1950

2.63

79

−2.21

1973–1988

1

Coast length (m)

Accretion rate (m/yr)

No. of transect

Erosion rate (m/yr)

Duration

Zone

0.2

−1.33

−0.11

0.14

0.91

0.24

−1.21

0.4

−0.6

1.12

0.88

−0.23

0.45

−1.35

1.82

−0.33

1.02

Mean rate (m/yr)

19

65

41

21

9

15

52

25

65

6

12

42

22

76

7

40

10

No. of eroded transect

46

0

15

35

47

41

4

53

13

72

66

36

57

3

72

39

69

No. of accreted transect

Table 3. Rate of shoreline changes for the eastern Giens tombolo between 1973 and 2015.

29.23

100

73.21

37.5

16

26.79

92.86

32

83.33

7.7

15.38

53.85

27.85

96.2

8.86

50.63

12.66

% of eroded transect

(continued)

70.77

0

26.79

62.5

84

73.21

7.14

68

16.67

92.3

84.62

46.15

72.15

3.8

91.14

49.37

87.34

% of accreted transect Assessment of the Shoreline Evolution at the Eastern Giens Tombolo 363

5

Zone

0.19

−0.29

1973–2015 0.05 0.28 2 1.99 0.3

−1.17 −1.04 −0.64 −0.4 −0.27

2000–2008

2008–2015

1973–2015

1973–1988

1988–2000

1700

2.6

−1.91

2008–2015

69

4.81

−0.19

2000–2008

Coast length (m)

Accretion rate (m/yr)

No. of transect

Erosion rate (m/yr)

Duration

0.02

0.71

0.89

−0.24

−0.5

−0.01

1.02

1.18

Mean rate (m/yr)

Table 3. (continued)

30

6

2

50

68

29

15

6

No. of eroded transect

39

63

67

19

1

36

50

59

No. of accreted transect

43.5

8.7

3

72.46

98.55

44.6

23

9.23

% of eroded transect

56.5

91.3

97

27.54

1.45

55.4

77

90.77

% of accreted transect

364 M. T. Vu et al.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

365

Fig. 11. Location of historical shorelines and rate of shoreline change of Zone 5 (Transects 279–347) using the EPR method in over the period of 1973–2015.

4.2 Estimation of Shoreline Changes During Period from 2015 to 2050 The positions of shoreline in 2015 and 2050 have been forecasted for the eastern part without taking into account the disastrous impacts such as storms. The brief results of this prediction are presented in Table 4 and shown in Fig. 13, Fig. 14, Fig. 15, Fig. 16, and Fig. 17. For zone 1, the result of future shoreline changes indicates the accretion trend is observed in most of transects (Fig. 13). The progradation trend is dominated by the mean rates of +1.8 m/yr, +0.45 m/yr, and +0.62 m/yr in the periods of 2015–2020, 2020–2050, and 2015–2050, respectively (Table 4). Over all periods, erosion is mainly concentrated in the area from Transect 5 to 25, whilst the remainders are accumulated by sediment supply from the Gapeau river. The advance of the southern part in this zone can result from the interference of Roubaud jetty in trapping the longshore sediment transport. In zone 2, mixed erosion and accretion was exhibited. However, accretion is dominant than erosion (Fig. 14). This positive trend is confirmed by the average change rates of 0.89 m/yr in 2015–2020, 0.4 m/yr in 2020–2050, and 0.46 m/yr in 2015–2050 (Table 4). Moreover, the % of deposited transects is increased from 62.82% in 2015–2020 to 67.95% in the next periods. Ceinturon beach will be continuously advanced seaward due to the presence of four groynes, which play the decisive role in accumulating the sediment

366

M. T. Vu et al.

Fig. 12. Rate of shoreline change of Zone 5 (Transects 279–347) using the LRR method in over the period of 1973–2015.

Fig. 13. Location of future shorelines and rate of shoreline change of Zone 1 (Transects 01–79) using the EPR method in over the period of 2015–2050.

fed by the Gapeau river. The maximum recession rates are often observed surrounding transect 114, immediately in the south of these groynes, whereas the maximum accretion rates mostly concentrate in the south of this zone. In zone 3, most the transects manifest recession, except transect from 158 to 165 corresponding to the shoreline between Hyeres port and the first groyne which are progradated (Fig. 15). During the period from 2015 to 2020, the average change rate is predicted approximately −1.17 m/yr, revealing a retreat trend. The highest loss rate of −2.59 m/yr is recorded around transect 206, while the highest deposition rate of 0.8 m/yr is seen in the northern part of this zone. In the period from 2020 to 2050, the negative trend is maintained almost continuously from the first groyne to the jetty of La Capte port, where the maximum recession rate of −0.44 m/yr is observed. Once again,

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

367

Fig. 14. Location of future shorelines and rate of shoreline change of Zone 2 (Transects 80–157) using the EPR method in over the period of 2015–2050.

the long-term analysis from 2015 to 2050 demonstrates the decline of shoreline at the average change rate of −0.24m/yr despite the presence of the groynes (Table 4).

Fig. 15. Location of future shorelines and rate of shoreline change of Zone 3 (Transects 158–213) using the EPR method in over the period of 2015–2050.

368

M. T. Vu et al.

Among five zones along the eastern Giens tombolo, the highest erosion rate of − 3 m/yr is predicted for zone 4, immediately in the south of geotube submerged breakwaters. This zone is absolutely dominated by a retreating trend with 100% of eroded transects during the period from 2015 to 2020 (Fig. 16). Nevertheless, this trend partly changes in the period from 2020 to 2050. The results of prediction show that the northern part of this zone has experienced erosion at the highest change rate of −0.29 m/yr, whilst the southern part is deposited at the highest change rate of +0.19 m/yr (Table 4). Generally, the shoreline along zone 4 is forecasted to decline by the average change rate of −0.25 m/yr over the period from 2015 to 2050, especially in the northern area of the La Capte beach.

Fig. 16. Location of future shorelines and rate of shoreline change of Zone 4 (Transects 214–278) using the EPR method in over the period of 2015–2050.

In the last zone of the eastern branch, the variation of the shoreline over time is the lowest compared with that of other zones. Results of statistical analysis carried out for all 69 transects indicate the erosion and accretion trend is shifted alternatively (Fig. 14). Between 2015 and 2020, 100% of transects is subjected to erosion at the highest recession rate of −2.21 m/yr around transect 313. The erosive tendency decreases in the period from 2020 to 2050, with only 42% of eroded transects. The northern part of the Bergerie beach would be suffered slight erosion at the highest recession rate of −0.28 m/yr, while accretion is predicted in the south of this zone at the highest change rate of +0.3 m/yr.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

369

The average change rate of all shoreline is about +0.02 m/yr, revealing a little accretion trend in this period. In addition, the analysis carried out from 2015 to 2050 notices that this zone could be undergone with an erosion tendency (82.6% of recession transects) at a mean change rate of −0.13 m/yr (Table 4).

Fig. 17. Location of future shorelines and rate of shoreline change of Zone 5 (Transects 279–347) using the EPR method in over the period of 2015–2050.

5

4

3

2

1.71

−1.25

2.3

−0.89

2015–2050

−0.14 0.3 0.16

−2.21 −0.28 −0.45

2015–2050

2015–2020

2020–2050

1700

−0.02

−0.61

2015–2050

69

0.19

−0.77

−0.29

2020–2050

2015–2020

−3

0.49

−0.66

2015–2050

1625

0.47

−0.44

65

0.8

−2.59

2015–2020

2020–2050

1400

2.1

56

4.26

−0.77

1925

−1.69

2015–2020

2020–2050

78

1950

2015–2050

79 1.37

Accretion rate (m/yr) 4.63

Erosion rate (m/yr)

−1.05

Coast length (m) −2.63

2015–2020

1

No. of transect

2020–2050

Duration

Zone

-0.13

0.02

−1.17

−0.25

−0.01

−1.92

−0.24

−0.11

−1.17

0.46

0.4

0.89

0.62

0.45

1.8

Mean rate (m/yr)

57

29

69

65

27

65

44

41

49

25

25

29

21

22

16

No. of eroded transect

12

40

0

0

38

0

12

15

7

53

53

49

58

57

63

No. of accreted transect

82.6

42

100

100

41.54

100

78.57

73.21

87.5

32.05

32.05

37.18

26.58

27.85

20.25

% of eroded transect

Table 4. Rate of shoreline changes for the eastern Giens tombolo over the period of 2015–2050.

17.4

58

0

0

58.46

0

21.43

26.79

12.5

67.95

67.95

62.82

73.42

72.15

79.75

% of accreted transect

370 M. T. Vu et al.

Assessment of the Shoreline Evolution at the Eastern Giens Tombolo

371

5 Summary The shoreline along the eastern Giens tombolo from 1973 to 2015 underwent alternating shifts of deposition and erosion. Erosion occurred in the north shorelines of zone 1 and zone 4 and the center part of zone 2 and zone 3; whereas the south shoreline of these zones was accreted with the change rate of +2 m/yr. The shore-normal structures, viz. Groynes and jetties, are attributed to be the main cause off sediment deficit in the downstream of these structures and erosion in their adjacent areas. The results of this work confirm that the beach nourishment plays a secondary role as the temporal method for limiting the decline of shoreline. Moreover, the estimation of shoreline changes during the period from 2015 to 2050 also reveals that Cabanes beach, Pesquiers beach, Ceinturon beach and La Capte beach are the most vulnerable areas to the severe erosion. Especially, in zone 2, the road of DR 42 may be disappeared in 2050 due to coastal erosion with the change rate of −0.89 m/yr. However, the accretion can be observed in some places in the upstream of jetties and breakwaters. The study result can be useful for future development and management of Giens tombolo coast as well as coastal zones around Hyères city. Acknowledgments. The authors would like to thank Vietnam International Education Development, Ministry of Education and Training, Vietnam (Grant No. 911) for the financial support and the organizations of EOL, CETMEF, CEREMA, SHOM, and REFMAR for providing the data of field investigation. The authors also wish to express sincere thanks to the USGS for sharing the Landsat series images as well as for the making the DSAS available on their website.

