Numerical Modeling of Crustal Deformation and Pore-Pressure Changes Associated with the 1999 Chi-Chi Earthquake, Taiwan by Gregory Bryce Dyer A thesis submitted to the Graduate Faculty of Auburn University in partial fulfillment of the requirements for the Degree of Master of Science Auburn, Alabama December 13, 2010 Copyright 2010 by Gregory Bryce Dyer Approved by Ming-Kuo Lee, Co-chair, Professor of Geology Lorraine Wolf, Co-chair, Professor of Geology James Saunders, Professor of Geology ii Abstract A new 3D time-dependent pore-pressure diffusion model PFLOW is used to investigate the response of pore fluids to the crustal deformation generated by strong earthquakes in heterogeneous geologic media. Using a carefully calibrated finite fault- rupture model (Ma et al., 2005), the coseismic pore pressure changes and diffusion induced by volumetric strain were calculated for the 1999 Chi-Chi earthquake (Mw = 7.6) in Taiwan. The Chi-Chi earthquake provides a unique opportunity to investigate the spatial and time dependent poroelastic response of near-field rocks and sediments because extensive field data of water level changes and crustal deformation are well documented and readily available. The integrated model provides a means to explore the various mechanisms responsible for hydrologic anomalies observed in Taiwan?s western foothills and the Choshui River alluvial plain. Of special interest is identifying which of the observed hydrologic changes can be explained by a coseismic strain hypothesis and whether the pore-pressure diffusion model can account for observed recovery (dissipation) rates of seismically induced water-level changes in the alluvial fan. Coupled strain-pore pressure modeling results show a strong correlation between areas of coseismic dilatational strain and water-level decline in regions where consolidated rocks are present in the foothills. However, in the Choshui River alluvial fan, water-level rises are observed in regions of dilatational strain, suggesting that other mechanisms, such as seismic shaking, compaction, or faulting- iii enhanced gravity flow may be responsible for hydrologic changes. Assuming pre-seismic hydraulic conductivity values, our modeling results also show that water-level recovery rates cannot be explained by simple diffusion processes, suggesting that seismic loading may have caused significant re-arrangement and compaction of sediments in the alluvial plain. iv Acknowledgements The author would like to thank Dr. Ming-Kuo Lee for his support and guidance in the field of hydrogeology and geologic modeling. I would also like to thank Dr. Lorraine Wolf for guidance in the field of geophysics, seismology, and manuscript writing. This thesis would not be nearly as well written without her careful critiquing. Special thanks go out to Dr. Amnon Meir for his PFLOW codes and guidance in the field of numerical modeling and taking the time to sit down with me whenever I had a question. I would like to thank Dr. James Saunders for his role as committee member. Many thanks go out to the staff and rest of the faculty in the geology department for their guidance and support. Specifically, I would like to thank Sheila Arrington, John Simms, and Dr. Uddin. A very special thanks go out to my parents, Edward and Diane Dyer, for their love and encouragement through these last two years. I would not have made it this far without them and I owe much of my success to them. Also, I would like to thank my lovely fianc?e, Ann Robbins, for all her love, support, and encouragement, and furthermore for providing me with the all the motivation in the world to achieve the most I can. Finally, I would like to thank the USGS, GSA, and APPG for their support and funding of this project. v Table of Contents Abstract???????????????????????????? ???? ii Acknowledgements???????????????????????????. iv List of Tables????????????????????????????? viii List of Figures????????????????????????????? .ix Introduction ..................................................................................................................? ....1 Background ....................................................................................................................? ..5 Geologic History .......................................................................................................? ..5 The Coastal Range ..................................................................................................? 7 Longitudinal Valley?????????????????????????. 7 The Central Range??????????????????...???? ? ? ? .8 The Western Foothills????????????????????? ? ? ? . 8 The Coastal Range?????????????????????????. .9 Chelungpu Fault Geology??????????? ??????????..? 