References 1. Winarso, G., Janto, J., Budhiman, S.: The potential application of remote sensing data for coastal study. In: 22nd Asian Conference on Remote Sensing, Singapore (2001) 2. Cassé, C., Pham, B.V., Pham, T.N.N., Hoang, P.P., Nguyen, L.D.: Remote sensing application for coastline detection in Ca Mau, Mekong delta. In: International Symposium on Geoinformatics for Spatial Infrastructure Development in Earth and Allied Sciences, Ho Chi Minh City, Vietnam (2012) 3. Tran, T.V., Tran, T.B.: Application of remote sensing for shoreline change detection in Cuu Long Estuary. VNU J Earth Environ Sci. 25, 217–222 (2009) 4. Alesheikh, A., Ghorbanali, A., Nouri, N.: Coastline change detection using remote sensing. Int. J. Environ. Sci. Technol. 4(1), 61–66 (2007) 5. Blanc, J.J.: Coastal and submarine sedimentological research in Western Provence. Annals of the Oceanographic Institute: Oceanographic Institute, Masson and Cie (1958) 6. Jeudy De Grissac, A.: Dynamic sedimentology of the Hyères and Giens bays (Var). Planning problems, p. 86 + annexes. University of Aix-Marseille II, Marseille (1975) 7. GEOMER: Coastal development - Feasibility study of an inland port at the mouth of the Gapeau: Currentological and sedimentological study (1996) 8. Capanni, R.: Study and integrated management of sediment transport in the Gapeau river/Hyères bay system. University of Aix Marseille 1 (2011) 9. Vu, M.T., Lacroix, Y., Nguyen, V.T.: Investigating the impacts of the regression of Posidonia oceanica on hydrodynamics and sediment transport in Giens Gulf. Ocean Eng. 146, 70–86 (2017)

372

M. T. Vu et al.

10. Lacroix, Y., Vu, M.T., Than, V.V., Nguyen, V.T.: Modeling the effect of geotextile submerged breakwater on hydrodynamics in La Capte beach. In: Vietnam-Japan Workshop on Estuaries, Coasts and Rivers, Hoi An, Vietnam (2015) 11. Courtaud, J.: Geomorphological dynamics and coastal risks case of the tombolo of Giens (Var, southern France), p. 263. University of Aix-Marseille I (2000) 12. SOGREAH: Sedimentological studies of the bay of Hyères. Coastline of Pothuau port and La Badine, p. 68 + appendices and maps (1988) 13. OCEANIDE: Study for the protection of Ceinturon beach and the southern sector of SaintPierre port - Phase 1: Synthesis of knowledge – Report (2010) 14. Vu, M.T., Lacroix, Y., Than, V.V., Nguyen, V.T.: Prediction of shoreline changes in Almanarre beach using geospatial techniques. Indian J. Geo-Mar. Sci. 49(2), 207–217 (2020) 15. Lillesand, T.M., Kiefer, R.W., Chipman, J.W.: Remote Sensing and Image Interpretation. Wiley, Hoboken (2008) 16. Vermote, E.F., Tanre, D., Deuze, J.L., Herman, M., Morcette, J.J.: Second simulation of the satellite signal in the solar spectrum, 6S: an overview. IEEE Trans. Geosci. Remote Sens. 35(3), 675–686 (1997) 17. Dewidar, K.: Changes in the shoreline position caused by natural processes for coastline of Marsa Alam and Hamata, Red Sea, Egypt. Int. J. Geosci. 2, 523–529 (2011) 18. Loos, E.A., Niemann, K.O.: Shoreline feature extraction from remotely-sensed imagery. In: IEEE International Geoscience and Remote Sensing Symposium (2002) 19. Canny, J.: A computational approach to edge detection. IEEE Trans. Pattern Anal. Mach. Intell. PAMI-8(6), 679–698 (1986) 20. Liu, H., Jezek, K.C.: Automated extraction of coastline from satellite imagery by integrating Canny edge detection and locally adaptive thresholding methods. Int. J. Remote Sens. 25(5), 937–958 (2004) 21. Douglas, B.C., Mark, C.: Long-term shoreline position prediction and error propagation. J. Coastal Res. 16(1), 145–152 (2000) 22. Michael, S.F., Dolan, R., Elder, J.F.: A new method for predicting shoreline positions from historical data. J. Coastal Res. 9(1), 147–171 (1993) 23. Li, R., Liu, J.K., Felus, Y.: Spatial modeling and analysis for shoreline change detection and coastal erosion monitoring. Mar. Geodesy 24(1), 1–12 (2001) 24. Fletcher, C.H., Romine, B.M., Genz, A.S., Barbee, M.M., Dyer, M., Anderson, T.R., Lim, S.C., Vitousek, S., Bochicchio, C., Richmond, B.M.: National assessment of shoreline change: historical shoreline change in the Hawaiian Islands. Open-File Report, Reston, VA, p. i-55 (2012) 25. E.O.L. Monitoring of the evolution of the beaches of the commune of Hyères-les-palmiers. Municipality of Heres-Les-Palmiers (2010) 26. Rajasree, B.R., Deo, M.C., Sheela Nair, L.: Effect of climate change on shoreline shifts at a straight and continuous coast. Estuar Coast Shelf Sci. 183, 221–234 (2016) 27. Jayson-Quashigah, P.N., Addo, K.A., Kodzo, K.S.: Medium resolution satellite imagery as a tool for monitoring shoreline change. Case study of the Eastern coast of Ghana. J. Coastal Res. 65, 511–516 (2013) 28. Hapke, C.J., Himmelstoss, E.A., Kratzmann, M.G., List, J.H., Thieler, E.R.: National Assessment of Shoreline Change: Historical Shoreline Change along the New England and Mid-Atlantic Coasts, p. 57. U.S. Geological Survey (2011)

The Evolution of Water Management in the Red River Delta of Vietnam and a Case of Chuc Son, Hanoi City Tuan Anh Pham1(B)

and Kelly Shannon2

1 Department of Landscape Architecture, Faculty of Architecture & Planning,

National University of Civil Engineering, Hanoi, Vietnam [email protected] 2 OSA/ Department of Architecture, Faculty of Engineering, KU Leuven, Leuven, Belgium

Abstract. For more than 2000 year, the Red River Delta has been created by (and always tied to) the regime of the Red River. The tradition of building the rive dike system in the Red River Delta has a very long interesting history. Historically, in the feudal period, the dike system was controlled and maintained by the local peasantry which revealed an understanding about the logics of the dynamics of this delta and its landscape. Humans had tamed the Red River’s water regime with quite low techniques, simple means, massive human endeavor and ingeniousness. Over time, acquired both indigenous and imported knowledge and new techniques, water management in the Red River Delta has become more refined and more complicated. However, the scope and speed of a swift urbanization nowadays, in combination of the environmental crises, predicted consequences of climate change as well as uncontrollable up-steam river constructions by neighbouring countries, has heralded a new era that demands a radical rethinking of water management. The paper is structured in three main parts. It firstly reviews the Red River Delta’s historical water management in order to understand the broader context and eventually draw lessons. Secondly, it discusses contemporary challenges in terms of water management for Vietnam in light of unprecedented modernization and urbanization. The article concludes with the case of Chuc Son, an area in the southern region of the Day River, which is an important tributary of the Red River and significant in terms of the capital city’s westward expansion. Keywords: Red River Delta · Day River · Water management · Urbanization · Climate change · Chuc Son

1 Introduction Vietnam’s Red River Delta has an extremely high population density (averaging 1,004 persons/km2 ) [6]) and hosts an ancient wet-rice civilization. The territory has a range of 0.1–1.5 km/km2 of naturally flowing rivers and 0.67–1.6 km/km2 of irrigated land [31]. As the geography which hosts most ancient human settlements in Vietnam, the Red River © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 373–395, 2021. https://doi.org/10.1007/978-3-030-60269-7_19

374

T. A. Pham and K. Shannon

Delta has more than 2000 year of land reclamation for paddy farming [4, 23]. Besides receiving the privileges of nature, due to both geography and climate, the Red River Delta endures extremely from natural calamities. The most dangerous disasters are related to water, including annual tropical typhoons, seasonal droughts and floods (Fig. 1). For over 1000 year, mankind has developed water management techniques aiming to reclaim and reconcile with the natural forces. The delta’s intricate system of water management has been built both for protection and irrigation since the 8th century. This water management system includes a dense and complex network of semi-permanent and permanent dikes, sluices, and pumping stations [20].

(a)

(b)

Fig. 1. Historically, the Red River Delta has been received natural disasters of seasonal floods and droughts: a) 1930 Extreme flooding in the Red River in Hanoi during rainy season [5]; b) 1927 extreme drought in the Red River in Hanoi during dry season. [Dépôt des archives d’outre-mer, Aix-en-Provence: FR/CAOM/30Fi119/70].

As the same as most large rivers in the world, the Red River’s water regime is varied. Its water level fluctuates between 1.41 to 14.3 m throughout the year with an average of flow of 2,640 m3 /s [19, 21]. Over 1902–1990, the average water discharge of the Red River was about 3740 m3 /s at Son Tay [32]. The water volume of the Red River totals approximately 83.5 Mio. m3 /year, and its water velocity can reach 3.45 m/s [19, 21]. This lets the Red River can become one of the world’s largest rivers regarding the water flow. Normally, the water level of the Red River rises quite slowly. However, its water level can very quickly rise from 1.0 to 4.0 m within 24 h after the typhoons [9]. Its water carries a great volume of alluvial sediment during floods. The water of the Red River can carry averagely about 1000g of silt/m3 and about 114–115 Mio. Tons/year [19]. The quantity of water and silt are different throughout a year, where there is 65–80% of water and 90% of silt [19] during the rainy season. The Day River (15 km to the west of Hanoi) is an important tributary of the Red River, and a strategic branch of the Red River in terms of water discharge during the rainy season and in terms of water supply to the low-lying agricultural fields downstream during the dry season. The Day River used to function as a flood diversion and retention area for the Red River to protect Hanoi’s center, as it forms a side-branch controlled by a spillway (Hat Mon), an old sluice (inoperable) (Van Coc), a new sluice (Cam Dinh), a dam (Day Dam) with six old gates and three new gates, and a new Cam Dinh – Hiep Thuan canal. Today, the Day River is considered a ‘dead river’ from the Day Dam, located 10 km downstream

The Evolution of Water Management in the Red River Delta

375

from its confluence with the Red River (Fig. 2). There is no longer enough flow in the river, and navigation is impossible, only leaving it meaningful for irrigation. Therefore, the main challenge for the Day River is to re-articulate its watercourse and return it to a healthy, flowing river. Its large adjacent agro-aquaculture area lies significantly lower than the Day River’s flood plain, yet suffers from severe droughts during the dry season. In the rainy season, the area is flood-prone. As well, the area is in the midst of rapid transformation from productive territory with villages towards an urbanized periphery of Hanoi. A great number of urban-industrial development projects and infrastructure are being carried out in the area which are dramatically altering the natural landscape and the existing water management; inevitably large areas will be replaced by hard surfaces that will accompany new development in the coming years.

Fig. 2. The Day River has been become a “dead river” since 1937, when the Day Dam was constructed. It results an extreme water pollution and lack of water during the dry season. [Pham 2011].

According to the Ministry of Construction, it is crucial to create an open landscape and a fresh environment for the Day River, considering the orientation of development of Hanoi towards the west [12]. Therefore, research is required in order to create renaturalization strategies for the Day River and develop new infrastructure, which works hand-in-hand with urban development and increased aqua-agricultural production to simultaneously address growth, innovation, economy and ecology.