10 Previous Work?????????????????????????.?? ..14 Coseismic Volumetric Strain Hypothesis? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..14 Fracture Induced Permeability Enhancement ...........................................................16 Compaction ...............................................................................................................18 Studies using Numerical Modeling ...........................................................................19 Observed Hydrologic Anomalies .......................................................................................20 Water-Table Changes ................................................................................................20 vi Persistent Water-Table Variations ............................................................................22 Middle Fan Response ................................................................................................27 Objectives and Research Significance ...............................................................................49 Methodology ......................................................................................................................51 Observed Field Data .....................................................................................................51 Water-Table Data Analysis ........................................................................................52 Post-Seismic Water-Table Analysis ..........................................................................53 Stream Discharge Analysis ........................................................................................54 Numerical Modeling .....................................................................................................55 Explanation and Application of Finite Fault Models ................................................56 Fault Parameters and Input File ................................................................................57 Fault Model and Geometry .......................................................................................58 PFLOW Modeling ........................................................................................................61 Poroelastic Governing Equations ..............................................................................61 Numerical Modeling Results .............................................................................................65 Introduction ...................................................................................................................65 Case Study 1: Simplified Right-Lateral Strike-Slip Fault ............................................66 Pore-Pressure Response to Simplified Reverse Fault ................................................69 Case Study 2: Chi-Chi Earthquake? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...85 Pore-Pressure modeling of the Chi-Chi Earthquake? ? ? ? ? ? ? ? ? ? ? ? . 88 Case Study 3: Choshui River Alluvial Fan .................................................................101 Choshui Pore-Pressure Modeling ...........................................................................104 Discussion ........................................................................................................................113 vii Factors Influencing Well Response ............................................................................113 Coseismic Response....................................................................................................115 Importance of Hydraulic Conductivity .......................................................................134 Summary and Conclusions ..............................................................................................135 References ........................................................................................................................138 Appendix A: Proximal Time-Series.................................................................................144 Appendix B: Middle-Fan Time-Series.............................................................................148 Appendix C: Distal-Fan Time-Series??????????????.?????.. 158 Appendix D: Strain Input file? . .? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 168 viii List of Tables Table 1. Co- to post-seismic water-levels for wells in the Choshui River alluvial fan.? .35 Table 2. Spatial distribution of hydraulic conductivity used in the Choshui alluvial fan pore-pressure diffusion model? ???? ? ?????.??? ???? .? 105 Table 3. Input parameters for strain calculations and finite fault models?????... . 