2 Evolution of the Water Management in the Red River Delta It is essential to understand the Red River Delta’s historical water management in order to comprehend the larger context and eventually draw lessons. Vietnamese customs and habits in the Red River Delta remain strongly tied to both water management and agricultural traditions of mono wet-rice [20]. The delicate relationship between human and nature structures both the physical environment and the cultural landscape shown as evident in the national literature as well as legends in Vietnam. Over millennia, water management has been improved as a seasonally complex system, which includes extensive river and sea dike and canal networks, to both regulate water and supply water for irrigation and protect both agricultural fields and settlement (Fig. 3). During the rainy

376

T. A. Pham and K. Shannon

Fig. 3. The entire delta is protected by a dense river and sea dike system. Without this dike system, two-thirds of the Red River Delta would be inundated in the monsoon season. (Pham 2010, adapted from [7, 25]).

season, river intakes of dikes are closed to prevent fields from flooding, while the main canal networks collect surface runoff from fields, which is then discharged into rivers downstream. During summer droughts and dry seasons, irrigation water is provided from

The Evolution of Water Management in the Red River Delta

377

the rivers. The main canals network acts as a reservoir system and conveys the water from the intake into the whole irrigated area [4]. During the pre-colonial period (from 939 to 1858), the water management was controlled and maintained by villagers. Rules and regulations of individual water accesses and maintenance of collective reservoirs were stipulated in customary law [4]. Generally, local people initiated small projects in their dwelling areas while larger territorial-scale projects were implemented under the guidance of the State and financed by taxes and inspected by Mandarins [10, 11, 14, 20, 24, 30]. Historically, the initial regional dike system was firstly mentioned in ancient Chinese documents [24]. However, after some centuries of development, the floodwater discharge was difficult due to this regional dike system which become an obstacle. Therefore, humankind concentrated to develop the systems of river dikes, that were particularly well-suited to address seasonal flooding. According to ancient annals, the first citadel (which worked as a dike) was erected in swampy lands in Hanoi in 767 AD. In 1077, the first river dike that was built and managed by the State was along the Nhu Nguyet River (now the Cau River). Building dike and hydraulic works were amongst the most important State concerns as evidenced in the Royal Proclamation of 1103 [24]. In 1248, a low soil-dike system along the main rivers of the entire delta was built. This dikes’ network played significant roles both in protecting the paddy fields from water surges and controlling monsoon water flows into the fields. It was smart control to deposit nutrient alluvial into the delta during the floods. The soil-dike system along the tributaries of the main rivers began construction in 1503 [24]. A gravity-based culverts system was tied to these dikes. They could open to allow water to flow into the fields during dry seasons, which were closed to protect the fields from excess water during the floods [24]. In 1472, the first sea dike was constructed along the coast [15]. Moreover, already in the 11th century, a number of large canals were dug for irrigation networks, such as Dan Nai Canal (1029), Lam Canal (1050), and Lanh Kinh Canal (1089) [17]. The hydraulic works and dike system were repaired yearly by compulsory participation in public works. A lot of people including students and soldiers was mobilized to move massive volumes of earth and to build bamboo embankments in order to reinforce the dikes system [14, 17, 20, 24] (Fig. 4). Nevertheless, after several centuries of water management and flood control, the Vietnamese people understood that there was a limitation of continual heightening of the dike system. It needed to be paralleled with lowering the water level during the floods. Therefore, the digging of new rivers to discharge floodwaters and reduce water levels was begun in 1729 [24] combined with the construction of reservoirs, dams, the widening and dredging of the existing rivers system in 1857, and even the destroying the dike system in the delta [18, 20, 24]. The first Vietnam’s complete water management policy was published in 1809. In this document, a hierarchical dike system based on river sizes was regulated clearly (see [24]). The document documented indigenous dike construction, which combined wood, bamboo, and soil. To compact soil of the dike, elephants were utilized, and finally, the dikes system was covered by grass [24]. However, the proclamation was quite generic. It did not specify any different rivers’ characters as well as their water levels. The highest water levels in various rivers were only mentioned when the 1838 annex was issued [24].

378

T. A. Pham and K. Shannon

Fig. 4. Pre-colonial dike repair and public works: The mandarins directed the population to repair the dike system and hydraulic works. In the instance above, the work is carried out on a dike in Thai Binh, Red River Delta. [ Source: Dépôt des archives d’outre-mer, Aix-en-Provence: BB/SOM//D3685].

With French colonization, there as a significant shift in flood control and water management in the entire Red River Delta. The dike system had become a substantial concern from both French and Vietnamese regarding to water management and flood control in the whole delta (Fig. 5). There were so many solutions debated in various degrees, such as upstream afforestation, construction new reservoirs, diversion of water from the Red River into the other its tributaries, complete canalization of the Red River and even destruction of the river dike system. These solutions had also been discussed often in the pre-colonial period [24] (Fig. 6). Ultimately, the French intensified and rationalized the whole infrastructural system including the dike system in the Red River Delta. There were 30 primary hydraulic units in the entire dike network system with respect to independence of water management and flood control [4].

Fig. 5. Evolution of dike’s typology: Ever since French colonization, the dikes in the Red River Delta have been almost continuously heightened and strengthened. (Pham 2010, updated from [5, 27]).

The Evolution of Water Management in the Red River Delta

379

Fig. 6. To canalize the Red River was proposed by Nguyen Canh in 1920s, aiming to improve the ability of water discharge to the East Sea in the rainy season. (Pham 2010, redrawn from [13]).

The quantity and quality of the dike system were increased significantly by using mechanical techniques which was the first-time implementation in Vietnam (Fig. 7). At the same time, techniques in dike construction were regularly improved and in 1926, a new profile (the step dike) was introduced. For the first time in Vietnam, new materials such as concrete was used to constructed modern waterworks, for example: Lien Mac Sluice (1937), Day River’s Dam (1932), and Vinh Yen Dam (1896), to control water that flowed into the Red River’s tributaries as well as into the delta strategically [22].

Fig. 7. In the 1930s, for the first time, heavy construction machinery was introduced in Vietnam which had increased the quantity and quality of the dike system in the Red River Delta [5].

There was massive destruction of the dike system in the Red River Delta during the Vietnamese struggle for independence from the French. Therefore, from the end of 1945

380

T. A. Pham and K. Shannon

until early 1946, the State mobilized an exceptional social labour to repair the dike system, see [17, 24]. More than 80% of the State direct investments were dedicated to water management improvements. Irrigation schemes and large drainage were constructed to complete a complex canal network classified in four levels [4]. However, the achievement of hydraulic works improvement and development was interrupted seriously during the First Indochina War. Once again, the method of water management and flood control in the Red River Delta changed siginificantly under the political framework of agriculture collectivization and cooperative settlement under the independent North Vietnamese State in 1954. After establishing a Ministry of Hydraulics in 1958, water management became a high priority, especially with respect to flood control in the Red River. Due to receiving the specific assistance of Chinese and Russian experts during this period, the development of hydroelectric plants was carried out in many studies which has purposes for both flood control and industrial development. At the same time, the system of river and sea dikes was substantially upgraded [24]. Nonetheless, during the Second Indochina War, the hydraulic works in the Red River Delta were again destroyed and interrupted seriously. Destruction the dike system become a strategic activity of the American military. This ambitious military action was impacted significantly the fledgling Vietnamese State. It also emphasised the vital importance of the dike system in the Red River Delta for both politics and economics in Vietnam. After receiving a devastating flood in 1971, it was concluded that flood control as well as water management for the Red River Delta had to be rethought fundamentally. Parallel with strengthening and heightening the dike system, a number of flood retention basins were created to divert floods from the Red River. At the same time, large-scale engineering solutions were concerned significantly including to plant new hydroelectric plants [24]. With the help of Russian experts, a largest hydroelectric project in Vietnam, the Hoa Binh Hydroelectric Plant, had been constructed from 1979 to 1994. It was also the first important one on the Red River Delta and contributed substantially to the Red River’s flood management.

3 Contemporary Water Management Challenges for the Red River Delta In 1986, due to recognizing the failure of the “great socialist agriculture” policy, the Vietnamese government decided to change the policy through economic liberalization (called “doi moi”) and reorient it towards a “socialist-oriented market economy”. Both cities and the countryside were radically transformed due to the territorial ordering was inseparable from the logics of settlement structure and production. At the same time, water management was changing in order to respond to a series of new challenges. Since “doi moi”, there has been massive rural to urban migration and the country witnessed a 3.17% average annual urban population growth from 2000 to 2019 [6]. As its unprecedented modernization and urbanization continue, both urban and rural territories are transforming at a scale and scope previously unseen. This fragmented and dispersed development reveals a challenge that is increasingly difficult to control. New settlement colonizes both agricultural low-land and floodplains everywhere in the Red

The Evolution of Water Management in the Red River Delta

381

River Delta. Land-filling on the low-lying land is changing significantly the permeability of the territory and affirming many pressures on existing water management. There has also been an explosion of unauthorized settlement outside the dike system where is unsafe and floodable areas. These areas and their populations have become the most vulnerable to the impending effects of predicted climate change. These settlements, mostly built by the poorest sector in society, are now part of the speculative market games of the rising entrepreneurial class. The dense occupation has led to the conversion of the permeable and natural flood plains into hard surfaces, seriously compromising natural discharge and increasing flood levels. Furthermore, the region’s aging water treatment system has been overloaded with a swift urbanization that brings polluted water in the whole delta, particularly in the large cities. Vietnam is one of the most vulnerable countries in the world due to predicted climate change (and more specifically to sea-level rise) [2, 21]. The effects are expected to be the most severe in the whole deltas in Vietnam including the Red River Delta. Already, the monsoon regime has become increasingly complicated: there is less water and unprecedented heavy rains in the dry season but more water in the rainy season. Additionally, the dike system retains floodwater that carries massive amounts of sediment during the time of floods. Therefore, the sediment in the floodwater cannot enter the delta; it drifts to the mouth of the main rivers and settles in their estuaries, blocking water from flowing to the East Sea. Otherwise, water locked inside the dike system causes further rising of rivers’ water levels, increasing the risk of dike failures. The risks of the predicted rising sea levels (inevitably reducing rivers’ capacity for water discharge), coupled with higher river floods (due to stronger storms and heavier rainfall) are inevitable. Moreover, in case of heavy rains, the water accumulating in the territory’s expansive lowlands cannot naturally drain due to a higher elevation of the riverbed. Finally, the frequency and severity of floods have been increasing. There was witness happened in November 2008, when Hanoi become a victim of the most extreme inner-city floods to date. There was a devastating death toll of 94 lives and damages were estimated at VND 7.3 trillion (USD 430 Mio.). As well crops and livestock were destroyed, including 210,000 ha of vegetables, 30,000 ha of rice, 10,000 ha of orchards, 40,000 ha of fishponds and nearly 200,000 livestock [3]. Additionally, before 1970, the frequency of heavy rains and floods in Hanoi occurred every 15–25 year. However, due to climate change, over the past 60 year, floods have become more frequent with a frequency of 5–7 year [16]. The Red River Delta is also confronted with many problems by the upstream reservoir and dam construction, both in Vietnam and outside of Vietnam (by China). These reservoirs and dams are needed for both electric power and water management. As throughout the world, the effects of the large engineering works are hugely controversial on local populations, whereas the flows of water are highly manipulated, and often there is a reduction of water for agro-aqua-culture. A number of reservoirs constructed in China reveal many troubles related to watershed management, which work irrespective of national boundaries and politics (Fig. 8). The Red River Delta is the mouth for water discharge of the whole hydraulic system of the Red River Basin into the East Sea. There is no transparency between the Vietnamese and Chinese governments with regards to water management due to contentious politics; the result is that the safety in the delta is endangered since water flows remain veiled. To date, the Red River Delta suffers the

382

T. A. Pham and K. Shannon

consequences, such as flood and drought, water pollution… of being the outlet of the Red River. Otherwise, the river fluctuations are no longer decided by both the vagaries of natural regime and the politics/economics of its neighbours (similarly by China’s and Laos’ contemporary dam building in the Mekong Delta).

Fig. 8. According to Ha 2010, on the upstream Red River system, China has a plan to construct 52 hydroelectric plants (23,24,5 and 2 dams that are on upstream Thao River, Da River, Lo-Gam River, and Bang Giang-Ky Cung Rivers in respectively), of which 8 dams were complete with the capacity of water restoration being about 2.0 billion m3 of water. [Pham 2011, addapted form google map, 1, 8].

Finally, once Vietnam entered its “doi moi” period, flood control and water management has become a part of the interests and investments of the larger international (donor and for-profit) community. A large number of Vietnamese experts have been given the opportunities to study abroad; at the same time, many international experts have become advisors in Vietnam, as in the French and Soviet-influenced eras. A number of studies, particularly in the Hanoi area, have continued with the old paradigm of hard engineering. It is clear that the shift to combine hard- and soft-engineering and to “make space for the water” and to “give room to the river” has not yet genuinely reached Vietnam. However, such notions are increasingly spoken about in academia and selected governmental departments.

The Evolution of Water Management in the Red River Delta

383

Between 2004 and 2006, “Pilot Project C: The Improvement Plan and Strategy for the Outside-of-Dike Area”, as a part of HAIDEP project (Hanoi Integrated Development and Environmental Program – The Comprehensive Urban Development Programme in Hanoi Capital City), studied 40 km long of flood plain area along the Red River’s banks in Hanoi, where has been illegally appropriated by housing development during the past decades [9]. The project investigated and proposed a new strategy for development of this area through a number of alternative scenarios. The research recognized that the area outside the Red River’s dike system is a precious asset for the entire city because of its historical and landscape value, as well as prevention capacity with respect to the natural calamities. They also value it for the potential of urban development, which could be argued to go against the environmental principles and the notion of giving more space to the water dealing with floods and the predicted consequences due to climate change (particularly increasing water in the rainy season). In this project, a second dike closer to the river was proposed to enhance the potential development of the areas outside the existing dike system and to secure citizen’s quality of life. To face up to the waves of swift urbanization, the study suggested that, in the future, the location of existing cultural areas and urban communes outside of the dike system would be allowed to have “controlled development” behind the second dike system. All the rest of land outside of the dikes system would be preserved as open space for recreation and agricultural production. Security was a high priority in the research and residential areas along the river terraces, around bridges, nearby the dike roads system would be relocated for safety concerns during the flooding times [9] (Fig. 9). From 2006–2007, a cooperation research between the capital cities of Hanoi and Seoul studied for new urban development along the Red River in Hanoi (RRPT). The project investigated how to integrate the flood control with new urbanization which can compromise the riparian recreation, economic impetus, and its identity. A suggested mega infrastructural system including the dike system with large arterial roads became the backbone of the study. There was a new incredible high-rise was grafted on the infrastructural system in a combination of new waterfront landscape system along the entire length of the Red River flowing through Hanoi [28]. A second dike system was also proposed as the same as the Japanese HAIDEP project’s strategy. However, this dike system was much more intrusive and the flood plain of the Red River was radically reduced in width (Fig. 9). As compensation, the banks of the river were excavated in some places, but the overall space for water was lost and replaced by new urban development. Inevitably, the risks of flooding would be exacerbated by land-filling [27]. Finally, in 2010, the Vietnamese themselves, through the Institute of Water Resources Planning (IoWRP), also made a proposal for the flood control project for Hanoi’s entire river system. Their project ambitioned to upgrade the whole dike system according to a new classification for river dikes (see [19]). A series of concrete embankments were suggested to construct and improve where there exists the most vulnerability resulting from the Red River’s flooding. Additionally, new landmarks of floodwater discharge points as well as new positioning dikes around dense settlements of the Red River and Duong River for flood control were proposed. All three projects relied heavily on hard engineering and mere protection from rising waters, rather than strategies of accommodating floods (Fig. 9).

384

T. A. Pham and K. Shannon

Fig. 9. The Japanese (A) South Korean (B) and Vietnamese (C) proposals for reconfiguring the dikes along the Red River all compromise severely the natural ecologies and dynamics of the water flows for the urban and economic development (Pham 2010, adapted from [9, 19, 27]).

4 Expansion to the West: The Case of Chuc Son 4.1 Contemporary Water Management Challenges for the Day River Based on the evidence receiving during fieldwork, since the Day River has been cut off from the Red River, the difference of water levels between the rainy season and dry season is only 3–4 m (while it is approximately 13 m on the Red River itself). The reduction of water flow has caused it to become an ecologically dead river. The naturally reduced water flow is compounded by floating agricultural production and natural floating plants, particularly fast-growing water hyacinth. Moreover, little control by authorities means that there is more freedom regarding the cut-and-fill activity of land on the flood plain. The new land is elevated for settlement as well as the planting of trees along riverbanks. All these activities further obstruct water flows and the entire water regime of the flood plain.

The Evolution of Water Management in the Red River Delta

385

Nonetheless, the Day River still retains its beauty and diversity as both a natural and artificial landscape. Over time, different uses of water and farmland transformed the riverbanks into a tranquil rural landscape [22]. Most of the area for rice cultivation is outside the dike system, while most of the area for vegetable cultivation is inside the dike system (on the regional flood retention areas). However, rice productivity is low due to poor soil quality and flood risk. The region’s paddy fields are idle from the beginning of October to the middle of January. However, many farmers use their paddy fields for other purposes such as fish ponds. Over the past decades, the landscape of Chuc Son has evolved into a rich landscape mosaic of aqua- agriculture due to the ingenuity and economic ambitions of local inhabitants who have developed different methods for irrigation and utilized rainwater harvesting, wells and ponds to work with the seasons. Historically, there was a clear hierarchical irrigation system in the territory, which included the Nhue, Day, and Tich Rivers. Today, water moves throughout the Red River Delta with extensive mechanical means; a series of pumping stations along the Day River dike supply water for irrigation and discharge waste and flood water. From the Day and Nhue Rivers, water is pumped into the main canals or smaller natural rivers, which then naturally flows downstream (by smaller irrigation systems) into the hinterlands. Because of the scarce and polluted water in the Day River, occasionally clean water is pumped from the hinterlands into the flood plain of Day River for irrigation during the dry season. Otherwise, famers only use black water (mostly from Nhue River and traditional agricultural production) for paddy fields. For vegetables and grain, they use either rainor well-water (from either individual or collective wells). However, these activities lead to the risk of underground water becoming polluted since this area is often flooded. During floods, water that carries toxins or harmful substances mix with underground water through the wells. The quality of underground water is thus dramatically reduced. In August 2008, the Prime Minister approved the area expansion of Greater Hanoi; there was almost a twofold increase in population (from 3.8 M to 6.2 M) and more than threefold increase in the territory (from 920.97 to 3,344 km2 ) [29] with an average density of 1,863 persons/km2 . Evidently, the environmental balance of the large territory is slated for major upheavals. There are five satellite cities and three “ecological towns” planned. As a result, the existing and very large low-lying natural-agricultural land will be replaced by new urban development. This, in turn, raises many new challenges concerning sustainable development, with new infrastructure demands, increased housing density, lost open space (green and water space) and degradation as well as homogenization of the rich environmental diversity of the region. Hence, there is not enough space for water and significantly reduced permeability of the land. Chuc Son, the southern area of the Day River, is a representative case in the transformation of the territory in the western expansion of Hanoi. It is obvious that the dike system of the Day River continues to have an important role in the relationship of settlement, productivity and infrastructural development. Typology of the Day River dike is quite simple. Since the severe flood in 1971, the dike system along the Day River has been carefully studied. While the existing left-bank dikes of the Day River were widened, strengthened and heightened, a series of new dikes on the rivers right bank was constructed on suitable for flood retention areas of the Chuc

386

T. A. Pham and K. Shannon

Son (as realized under a 2011 Prime Minister’s Decree) [26]. The new dikes were constructed further to the west and incorporated the existing mountains as a natural defence system. Moreover, a number of new spillways were created by lowering existing dikes downstream (Fig. 10). The dikes along its banks are smaller than those of the Red River dike but have innumerable more (in)formal functions and activities. Due to the combination of a number of contradictory elements (natural/artificial topographies, commercial areas/settlement and aqua- and agricultural/industrial production areas), a diversity of landscapes and spaces along the dike system exist. In addition, with the new demands of the free market and diversified economy, there is an increase in local demands for space, which is as close to the dike and infrastructure system as possible. A process of elevating land up to the height of the dike system is presently occurring in the area, which causes fundamental changes in both the function of the landscape and the dike system. New challenges in terms of water management require annual strengthening and frequent stability checking of the dike system will continually need to be dealt with.

Fig. 10. Typology and movement of dike in Chuc Son: The profile of the Day River dikes is similar to others in the Red River Delta and historically were adopted to the dynamic flows of the river’s flow [22, 23].

Between 1907 and 2005, there was little change in the urban footprint of Chuc Son. Afterward, however, there was a swift change of both rural and urban morphologies. A great number of projects for urban and industrial development as well as water management were carried out on the left side of the Day River dike and territory will

The Evolution of Water Management in the Red River Delta

387

inevitably continue to change dramatically in the coming years. The agro-natural landscape will continue to be replaced by the hard surfaces of new development. A larger looming threat than the legal construction (including administrative buildings, schools, markets,…) made by local government is the illegal constructions inside the dike system that fills the floodplain and further alters the already severely damaged hydrology of the region (Fig. 11).

Fig. 11. Chuc Son as envisioned in expansion plans: As Hanoi expands, so to does its agricultural hinterlands—at least in the dreams of the planners for the new master plan for the capital city. Chuc Son is slated to become an “ecological urban area” between the Day River’s flood plain and its hinterland. (Pham 2020, adapted from [12]).