112 Table 4. Aftershock timing and magnitude.???????????..?????... 122 ix List of Figures Figure 1. Geologic map and cross-section of Taiwan???????????? ?? .. 6 Figure 2. Choshui River alluvial fan cross-section ????????????? ?... 11 Figure 3. Location map and geologic cross-section of Chelungpu fault????... ....... 13 Figure 4. Location map for monitoring wells in the Choshui River alluvial fan..? ?... 21 Figure 5. (a) Time-series plot illustrating up-up water-level response (FY) (b) time-series plot illustration down-down water-level response (LY) ? ????????. .. 23 Figure 6. (a) Time-series plot illustrating up-down water-level response (YL) (b) time- series plot illustration down-up water-level response (CS) ????. ????. 24 Figure 7. Map indicating well locations and the three divisions of the Choshui River alluvial fan? ???????????????.?????????...?. 26 Figure 8. Coseismic well response map????????????????? ??. 29 Figure 9. Coseismic time-series plots representative of wells located in the distal, middle, and proximal zones??????????...???????????? .? 30 Figure 10. 50-day well response map?..???????????????? ?? .. 32 Figure 11. 100-day well response map?..???????? ????? ?... ???. 34 Figure 12. Stream gauging station location map???????????? ?... ?... 37 Figure 13. Pre- to post-seismic response of stations SS (a), TL (b), and YL(c) along the Ta?ang River??????? ?????????????????? ..? 38 Figure 14. Pre- to post-seismic response of stations CN(a),TD(b), and WS(c) along the Wu River??? ???????????????????????? ... .40 Figure 15. Pre- to post seismic response of stations CF(a), NB(b), and KI(c) along the Wu River?????????????????????????? ..?. .41 x Figure 16. Pre- to post-seismic response of stations TT(a) and NMP(b) along the Choshui River?????? ???????????????????? 43 Figure 17. Pre- to post-seismic response of stations TC(a) and CY(b) along the Choshui River??????????????????????????? ..?? .44 Figure 18. Chelungpu fault model and geometry as used in this research??? ?... ?. 61 Figure 19. Strain map and cross-section for simplified right-lateral fault model? ??????????????????.????? ?? .?? ? 69 Figure 20. Coseismic pore-pressure field for simplified right-lateral fault? ?... ??? 71 Figure 21. Maximum coseismic pore-pressure field for simplified right-lateral fault???? ?????????????????????????? 72 Figure 22. Resultant pore-pressure field for simplified right-lateral fault after 1 day with a homogeneous hydraulic conductivity of 3 x 10-4 m/d? ??? ? ? ........74 Figure 23. Pressure field and pressure gradient map for right-lateral fault?? ?... ?? 75 Figure 24. Resultant pore-pressure field for simplified right-lateral fault after 10 days with a homogeneous hydraulic conductivity of 3 x 10-4 m/d???? ???..... 77 Figure 25. Resultant pore-pressure field for simplified right-lateral fault after 20 days with a homogeneous hydraulic conductivity of 3 x 10-4 m/d?? ?????..... 78 Figure 26. Resultant pore-pressure field for simplified right-lateral fault after 1 day with a homogeneous hydraulic conductivity of 3 x 10-5 m/d?? ????? ?? ....? 80 Figure 27. Resultant pore-pressure field for simplified right-lateral fault after 10 days with a homogeneous hydraulic conductivity of 3 x 10-5 m/d?????? ?.? 82 Figure 28. Resultant pore-pressure field for simplified right-lateral fault after 20 days with a homogeneous hydraulic conductivity of 3 x 10-5 m/d??????? ? . 83 Figure 29. Resultant pore-pressure field for simplified right-lateral fault after 100 days with a homogeneous hydraulic conductivity of 3 x 10-5 m/d?????? ??. 85 Figure 30. Calculated strain map and cross-section for Chi-Chi earthquake? ?... ?.? 88 Figure 31. Coseismic pore-pressure field calculated for Chi-Chi earthquake with a homogeneous hydraulic conductivity of 3 x 10-4 ?????????? ..?.... 91 Figure 32. Coseismic pore-pressure field calculated for Chi-Chi earthquake after 4 days with a homogeneous hydraulic conductivity of 3 x 10-4 m/d??? .? ???.... 92 xi Figure 33. Coseismic pore-pressure field calculated for Chi-Chi earthquake after 25 days with a homogeneous hydraulic conductivity of 3 x 10-4 m/d???.????.... 93 Figure 34. Coseismic pore-pressure field calculated for Chi-Chi earthquake after 50 days with a homogeneous hydraulic conductivity of 3 x 10-4 m/d??? .??? ?.... 94 Figure 35. Schematic diagram of the layered geology used for layered Chi-Chi pore- pressure model?????? ?????????????? .?????. .95 Figure 36. Resultant pore-pressure field calculated for Chi-Chi earthquake after 25 days with a layered geology??????? ???????.............................. .....97 Figure 37. Resultant pore-pressure field calculated for Chi-Chi earthquake after 50 days with a layered geology......... ..................................................................................99 Figure 38. Resultant pore-pressure field calculated for Chi-Chi earthquake after 100 days with a layered geology .........................................................................................100 Figure 39. Resultant pore-pressure field calculated for Chi-Chi earthquake after 200 days with a layered geology???????????????... ........... ...............101 Figure 40. Close-up view of strain calculated for the Choshui River alluvial fan and location map? ?????????? ???????????????. 103 Figure 41. Strain calculated for the Choshui River alluvial fan and strain cross-section ????????????????????? ???. ??????? ???. 103 Figure 42. Coseismic pore-pressure field calculated for the Choshui River alluvial fan with 8 layered geology????? ??????? ...?????????. .107 Figure 43. Resultant pore-pressure field calculated for the Choshui River alluvial fan with layered geology after 25 days??? ??????? .?????????. 109 Figure 44. Resultant pore-pressure field calculated for the Choshui River alluvial fan with layered geology after 50 days??? ???????????????..? 110 Figure 45. Resultant pore-pressure field calculated for the Choshui River alluvial fan with layered geology after 100 days??? ???????????????.... 111 Figure 46. Resultant pore-pressure field calculated for the Choshui River alluvial fan with layered geology after 200 days???? ??????????????? 112 Figure 47. Coseismic well-response versus calculated strain field for the Chi-Chi earthquake? ??????? ??????????????????? 119 xii Figure 48. Coseismic well-response versus calculated pore-pressure field for the Chi-Chi earthquake??? ??????????????? ...????????. 120 Figure 49. 100-day well response for wells located in proximal fan/slope region??????????????????????????? ? ... 124 Figure 50. 100-day well response for wells located in the middle fan?? ?... ??... ..126 Figure 51. 100-day well response for wells located in the middle fan??? ? ?... .... 127 Figure 52. 100-day well response for wells located in the distal fan..????... ........ .128 Figure 53. 100-day well response for wells located in the distal fan..???? ?? .... 129 Figure 54. 100-day precipitation data for the towns of Douliou and Changhua? ?? ..130 Figure 55. Modeled JL well recovery versus observed recovery??? ?... ???? ..134 1 Introduction The damages from earthquakes can be extensive in terms of money required to repair structures and more importantly the lives lost to these catastrophic events. For these reasons, geoscientists have dedicated much research to understanding the mechanisms that produce mainshocks and aftershocks. The dynamics of earthquake triggering mechanisms are highly variable and intrinsically related to the regional tectonic stress environment as well as to hydrologic changes (e.g., Healy et al., 1968; Simpson et al., 1988; Lin and Stein, 2004). Major breakthroughs relating stress transfer generated by earthquakes to neighboring faults and associated hydrologic changes has increased geoscientists? abilities to locate potential areas of seismic activity, yet many aspects of these predictions rely on knowledge and understanding of the seismic history for a given area (Stein, 1999). An unparalleled opportunity to study earthquake dynamics and associated hydrological responses arose on September 21, 1999. On this date a magnitude Mw= 7.6 earthquake occurred near the town of Chi-Chi in central Taiwan. The epicenter was located at 23.85?N, 120.82?E, with a focal depth of approximately 8 km. The seismic event occurred along the Chelungpu thrust fault, which trends north-south for 100 km (Ji et al., 2005). The Chelungpu fault is a bedding-parallel thrust that separates central Taiwan?s foothills and coastal plains (Lee et al., 2002). The rupture generated dual slip components, having both a reverse dip-slip and also a strike-slip motion. Horizontal 2 displacements along the hanging wall increased from 1 m in the south to 9 m at the fault?s northern terminus. Vertical displacements along the hanging wall were greatest near the fault trace, where a maximum value was found to be around 4.4 m. Along the footwall, smaller horizontal displacements were observed to be from 0.2 m to 1.5 m, and subsidence was observed to 0.32 m (Johnson et al., 2001). The effects of the Chi-Chi earthquake?s large surface rupture were devastating to the population in the region. Official casualty reports indicate that 2470 people died, 11,305 others were injured, and greater than 100,000 buildings were damaged (Shin and Teng, 2001). During this catastrophic event, a network of 60 broadband strong-motion stations recorded ground-motion data (Lee et al., 2001; Shin and Teng, 2001; Wang et al., 2003) and 