388

T. A. Pham and K. Shannon

4.2 The Case of Chuc Son The case study area of Chuc Son has relatively flat topography. Most of the land is a flood plain of the Day River, which is a bit higher than its hinterland due to sedimentation (Fig. 12). This character is one of the reasons why water flows quite easily from the flood plain into hinterland but, at the same time, has difficultly discharging back into the river in the rainy season. The little water in the Day River can hardly flow into paddy fields in the dry season. Since a natural water flow no longer exists, a huge artificial system for water control in this area is required. The relatively flat topography has slight differentiation in levels, which, in turn, determines distinct (and often conflicting) land designations such as agricultural production/settlement, flood-prone/safe land, natural/artificial water management, etc. A diversity of agricultural land-use reflects inherent landscape logics: low-lying land is cultivated as paddy, naturally higher lands are used for more floodvulnerable crops such as vegetables, flowers, fruit trees or ornamental trees, and the highest natural (and artificial) lands are occupied by humans (residences, industries, social spaces,…). On a relatively smaller scale, the topography is often artificially modified following a ‘cut-and-fill’ logic. The earth that is cut to make ponds or components of the irrigation system is used to create raised and safer lands. The design research proposes alternatives to business-as-usual and seeks to capitalize on interplays and synergies between urbanization, infrastructure, and landscape. The vision for Chuc Son provides a crucial frame for development through enhancing the existing water-based infrastructure of the territory. The ultimate aim of the design research is to protect the flood plain of the Day River as the main water discharge area for the expanding urban area of Hanoi. Since the city is expanding westwards, and in case of failure reservoirs upstream and unpredictable extreme floods, existing villages outside the Day River’s dike system should be consolidated and expansion should be limited while existing villages inside dike system should not be allowed to develop and expand. The nature of the Day River, and more specifically address the unique character of Hanoi’s urban and rural periphery, is structured by strategic, yet flexible, a development that relies on the water system as an ecological backbone. ‘Soft engineering’ is proposed to work with the dynamics of water and reduce and mitigate the predicted impacts of natural disasters due to climate change (Fig. 13). The ultimate aim of the design research is to protect the flood plain of the Day River as the main water discharge area for the expanding urban area of Hanoi. Since the city is expanding westwards, and in case of failure reservoirs upstream and unpredictable extreme floods, existing villages outside the Day River’s dike system should be consolidated and expansion should be limited, while existing villages inside dike system should not be allowed to develop and expand. In order to counter pressure from urbanization on the eastern side of the Day River, an elastic linear park is proposed along the left dike to protect the Day River and its flood plain. The elastic nature of the green zone has to do with its capacity to respond to season variations in water levels and to act as a flood mechanism. As well, the flexible/seasonal water network is planned to addresses both water quantity (including flooding, storm/rainy water retention, drainage and irrigation) and water quality (including sewage and purification) issues and the recreational use of water. Amongst the elements

The Evolution of Water Management in the Red River Delta

389

Fig. 12. Location of the Chuc Son case study in the Hanoi Construction Master Plan to 2030 and vision to 2050 (Pham 2020).

Fig. 13. Three water retention strategies: More water space gives more potential for agroaquaculture production and reduces the unused land of paddy fields in the case study [22, 23].

390

T. A. Pham and K. Shannon

in the water system are storm water management, a decentralized waste water purification network which combines constructed wetlands and chemical treatment plants, and a system of water retentions such as: lakes, ponds and alluvial traps for agriculture. The same area would be physical structure to direct and limit urban development and protect nature along the Day River. New towns should be developed at the intersections of the main roads. This can help to reduce construction and expansion in the surrounding villages (Fig. 14). This new urbanization should be built on safer land (by inter-cut-and-fill process) on both sides, strongly related to the existing topographical conditions. Hence, a balance for development on both sides of the Day River would create an ecological (and recreational) corridor as well as part of an expanded and protected flood plain.

Fig. 14. Frame structure for urbanization: An alternative linearity along and outside the Day River dikes (with the possibility of new interplays of landscape, infrastructure, and urbanization) creates a series of different centralities [22, 23].

To emphasize the elastic green zone between new urban development and the recovered nature of the Day River and to simultaneously improve the living standard in present villages along Day River dikes, a new tramlines network (which could help to reduce the pressure of transport on the present-day dike system) could be configured for the area. Apart from melding with existing dikes to become a more functional hybrid dike

The Evolution of Water Management in the Red River Delta

391

in places, new tramlines will pass villages and paddy fields with the aim to link existing villages together as much as possible and connect to region’s crucial infrastructure. Hence, the Day River dikes will also serve more functions than mere technical roles of water management. In addition to the primary linear park along the Day River, a system of secondary linear parks along small rivers and main canals for irrigation which connect the Nhue, the Tich, and the Day Rivers could also be created. Such a system could improve not only environmental condition but also contribute to create a hierarchical system of green and open spaces for the whole Hanoi region. As well, the system both could preserve and provide new open spaces for compensation of nature for new urban development and work as links to maintain an ecological balance and create gradual alteration between urban and rural areas. Waste water management is also a very important issue in the case study; particularly since urbanization is juxtaposed with the Day River. At the confluence with smaller rivers and with the meetings of villages (traditional agricultural production), such as the So village (one of the main sources of waste water discharging directly into the Day River), aerated/green lagoons could be located (combined with the secondary linear parks in one). Such lagoons could be integrated into the flood management for the plain and work as both flood mitigation devices and retention basins.

Fig. 15. Diagram sections guiding urbanization: Existing condition (A); expected urbanization in Hanoi Construction Master Plan to 2030 and vision to 2050 (B); new strategies across new urban district (C) and lagoon (D) [23].

392

T. A. Pham and K. Shannon

A combination of re-naturalizing the Day River, preserving its flood plain, and creating a water retention system is one of main solutions to protect Hanoi in water management. The ecosystem of the Day River and its flood plain will become an important ‘green lung’ for Hanoi’s sustainable development. In terms of urbanization, on the eastern side of the Day River, a dispersive development of new urban districts on the low-lying agricultural lands lead to increasing risks of inundation due to the proliferation of hard surfaces. Therefore, it should be developed more compactly to save land for blue-green spaces, which are used for urban agriculture, recreation, decentralized bio-filters, or seasonal/flexible water areas (Fig. 15). New ecological towns proposed in the Hanoi Construction Master Plan to 2030 anh vision to 2050 on the other side should develop outside the Day River’s flood plain. Parallel with the water retention system on the hinterland could be another system along the Day River banks to increase the capacity of retaining water in the area during the monsoon season. It would be an opportunity to improve waterscape and also work as a system to trap fluvial during floods to improve the quality of the soil. However, construction of new forms for water retention would cause complicated water flows and therefore the forms of the traps need to be studied carefully in order to protect riverbanks in terms of hydraulics.

Fig. 16. Space for the Water: Interpretative mapping is not merely descriptive but mirrors realities. By the form of collage, it reduces to the essence and unfolds the hidden potentials and discloses conditions to emerge new realities that are suitable with new strategic interventions. Water retention along the Day River is an opportunity to improve the landscape, water management, and agricultural production [22, 23].

The Evolution of Water Management in the Red River Delta

393

At a larger territorial scale, the Day River flood plain is extremely important for flood diversion and retention in case of extreme failures of upstream Vietnamese reservoirs and for unexpected water coming from China. As part of the proposal, a series of new water retentions basins would also be created. The locations and configurations of the basins would be based on the existing elevations of land and water demands for irrigation on areas in the dry season; they would also serve as rainwater harvesting basins and thereby improve the aqua- agriculture economy of the region. Indirectly, the quality of soil due would also be improved due to the fluvial in the flood water, increasing areas for vegetable and aquaculture by reducing unused land of paddy (Fig. 16).

5 Conclusion The water management and dike building traditions for flood control in the Red River Delta have a long and distinguished history. Historically, people had controlled and maintained the dike system by the local peasantry, which revealed an intimate understanding of the logics of the delta’s landscape and its dynamics. Mankind had reclaimed the Red River’s water regime with quite low techniques, simple means, massive human endeavour and ingeniousness. Over time, acquired both indigenous and imported knowledge and new techniques, water management in the Red River Delta has become more refined and more complicated. Nonetheless, nowadays, a combination of existing position of urbanization, environmental crises, predicted consequences of climate change as well as the uncontrollable up-steam river constructions by neighbouring countries, has brought these traditions of water management into a new era that requires a fundamental rethinking. The Day River is one of the most important natural rivers in the Red River Delta regarding both water management and environmental protection. It is a specific and fragile spatial edge of Hanoi’s development and in the midst of rapid transformation from an aqua- agricultural territory with traditional villages towards a fully urbanized periphery. It is receiving massive pressure to urbanize and infrastructure construction has already begun. However, since it has ceased to exist as a functioning ecological entity it demands urgent attention. Its ecological integrity needs to be recovered and it should be reconfigured to work as Hanoi’s ‘green lung’. Through the lens of the water urbanism, the research results of the Chuc Son case study suggested several solutions that can become lessons for others areas in Hanoi as well as other cities in Vietnam. There is a need of multi discipline co-operation, such as urban planning, water management, and landscape architecture… to enhance the environmental and landscape quality, values of agro-aqualcutural production, and quality of citizens’ life. They can be adaptive solutions to deal with the Hanoi’s future growth and predicted climate change.

References 1. Dang, Q.T.: Urban floods – Challenge, presentation at World Conference on Disaster Reduction, Kobe: 18–22 January 2005 (2005)

394

T. A. Pham and K. Shannon

2. De Niijs, A., Pham, A.T.: Rising the dykes & taming the swamp: water management in Vietnam’s red river & mekong deltas. In: De Meulder, B., Shannon, K. (eds.) Water Urbanisms - East, pp. 208–221. Park Book, Zurich (2013) 3. Development Workshop France: Viet Nam 2008: a year of unpredictable economic volatility, 30 December 2008 http:/www.dwf.org/blog/lists/Posts/Post.aspx?ID=213. Accessed 19 Apr 2009 4. Fontenelle, J.: Vietnam red river delta irrigation management: incomplete recognition of local institutional innovations. Working Paper No. 27. GRET, Paris (2001) 5. Gauthier, J.: Digues du Tonkin. Imprimerie D’Extrêm e-Orient, Hanoi (1930) 6. General Statistics Office of Vietnam: Area, population and population density by province by Cities, provinces, Year and Items (2019). https://www.gso.gov.vn/default.aspx?tabid=714. Accessed 8 Aug 2019 7. Gliedman, J.: Terror from the sky-North Viet-Nam’s dikes and the U.S. Bombing. Vietnam Resource Center, Cambridge (1972) 8. Ha, V.K.: Synthesis Report concerning the scientific results of the basics scientific study for abrogating flood retentions along the Red, Day, and Hoang Long rivers (DHTL 2008/34G). Thuy Loi University, Hanoi (2010). (in Vietnamese) 9. Hanoi Integrated Development and Environmental Program (HAIDEP): Pilot Project C: Improvement Plan and Strategy for Outside-of-Dike Area, The Comprehensive Urban Development Programme in Hanoi Capital City of the Socialist Republic of Vietnam. Hanoi (2007) 10. Le, B.T.: Vietnam - the country and its geographical regions. The Gioi publishers, Hanoi (1997) 11. Logan, W.S.: Hanoi Biography of a City. University of NDW Press Limited, Sydney (2000) 12. Ministry of Construction (MoC): Hanoi Construction Master Plan to 2030 and vision to 2050 Ministry of Construction Report, Hanoi (2010) 13. Nguyen, C.: L’hydraulique fluviale & solution de la question du fleuve rouge, tome II. Géomètre Diplômé du Gouvernement, Nam-Dinh, Tonkin (no date) 14. Nguyen, D.N.: Do the urban and regional management policies of socialist vietnam reflect the pattern of ancient mandarin bureaucracy? Int. J. Urban Reg. Res. 8(1), 73–89 (1984) 15. Nguyen, H.K.: The Hong Duc Dike and salty land reclamation along the coast areas on the south of the Red River under the early Le Dynasty. NCLS 5(224), 35–42 (1985). (in Vietnamese) 16. Nguyen, H.T. Cities confront challenges of responding to natural disasters (2018). https:// www.nhandan.com.vn/cuoituan/item/38426802-ðo-thi-truoc-thach-thuc-ung-pho-thien-tai. html. Accessed 28 Apr 2019 17. Nguyen, K.V.: Vietnam-a long history 7th revised and, expanded The Gioi Publishers, Hanoi (2007) 18. Nguyen, V.P.: The low river, the high dike. In: Nguyen, V.P. (ed.) Land, People, Hanoi, pp. 49–65. Youth Publishing House, Hanoi (2009). (in Vietnamese) 19. Institute of Water Resources Planning: Detail planning of flood control for the rivers which has dike system in Hanoi to 2020, General Report [in Vietnamese]. Institute of Water Resources Planning, Hanoi (2010) 20. Pham, A.T., Shannon K.: Water management in Vietnam: indigenous knowledge and international practices: the case of the Red River Delta. In: N-AERUS XI, Urban Knowledge in Cities of the South, pp. 285–301. Brussels (2010) 21. Pham, A.T., Shannon, K.: Urbanization and Climate Change in Vietnam: A Case Study of Hanoi. In: International Symposium Developing Countries Facing Global Warming: a PostKyoto Assessment, pp. 203–222. Royal Academy for Overseas Sciences, United Nations, Brussels (2010)

The Evolution of Water Management in the Red River Delta

395

22. Pham, A.T.: Recovering the Day River, Hanoi, Vietnam: Interplays of infrastructure, urban development, and agricultural production. In: 6th International PhD Seminar Urbanism & Urbanization: The next Urban Question - Themes, Approaches, Tools, pp. 242–253. Venice (2011) 23. Pham, A.T.: Water Urbanism in Hanoi, Vietnam: An Investigation into Possible Interplays of Infrastructure, Urbanism and Landscape of the City’s Dyke System. Ph.D. dissertation. KU Leuven (2013) 24. Phan, K., Tu, M., Nguyen, G.Q.: Dike system in Vietnam (Brief History) [in Vietnamese]. Agriculture Publisher, Hanoi (1995) 25. Pouyanne, A.A.: L’Hydraulique Agricole au Tonkin. Atlas. Imprimerie D’Extrêm e-Orient, Hanoi (1931) 26. Prime Minister’s Decree: No.04/2011/NÐ-CP on 14 January 2011 concerning the abrogation using flood diversion areas and flood retention areas on Red River System, Hanoi (2011) 27. Red River Project Team: Basic development Planning, Red River area–section flows through Hanoi, main report [in Vietnamese] Red River Project Team Report, Hanoi (2007) 28. Shannon, K., et al.: Hanoi: The City of Lakes and Rivers. In: De Meulder, B., Shannon, K. (eds.) Water Urbanisms - East, pp. 92–117. Park Book, Zurich (2013) 29. The National Assembly of the Socialist Republic of Vietnam: Resolution No 15/2008/QH12 on 23 May 2008 adjusting the administrative boundaries of Hanoi and some provinces, Hanoi (2008) 30. Tran, H., Nguyen, Q.T.: Thang Long-Hanoi throughout its ten centuries of urbanization [in Vietnamese]. Construction Publishing House, Hanoi (2004) 31. Tran, L.D., Vu, T.T.L., Hoang, T.S.: Resources, water in Hanoi and exploiting and utilizing problems, Conf. Utilizing socio-economic, resource and natural potentials in the process of sustainable development and urbanization of Hanoi’s region (October Hanoi) (2006). (in Vietnamese) 32. Vinh, V.D., et al.: Impact of the Hoa Binh dam (Vietnam) on water and sediment budgets in the Red River basin and delta. Hydrol. Earth Syst. Sci. 18, 3987–4005 (2014)

Building Climate Change Resilience Indicators for the Rural Commune in the Northern Delta, Vietnam Toan Duong Thi(B) , Duc Do Minh, and Luu Tran Thi University of Science, Vietnam National University, Hanoi, Vietnam [email protected]

Abstract. The rapid development and change of agriculture in the context of climate change have caused significant environmental impacts in rural areas. This paper has the purpose to build a resilience indicators set and a framework to quantitative the resilience value to the rural commune in the North Delta, Vietnam. These indicators respond to the requirements of economic development, sustainable livelihoods, environmental issues, and increased resilience to climate change. For this purpose, a database was built based on questionnaire results, monitoring data collecting, and expert’s elicitation. Finally, these resilience indicators set was built with three capitals (natural, social and infrastructural capital), 17 sectors, and 50 indicators, as following: (i) the natural capital includes 1 sector as land use having 11 indicators of types land-use area; (ii) the social capital includes 9 sectors, and 19 indicators belong to Income, Poor rate, Employed labor rate; Education, Cultural, Health, Environment, Policy; (iii) the infrastructure capital includes 7 sectors and 20 indicators as components of Road condition, Irrigation system, House, Electricity; Community Facilities; Communication systems, and Energy. A framework for calculating resilience value also was built. The proposed resilience indicators set and the calculated framework was employed for two communes as Hai Dong (coastal commune, in Nam Dinh province) and Lam Dien (non-coastal commune, in Hanoi). The resilience results are 0.64 and 0.58 for the two communes, respectively. The presents detailed processes of the resilience indicator building and calculating should be suggested to apply widely in the rural communes in Vietnam. Keywords: Resilient climate change · Rural commune · Natural capital · Social capital · Infrastructure capital

1 Introduction There are some types of indicators such as climate change indicators, climate impact indicators, climate adaptation indicators, vulnerability indicators, and resilience indicators which are used to indicate the ability of resilience of a system to climate change and disasters. Resilience was firstly envisioned by Holling (1973) and elaborated upon © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 D. Tien Bui et al. (Eds.): Proceedings of the International Conference on Innovations for Sustainable and Responsible Mining, LNCE 108, pp. 396–428, 2021. https://doi.org/10.1007/978-3-030-60269-7_20

Building Climate Change Resilience Indicators for the Rural Commune

397

by natural and social science researchers provides a way to think about developing and managing social-ecological systems as integrated systems. The other element relates to human activities (e.g. institutions, infrastructure, culture), and the environmental elements (e.g. geological, climatological, biological) were included to create a coupled complex system. The programs and projects of development practitioners and agencies are often aimed at improving these complex systems, which include overlapping and interacting geographic, administrative, and environmental factors at different, competing boundaries, and scales [1]. The resilience can be applied to any aspect and scale in a country or region. In this paper, the term of resilience as the resilience of a rural commune scale in the context of climate change, the definition of resilience will follow by the Intergovernmental Panel on Climate Change (IPCC): “The resilience is the ability of a system and its component parts to anticipate, absorb, accommodate, or recover from the effects of a hazardous event in a timely and efficient manner, including through ensuring the preservation, restoration, or improvement of its essential basic structures and functions” [2, 3]. Building resilience indicators in the field of climate change are quite late, there were no indicators of resilience to climate change until 2008 [4, 5]. Recently, resilience indicators are created in many countries, especially in the country has strong impacts on climate change [6, 7, 8, 9], and disasters [10, 11, 12, 13].Community resilience is part of national resilience, is an important part of many poor and developing countries like Vietnam. This is a rural region with main livelihood relating to agriculture and fisheries [13, 14, 15], which are subjects affected greatly by natural disasters and climate change. The indicator system becomes an effective tool to enhance the resilience of these areas. The issues covered in indicators are socio-cultural, economic, environment, infrastructure, and governance considerations. Many research works tried to improve the tools and methods or processes for building the resilience indicator sets and the method to measure or quantity the resilience value. Scherzer (2019) [16] proposed a baseline, a community resilience index, using 47 indicators to describe the resilience capacities of the Norwegian municipalities. A simple and efficient resilience assessment framework including 27 questions was proposed to assess the resilience for a community in Cambodia by Jacobson, 2020 [14]. In those indicators set, the role of every indicator was considered having equal weight, only a few research mention on the weight of the different indicators [10]. The livelihood diversification was still limited in those resilience indicators of Cambodia [14]. The indicators of infrastructure development were not covered to assess the resilience of the rural community in Bangladesh [15]. There is no reviewing and agreement on how best to measure resilience [15]. The numbers of indicators also change quite differently between countries, is from a few of 15 indicators in [17], to quite large of 87 indicators in [19]. Jacobson, 2020 [14] reviewed the resilience indicators building for rural areas and commended that almost resilience framework is quite complex to empower communities to build resilience and take ownership of adaptation efforts. Thus the numbers of indicators may not important, but the subject and scope and local condition are the main issues that need be considered when making a list of indicators. For the area were affected by natural and geo-hazards, the building of resilience dominantly mention improving the indicators of infrastructure

398

T. Duong Thi et al.

institutions, and insurance and developing technology solution such as early warning system [11, 12, 13]. Other indicators on the condition of nature, social-economic, human, and including infrastructure are often covered in the resilience indicators to climate change. To improve the methods for measuring resilience, Asadzadeh (2017) [10] suggested that the theoretical background of resilience measures should be going to solve the issues relating to the semantic completeness (why resilience), measurement focus (resilience for when), operationalized domain (resilience of what), and unit of analysis (resilience for whom). Good practice in building resilience indicators requires responding to these issues which depending on the local condition. Then the factors relating to causes (as from climate change, disasters), subjects, and level (rural community) and all their capital (nature, social-economic, human, infrastructure, technology) need to expose and cover in the resilience indicators set. In Vietnam, there are a few research building climate indicators. The first research built an indicator system as the climate adaption indicator for the natural capital [20] and the second research also built climate adaption indicators for the city model [21]. Those researches [20, 21] become effective references for building climate resilience indicators in Vietnam. The resilience indicators system in this paper is the main result in the project “Building and applying a model of low carbon and high resilience communities to respond to climate change in rural areas in the Northern Delta” [22]. The initial result in assessing the resilience carried out for Hai Dong commune [23]. These resilience indicators will be continuously updated and presented in this paper. This paper has the objective is to propose resilience indicators set and a framework to quantitative the resilience value to the rural commune in the North Delta, Vietnam. The resilience indicators set can be used to determine the resilience of a rural commune in a specific time and condition. The results in resilience value will help us to show which factor needs to be improved also as suggesting the solutions to increase the resilience of a commune.