70 hydrological stations with 188 monitoring wells documented considerable changes in water levels across the large sedimentary basin to the west of the fault (Chia et al., 2001; Wang et al., 2001; Wang et al., 2003). The availability of such data provides researchers with a rare opportunity to investigate possible correlations between coseismic crustal deformation and hydrological changes, such as pore pressure diffusion, water- level change, increases in stream discharge, and liquefaction (Wang et al., 2003). Despite a large number of investigations on the source mechanisms of the Chi-Chi earthquake and on rupture propagation, little attention has been given to possible poroelastic coupling in this large earthquake event. In general, the role and interactions of pore fluids with large magnitude earthquakes and aftershocks is still a deficient aspect of earthquake investigation. However, this realm of investigation should not be ignored, as the expulsion or diffusion of overpressured fluids out of the fault zone could significantly impact the pore-pressure distribution and aftershock activity in the seismic zone. 3 Similarly, the study of coseismic water-level changes and pore- pressure diffusion is also important because of its relation to other earthquake-related natural hazards, such as landslides and debris flows, liquefaction of soil, and groundwater pollution caused by the mixing of radioactive wastes or other hazardous wastes (Lai et al., 2004). Thus understanding the mechanisms that generate hydrologic anomalies and being able to quantify these relationships is important for many reasons. Three plausible hypotheses have been proposed to explain the possible mechanisms accounting for field observations that indicate a strong relationship between hydrological changes and coseismic crustal deformation and ground motion. The first hypothesis claims that water can be expelled from the crust as a result of coseismic elastic strain (Muir-Wood and King, 1993). The rupturing of faults transfers stress to other areas of the crust, and therefore generates areas of strain, which may be of a compressional or dilatational nature. Pore fluids flow from areas of compression to areas of dilatation, accounting for some of the hydrological variations associated with large magnitude earthquakes. The second hypothesis states that the properties of the fluid- containing matrix can be altered by seismic activity. Fractures opened during an earthquake could increase the permeability and conductivity of the rock as long as these fractures are well-connected (Briggs, 1991; Rojstaczer and Wolf, 1992). This increase in permeability can lead to both changes in flow direction as well as increased flow and thus possibly account for observed hydrological anomalies. A third hypothesis claims that water can be released from storage through coseismic liquefaction or sediment compaction by strong ground-motion (Manga, 2001; Manga et al., 2003; Montgomery et 4 al., 2003). The expulsion of pore fluids caused by a reduction in pore space could account for rises in water-table and increasing base flow to streams. This research project explores the three major hypotheses for explaining time- dependent changes in stress/strain and pore-pressure associated with the 1999 Chi-Chi, Taiwan, earthquake. Results from poroelastic modeling are compared with field data to examine the relationship between pore pressure build-up/relaxation and crustal deformation during and after the earthquakes. Of special interest are (1) which of the observed water-level changes can be explained by the coseismic strain model and (2) whether the pore-pressure diffusion model can adequately explain the various recovery (dissipation) patterns of seismically induced excess pore-pressure in the alluvial fan. These results have significance for understanding hydrologic anomalies related to other large magnitude earthquakes. 5 Background Geologic History The island of Taiwan developed as the result of the complex interactions between the Philippine Sea plate and the Eurasian plate over the last 12 Ma (Teng, 1990). Geographically, Taiwan is located at the convergent boundary between these two plates. The island?s formation began when the ongoing consumption of the oceanic crust of the South China Sea led to the oblique collision of the Eurasian continental margin with the Luzon Arc around 6.5 Ma (Lin et al., 2003). Seismological studies have shown that this region is one of the most complex consumptive plate boundaries in the world (Katsumata and Sykes, 1964). To the south, the Eurasian plate subducts underneath the Philippine Sea plate along an east-dipping seismic zone; however, to the northeast, the polarity of subduction reverses and the tectonic style reflects the subduction of the Philippine Sea plate along a north-dipping seismic zone (Shin and Teng, 2001). In between these subduction zones, the island of Taiwan developed as a small fold-and-thrust belt as the result of oblique collision between the northern Luzon Arc Complex and the South China Sea rifted margin. Taiwan can be broken down into five major geological provinces, the Coastal Range, the Longitudinal Valley, the Central Range, the Western Foothills, and the Coastal Plane, described below from east to west (Fig. 1). 6 Figure 1. (a) Geologic map of Taiwan shows the five major geological provinces developed from the oblique collision of the Philippine Sea plate with the Eurasian plate. (b) Geologic cross-section provides a generalized view of changes in lithology and structure along the transect A-B (after Shin and Teng, 2001). 7 The Coastal Range The Coastal Range in eastern Taiwan was formed from the collision of the northern Luzon arc and fore-arc basin with the Eurasian margin during the Miocene (Ho, 1986). The lithology is dominated by basal andesitic volcanic units and vertically grading volcaniclastic deposits. Volcaniclastic deposits are composed mainly of turbiditic sandstones, mudstones, siltstones, and conglomerates containing large amounts of disseminated volcanic fragments. Found within the upper sections of volcaniclastic units are m?lange type deposits (Ho, 1986). These m?lange deposits are composed of detritus from the Luzon arc and the Eurasian margin. Ophiolites derived from the uppermost mantle of the South China Sea are commonly found within these m?lange deposits (Page and Suppe, 1981). Generally, the structure of the Coastal Range is dominated by NNE- trending anticlines and synclines, and low-angle east-dipping imbricate thrust faults (Ho, 1986). Longitudinal Valley The Longitudinal Valley separates the eastern Coastal Range and the Central Range. This zone is approximately 6 km wide and trends N-S for 150 km. Geologically, it is considered to be the suture zone associated with the Luzon Arc-Eurasian margin collision, where the Eurasian margin corresponds to the western boundary and the Luzon Arc of the Philippine Sea Plate is the eastern boundary (Chai, 1972). 8 The Central Range The Central Range forms the core of the island and rises to a height of 4000 m. The Central Range is dominated by unroofed pre-Tertiary continental basement and Cenozoic sedimentary and metasedimentary cover. The eastern flank is dominated by pre-Tertiary metamorphic basement complexes, made up of mostly schists, marble, gneiss, and amphibolites bodies (Ho, 1986). The western flanks and highest ridges are underlain by argillite, slate, and phyllite (Ho, 1975). The Western Foothills The Western Foothills are part of the modern day foreland fold-and-thrust belt formed from pre-collisional Cenozoic siliciclastic deposits including Oligocene to Miocene continental margin sequences and Pliocene to Quaternary synorogenic foreland basin deposits (Chou, 1973; Covey, 1986; Teng, 1987a). Continental margin sequences consist mostly of shallow marine siliciclastic sediments (i.e., shale, siltstone, and sandstone) derived from the Asian continent and correspond to the original depositional setting on the continental margin (Teng, 1990). The Cenozoic sediments in the western Taiwan basin were deformed into mountains during the Plio-Pleistocene by a combination of folds and thrust faults (Ho, 1986). 9 The Coastal Plain The Coastal Plain, west of the Western Foothills, consists of alluvial sediments and well-bedded, but poorly consolidated, clastic deposits. The Neogene sediments underneath the alluvial cover have a regional dip towards the east (Ho, 1986). The lithologies of the Coastal Plain mimic that of the tectonic development of Taiwan and the foreland basin. Foreland basin deposits are underlain by passive margin siliciclastic sediments and syn-rift deposits originating from the rifting of the South China Sea. Foreland basin deposits show a generally coarsening and shallowing up sequence. Deeper marine deposits are overlain by more shallow water and coastal deposits as the orogenic front migrated forelandward during the Plio-Pleistocene. In general, facies show a trend from more coarse in the north to fine in the south, indicating the relative direction of the collision is propagating towards the south as new depocenters evolve with the migrating orogenic front. A major unconformity separates the foreland sequences from the Miocene passive margin sequences. The stratigraphic gap increases westward into the distal portions of the foreland basin and provides evidence for uplift and erosion caused