2 The Study Area Preparation of Your Paper Figure 1 shows the study, the North delta, and the location of two communes as Lam Dien (Chuong My, Hanoi city) and in Hai Dong (Hai Hau, Nam Dinh province as two case study for calculating the resilience values. The North delta includes 10 provinces as Vinh Phuc, Hanoi, Bac Ninh, Hai Duong, Hai Phong, Thai Binh, Nam Dinh, Ninh Binh, Ha Nam. There are two river systems flow through delta as the Red River and Thai Binh River. The coastal zone area has a length of about 200 km from Hai Phong, Thai Binh, Nam Dinh. The North delta area and its coastal area have been greatly affected by climate change and annually natural disasters such as the seawater level (SWL) rise, critical storm, and typhoon, dramatic change of rain intensity as heavy rain, or no rain in longtime. The monitoring data at the stations in Vietnam show that the SWL in Vietnam rose with an average rate of 3.5 ± 0.7 mm/y, and about 2.5 mm/year in the northern of Vietnam [24]. According to the Vietnam Hydrometeorological data center [25], Vietnam has the average number of storms and typhoons as about 7–8 storm events per year, that is a number of the storm the coastline of the North delta has effects. In the storm event,

Building Climate Change Resilience Indicators for the Rural Commune

399

the heavy rain and strong wind and also caused high tide and SWL. In those numbers of storms, there is more than 50% of storms cause the SWL to rise 1 m, 30% of storms cause the SWL rising 1.5 m, and 11% of storms cause the SWL rising 2.5 m. Recently, an increasing number of the intensive typhoon and high tide level during typhoon events has been recorded, causing the sea-dike failure or increasing distance of seawater intrusion. The coastal area of the Nam Dinh area is the most dangerous by attacking annually typhoons causing strong erosion and sea dike instability. The structure and material of protected construction and sea dike are also important factors affecting the instability of the coastline. The sea dike and slope of coastal are often covered by concrete, but the underlayer or dike body is local material by the sand or sandy silt in local, that material may compact not responding standard. That dike or coastal consequently broken by the strong wave and tide [26]. The average elevation of the North delta is from 0.4Z m to 9 m above the seawater level, with nearly 60% area has elevation lower than 2m, where an area will be in the potential of the flood if not protected by strong dike system. Four provinces as Hai Phong, Thai Binh, Nam Dinh, Ha Nam have 80% are lower than 2 m. Along the Red River and Thai Binh River, the river dike system was built in long-term ago create a river mudflat which is higher 3–5 m than the main river flow. The human lives with high density in both sites of river dike and large are of the river mudflat will be in danger of flooding in the rainy reason. The river dike and riverbank where are in the high instability potential will affect human living and other activities.

Fig. 1. The study area

There are about 40% land area in the North delta used for agriculture, 25% area for the forest, 7% for human living, the left is used for other activities. Based on the percentage of land use, the area of agriculture is the largest area and is greatly affected

400

T. Duong Thi et al.

when natural disasters happen. In this paper, the resilience of a rural commune will define all components of livelihood and all capital as well as a factor in the context of the effects of climate change, natural disaster, land use, infrastructure, and all human activities. The weight and the resilience value of all the attending factors will be assessed to indicate the role of these factors, then can finding the solution to enhance the resilience of a commune.

3 The Frameworks, Approach for Building Resilience Indicators 3.1 The Framework Figure 2 shows the brief of the framework to build the resilience indicators with five main stages.

Fig. 2. The framework including the research stages for building the resilience indicators

The first stage is carried out by two processes of overviewing the current approach and leaning models on resilience indicators built in some countries. The first draft categories resilience indicators were listed, then continues shortcut by the consultancies from scientists and professional experts on climate change topic. The scope applies area

Building Climate Change Resilience Indicators for the Rural Commune

401

in this study is the rural communes in the North delta. Then all information in categories responding to the study area will be pointed out and checked. In this process, the new indicators may be added responding to the local condition. The process of exposure to the character and impact of climate change was assessed to find out the indicators and elements affected by climate in North delta. This process also needs to determine the contributed weight and the resilient index in the next stages. The methods used in the first step including the operation of seminars, workshops, and build the questionnaire sheets to overview, collect, and analyze the database. The second and third stages are the important stages in the framework of the resilience indicators building. The objective of these stages is deciding to select which indicators and elements, then building the processes to determine the resilience value. The resilience value depends on the contribution weight, resilient index, and the criteria for indicators and elements. While the contribution weight indicates the role of indicators or elements in resilience to climate change. The indicator with high contribution weight means to play a higher role in resilience. The method to determine the contribution weight and the resilience index, and criteria will be described in the next part. The stage fourth and fifth is applying the resilience indicator in some specific communes and continue to confirm the effectiveness of indicators, then assessing the potential to a wide application on a larger scale. 3.2 The Approaches Based on the Background Knowledge Although the development of the climate resilience indicators started only after 2008 [4], it seems interested in quite large researches. In most researches, there are four broad approaches to defining and operationalizing resilience: (i) Vulnerability approach; (ii) Adaptive capacity approach; (iii) Formal capitals approach; and (iv) components or determinants of resilience approach as described in Table 1. In this paper, these four approaches were considered as a background knowledge for building a resilience indicator system of the rural commune scale in Vietnam. In these approaches, all elements of a system, which are suffered the influence from climate change will be mentioned, however, with some differences in meaning. In our opinion, three latter approaches as the Adaptive Capacity Approach, Formal Capitals Approach, Components, or Determinants of Resilience Approach have many agreements and conjunction with each other. The components and capital approaches have the most similar. However, the capital approach may more clearly in the capital blocks. The approach in this paper will learn from the adaptive capacity approach, formal capitals approach, and combine with the defining of IPCC. Finally, the resilience here is the ability of a system and its parts to anticipate, absorb, recover or recover from the effects of a hazardous event and improvement its structures and functions for future climate, including surprises. Based on the Study Area Condition The methods used here as known as the hybrid resilience framework as shown in Fig. 2. The hybrid resilience includes both the theory science concepts and local conditions with its components parts (nature, social, human, infrastructure, policy management), and finally, using the quantitative methods to determine the resilience in a specific value.

402

T. Duong Thi et al. Table 1. Overviewing on the approach for building resilience indicators [4]

Approach Vulnerability approach

Adaptive capacity approach

Formal capitals approach

The objective and main content Vulnerability is considered to be a combination of exposure, sensitivity and adaptive capacity Vulnerability is reduced through improving both resistance and resilience. Resilience does not include exposure, which is the third component of vulnerability Adaptive capacity is the ability to design and implement effective adaptation strategies, or to react to evolving hazards and stresses so as to reduce the likelihood of the occurrence and/or the magnitude of harmful outcomes resulting from climate-related hazards. The adaptation process requires the capacity to learn from previous experiences to cope with current climate, and to apply these lessons to cope with future climate, including surprises It considers vulnerability as a lack of capabilities or physical, financial, social, human and natural “capitals”. Resilience becomes the opposite of vulnerability and is considered to be the possession of these capabilities or capitals The resilience to climate change followed (Elasha et al., 2005) as the combination of five groups of livelihood capitals: Natural capital: Physical capital: Financial capital: Social capital; Human capital

References (Malone 2009) [27]

Berkes and Jolly (2001) [28] Carpenter et al. (2001) [29] Elasha et al. (2005) [30] Malone (2009:6)

Malone (2009) Elasha et al. (2005)

(continued) Table 1. (continued) Approach Components or determinants of resilience approach

The objective and main content This is similar to the capitals approach. The components or determinants of resilience are defined specifically for each local area and indicators are selected based on the determinants

References

Building Climate Change Resilience Indicators for the Rural Commune

403

The methods include expert elicitation; interview sheet (questionnaires) and surveying field sheet; collect and analyze the climate impacts database, calculate resilience value. The methods of building questionnaires and collecting local data were presented in the [23], then will not remain here. The processes and methods to determine the resilience will be presented in detail below (part 3.4). Expert elicitation is a method used in aspects related to climate change based on a wide variety of information and data, including observations, model results, insights from research into underlying processes (socioeconomic or natural), and other sources. Expert panels and elicitation are thus particularly valuable in understanding climate resilience, where empirical knowledge of processes and relationships among potential causal factors is limited, minimal baseline data exist, and short-term collection of primary data is difficult. Table 2. Identification of the capitals and components Capitals

Sectors = the components of the capitals

Natural capital

Soil, water and mineral resource

Social capital

Population, Income, Poor rate, Employed labor rate; Education, Cultural, Health, Environment, policy

Infrastructure capital

Road, Irrigation system, House, Electricity; Community Facilities; Communication systems

Table 3. The climate change factors and the hazards The climate change factors

Hazards

Sea level rise – Seawater intrusion Critical storm, typhoon, extreme tidal – Flooding – Erosion coastal area – Instability of the dike system and coastal infrastructure Heavy rainfall

– Flooding – Erosion in riverbank – Instability of the dike system

Critical change in weather

– Drought – Seawater intrusion

As following in Fig. 2, the building resilience indicators are the following: Firstly, the capitals and their components were identified based on overviewing the resiliencebuilding concepts as shown in Table 2. Secondly, the database on local conditions is investigated and collected in the context of climate and covering economic - environmental policy developments. Based on the monitoring climate change, the results of researches, and the annual reports at local communes on the climate impacts [25, 31, 32, 33, 34], it can be summarized the main climate change factors and common hazards

404

T. Duong Thi et al.

happened in the North delta as shown in Table 3. The components of the three capitals identified from the resilience approach are assessed to the impacts of climate change processes such as natural hazards (seawater intrusion, flooding drought, and instability of river/coastal infrastructure). The analysis and evaluation will derive components and criteria concerning the impacts of climate change. Moreover, some requirements are added to the resilience indicators built for the rural area in the North delta. The resilience indicators have to ensure continuously three objectives: (1) socio-economic development, (2) ensuring sustainable environmental development, (3) reducing carbon emissions, and respond to climate change. The resilience indicators ensure that will not cause any trouble for people, occupational structure, and especially supporting to develop the livelihoods in the rural area. The criteria of resilience indicators are consistently developed and do not conflict with the ongoing programs for rural areas as the new rural criteria indicators; ensuring environmental conditions, decreasing the carbon emission, and impaction of climate change. Figure 3 shows the close conjoining of all components in the structure of the resilience model for the rural commune in the North Delta, Vietnam.

Fig. 3. Components in the structure of the resilience model

4 Results on the Building the Resilience Indicators Set 4.1 Proposed Resilience Indicators Following the framework in 3.1 and the approach of the capitals and components resilience approach in 3.2, the process combines between the resilience approaches, the expert’s experience, and elicitation, and the creating field database methods, the decided resilience indicator building for the rural area in the North of Vietnam as shown

Building Climate Change Resilience Indicators for the Rural Commune

405

in Table 4. This resilience indicator includes threes capital, 17 sectors and 50 indicators, as following: (i) the natural capital includes 1 sector as land use having 11 indicators of types land-use area; (ii) the social capital includes 9 sectors, and 19 indicators belong to Income, Poor rate, Employed labor rate; Education, Cultural, Health, Environment, Policy; (iii) the infrastructure capital includes 7 sectors and 20 indicators as components of Road condition, Irrigation system, House, Electricity; Community Facilities; Communication systems, and Energy (Table 4). 4.2 Guideline for Employing the Resilience Indicators Set to Calculate the Resilience Figure 4 shows the processing of calculating the resilience value using the resilience indicators. The processing includes 5 steps: – Firstly, in step 1, selecting the indicators for the case study area following the type of rural commune as the coastal or non-coastal commune. The categories’ resilience indicators are divided into three levels as adopting from the framework of resilience indicators building. The first level includes three capitals, the sectors of these capitals as the second level, and the indicators are the third level. The resilience of a commune is derived from three levels. The role of indicators plays in the resilience measure through the characteristic of these indicators in a specific area involving the impacts of climate change. – Step 2 as building local database by using the surveying data, analysis questionnaires performing at the specialized departments at the commune People’s Committee and to the local peoples living communes. – Step 3 as determining the contribution weight of elements in three levels. In these steps, expert elicitation, which is often used in climate-change research, plays an important role in deciding to determine the contribution weight of indicators. Based on the case study natural condition, the local collected database, and the experience from expert elicitation on assessing the resilience potential of elements in resilience indicators, the contribute to weight is determined as follows: For the level 1: Three capital areas considered as the same weight to resilience value. Then the weight of the capital is Watural = Wsocial = Winfras = 1/3. For level 2: the weight of sectors is deferent in deferent capital. The sectors in the natural capital as (i) topography and (ii) and using have the same the contribute weight, equal to ½. The sectors in the social capital as (i) also have the same the contribute weight, equal to 1/8. The sectors in the infrastructure capital have different weights and were determined by the Analytic Hierarchy Process (AHP) [35], which also presented in [23] then not remain here. The results for calculating the weight of sectors in the infrastructure as shown in Table 5. For level 3: determining the weight of the indicators, which depend on the characteristic of the indicators: For the indicator in the natural capital, the weights of land using or soil use condition is the percentage of the soil type using. This value will deference in every commune and time. For the indicators in the social capital and infrastructure

406

T. Duong Thi et al.

Fig. 4. The processes of calculating the resilience values

capital, the indicators play the same role in the resilience of the sector, then the weight has the same value in each sector. – Step 4: Determining the resilience index (Mi) and the indicator criteria. The resilience index and criteria are measures presenting the strength of climate impactions to indicators. The resilience index is arranged value ranging from 0 to 1 and divided into 4 levels as shown in Table 6. The resilience criteria are the basement to determine the value of the resilience index. The criteria are deferent in different indicators and elements, based on characteristics of indicators such as the exposure to the impaction of climate, the development in the local area, and the policy rules. For the impaction of climate, the criteria and resilience index are determined based on the exposure level to the climate change factors as shown in Table 3. It can be seen that a hazard may cause different factors or a factor that can cause some hazards. In general, the hazards in the North delta can be grouped into four types:

Building Climate Change Resilience Indicators for the Rural Commune

407

Table 4. The resilience indicators suggested for the rural area in the North of Vietnam Capitals Natural capital

Sectors Land using

Social capital

Income Poor rate Labor rate Education

Health Cultural Environment

Policy Low carbon agricultural producing

Indicators and descriptions Types of and land use in a communes: 11 types of land-use Total income in million for a person/year Rate of poor households Percentage of employed people over the population in working age Rate of universalize preschool education for 5-year-old children Percentage of students graduating from the secondary school continuing the high education Percentage of employed peoples who have been attended trainings Obtaining the national criteria for health Percentage of villages, hamlets obtain cultural standards Percentage of households using clean water and being proactive in accessing clean water sources as prescribed Percentage of households to product, business apply waste collection and wastewater treatment, and checked by Vietnam standard (QCVN 62-MT:2016/BTNMT) Having landscape planning in criteria of green environment - clean - beautiful, safe Burial area is in accordance with regulations and planning Percentage of solid waste is collected and treated Percentage of households having standard bathrooms and toilet Percentage households having breeding facilities to ensure environmental sanitation in terms of waste collection and water treatment The percentage households product fast-food obtain safety assurance Application of regulations, policies, and programs related to climate change (Budget, training) Percentage of organic fertilizer replacing urea fertilizer in agriculture Percentage of animal wastes being reproduced into fertilizer

(continued)

408

T. Duong Thi et al. Table 4. (continued)

Capitals Infrastructure capital

Sectors Traffic road condition

Irrigation system

House condition

Electricity

Community facilities used in disaster prevention

Communication system

Energy

Indicators and descriptions The commune roads and roads from commune centers to district roads are asphalted or concreted, ensuring convenient and convenient cars all time in year Percentage roads in hamlets, inter-hamlet are solidified, ensuring cars to be convenient for travel all time in year Percentage roads in hamlets are clean and not muddy in the rainy season Percentage main roads in the fields ensures convenient transportation all time in year Percentage of agricultural land area that is irrigated and drained actively Dyke system and irrigation works ensure disaster prevention No tabernacle, dilapidated Percentage of stabilized houses built in clusters Obtaining the standard of electrical system Percentage of households using electricity regularly and safely School was built stability and obtain national standards Commune has a stability culture construction Percentage of villages having cultural houses or sports activities area Preparing the equipment, supplies, vehicles and logistics according to local or local disaster prevention regulations Having postal service points Having telecommunication and internet services Percentage of villages radio and loudspeaker system Having website and applied information technology in management and administration Percentage of households using clean biogas Percentage of households using alternative energy (solar energy)

(i) seawater intrusion; (ii) flooding; (iii) instability of river/coastal infrastructure; and (iv) drought. In the natural capital, the resilience index the indicator is determined by the resilience potential when these hazards happen as shown in Table 7. The rural area here is divided into two types as the coastal commune and the non-coastal commune. Because these types of the commune have difference the climate impacts, that classification support to determine the resilience index and value more accurately. The concept for

Building Climate Change Resilience Indicators for the Rural Commune

409

that classification according to the Government’s Decree No. 161/2003 / ND-CP of December 18, 2003, clearly defining the territorial boundaries of coastal communes, districts, and provinces. Table 5. The results of the contribute weight for the indicators of infrastructure capital Indicators Traffic Irrigation House Electricity Community Communication Energy road system condition facilities system condition used in disaster prevention Weight

0.159

0.382

0.218

0.031

0.092

0.079

0.041

Table 6. The resilience index levels The Resilience Index (Mi)

The resilience levels

The climate impaction intension

1

High resilience

Not impact

0.67

Medium resilience

Negligible impact

0.33

Low resilience

Significant impact

0.1

Very low resilience

Great impact

For the social capital and infrastructure capital, the criteria for indictors were defined based on the climate impacts, the current programs for rural development as the new rural model, and other policy rules as well as the development characteristic of local area. Table 8 and Table 9 descript the criteria and basements to determine the resilience index for the indicators of the social capital and infrastructure capital, respectively. – Step 5: Calculating the resilience value: The resilience value is calculated for three levels of indicators following the equations as shown in Fig. 4. Two essential components play the roles to resilience value as the contribute weight and the resilience index, which are determined as the description in Step 3 and Step 4, respectively. The sectors in the social capital have same the weight value, but that in the natural capital and infrastructure capital have a difference in the weight value. Then the resilience values of indicators, sectors, and capital for the social capital will be carried out by the Eq. 1, 2, 4; for the natural capital and infrastructure capital will be carried out by the Eq. 1, 3, 5. Finally, the resilience value of all capitals or the resilience value of a commune will be completed by using Eq. 6. In summary, the processing of applying the resilience indicators for a commune in the North delta can be followed steps: – Selecting the indicators set (the coastal commune or non-coastal commune)

410

T. Duong Thi et al. Table 7. The resilience index for indicators of the natural capital

Sectors

Land using

Indicators

The coastal aquaculture area

The climate impact level The resilience index Coastal commune*

– Seawater intrusion: medium resistance (M = 0.67) – Floods, erosion, dike break: low resistance (M = 0.33 × 2) – Critical change in weather: medium resistance (M = 0.67) The freshwater – Seawater intrusion: aquaculture area very low resistance (M = 0.1) – Floods, erosion, dike break: low resistance (M = 0.33 × 2) – Critical change in weather: medium resistance (M = 0.67) Salt making area – Seawater intrusion: high resistance (M = 1) – Floods, erosion, dike break: very low resistance (M = 0.1 × 2) – Critical change in weather: medium resistance (M = 0.67) The rice agriculture – Seawater intrusion: area low resistance (M = 0.33) – Floods, erosion, dike break: very low resistance (M = 0.1 × 2) – Critical change in weather: low resistance (M = 0.33) The crops vegetables – Seawater intrusion: area low resistance (M = 0.33) – Floods, erosion, dike break: very low resistance (M = 0.1 × 2) – Critical change in weather: very resistance (M = 0.1) The perennial crops – Seawater intrusion: area medium resistance (M = 1) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: medium resistance (M = 0.67)

0.42

Non-coastal commune * –

0.36

0.58

0.47

–

0.27

0.44

0.22

0.38

0.59

0.67

(continued)

Building Climate Change Resilience Indicators for the Rural Commune

411

Table 7. (continued) Sectors

Indicators

The protection coastal forest area

The climate impact level The resilience index Coastal commune*

– Seawater intrusion: high resistance (M = 1) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: high resistance (M = 1) The other – Seawater intrusion: agricultural land high resistance (M = (breeding) 1) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: high resistance (M = 0.67) Non-agricultural area – Seawater intrusion: medium resistance (M = 0.67) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: medium resistance (M = 0.67) Unused land area – Seawater intrusion: medium resistance (M = 0.67) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: high resistance (M = 1) Water surface area – Seawater intrusion: (natural water high resistance (M = ecosystem, flood 1) drainage) – Floods, erosion, dike break: medium resistance (M = 0.67 × 2) – Critical change in weather: high resistance (M = 1)

0.84

Non-coastal commune * –

0.44

0.53

0.75

0.84

0.84

0.84

0.75

0.84

(*the resilience index of indicators the average of the resilience index to climate impaction; ** for non-coastal commune, the resilience index is determined in the same process, but different the climate impact).

412

T. Duong Thi et al. Table 8. The criteria and resilience index for indicators of the social capital

Sector

The criteria and basement of indicators

Sectors No. 2 Income

Total income in million for a person/year: Based on statistics of the average income of the Northern Delta: Based on the sector No. 10 of the new rural criteria indicators Based on poverty statistics of the Northern Delta Based on the sector No. 10 of the new rural criteria indicators (Multidimensional poverty rate) Percentage of employed people over the population in working age: Based on statistical data of employed population of the Northern Delta; Based on the sector No. 12 of the new rural criteria indicators Percentage of universalize preschool education for 5-year-old children: Base on No. 14.1 of the new rural criteria indicators Percentage of students graduating from the secondary school continuing the high education: Base on No. 14.2 of the new rural criteria indicators Percentage of employed peoples who have been attended trainings: Base on No. 14.2 of the new rural criteria indicators Obtaining the national criteria for health: Base on No. 15.2 of the new rural criteria indicators Percentage of villages, hamlets obtain cultural standards: Base on No. 16 of the new rural criteria indicators

Sector No. 3Percentage of poor households

Sector No. 4 The rate of employed labor

Sector No. 5 Education

Sector No. 6 – Health

Sector No. 7 Cultural

The resilience index (Mi) 1 0.67 >55 44–55

0.33 35–45

0.1 ≤35

≤1%

1–3%

3–5%

>5%

100%

100–75%

50–75%

95

90–95%

90%

80–90%

70–80%

45%

30–45%

15–30%

70

50–70%

30–50%

70%) (65–70%)

0.33 100% (50–65%)

0.1 100% (70%)

100% (60–70%)

100% (50–60%)

100% (90%

80–90%

70–80%

30% carbon fertilizer replacing urea agricultural fertilizer in agriculture production Percentage of animal >70% wastes being reproduced into fertilizer

0.33 80–90%

0.1