by a migrating forebulge during the last 3 m.y (Yu, 2001). The 1800 km2 Choshui River alluvial fan lies within the Coastal Plain province and provides the southwest Taiwan with its main source of freshwater. It is composed mainly of alluvial and shallow-marine deposits of Holocene to Pleistocene age, overlying a sedimentary basin containing alternating layers of Pleistocene to Miocene sandstone and shale (Chen and Yuan, 1999). The Western Foothills also contain these same alternating layers; however, they have been faulted and folded (Wang et al., 2004). The Choshui River alluvial fan contains four interfingering aquifers separated by three clay- 10 rich confining aquitards (Fig. 2) (Water Resources Bureau, 1999). The aquifers extend to a depth of approximately 300 m. The aquitards consist of thin, fine-grained sand, silt and mud that pinch out towards the fan?s eastern edge (e.g., Wong and Wung, 2007). Layer F1, the shallowest aquifer, is an unconfined or partially confined aquifer, whereas Layers F2 through F4 are mostly well-confined aquifers (Hsu et al., 2000). The Choshui River alluvial fan is also divided into the upper-fan, middle-fan and lower-fan areas, based on the different hydrological and geological settings. The upper-fan (proximal) is closest to the Western Foothills and is comprised of thick gravel layers with high hydrological conductivities (10.81 x 10-4 to 8.37 x 10-4 m/d). The middle-fan is composed of thick sands and gravels with some interbedding of silts and muds to the north and mostly medium to fine-grained sands and silts to the south. The distal-fan is mostly composed of marine facies silt and mud deposits of low hydrologic conductivity (5.99 x 10-4 to 1.52 x 10-4), with a few thin layers of sands and gravels in the north-central area (Water Resources Bureau, 1999). Chelungpu Fault Geology Most of Taiwan?s seismic activity is attributed to rapid northwest convergence of the Philippine Sea plate towards the Eurasian plate (e.g., Shin and Teng, 2001). The average rate of convergence is believed to be approximately 82 mm/yr (Wong and Wung, 2007). Seismicity coincides with the growth of the Central Range and Western Foothills by frontal accretion and internal thickening as the convergence between the two plates continues. The 1999 Chi-Chi earthquake was the result of rupture along the Chelugpu 11 Figure 2. Geologic cross-section of the Choshui River Alluvial Fan (after Baird, 2002) 12 fault in west-central Taiwan (Fig. 3). The Chelungpu fault is part of the foreland basin fold-and-thrust belt that accommodates this internal shortening of the island, and also marks an important geological boundary between the Western Foothills and Coastal Plains provinces (Shih et al., 2000). It is considered to be a bedding-parallel thrust (Fig. 3) that follows and slips along the 150 to 300 m-thick Chinshui shale layer (Wang et al., 2004). The fault strikes roughly N-S for approximately 75 km, and then bends to the east for another 25 km; it dips towards the east at approximately 30 degrees (Chen et al., 2001). At greater depths the Chelungpu fault is thought to be a basal decollement in the critical wedge-type mountain belt and thin-skinned model (e.g., Barr and Dahlen, 1989; Barr et al., 1991, and Wang et al., 2000). Early studies suggest that the Chi-Chi event involved multiple fault planes with variable slip and strike components. 13 Figure 2. (a) Location of the Chelungpu fault and Chi-Chi epicenter. (b) A generalized geologic map of the fault zone area. (c) Cross-section of the Chelungpu fault zone shows that the Chelunpgu fault dips towards the east at approximately 29 degrees along the underlying Chinshui shale layer (after Johnson and Segall, 2003) 14 Previous Work A significant amount of research has been aimed at understanding the mechanisms that trigger anomalous hydrologic variations observed with the onset of seismic events. A fundamental understanding of these mechanisms proves vital as earthquakes have been observed to disturb groundwater supplies, change pore-water chemistry, and drastically alter pore-pressures. Advances in geodetic observation tools, hydrologic monitoring stations, and computer modeling software, combined with a practical importance to understand the link between crustal deformation and hydrologic anomalies, have provided the assets necessary to explore poroelastic coupling in great detail. Coseismic Volumetric Strain Hypothesis Muir-Wood and King (1993) first proposed the volumetric strain hypothesis. This hypothesis investigates how stress resulting from fault ruptures is transferred to pore fluids. When a fault slips, it relieves a portion of the regional tectonic shear stress. Because of heterogeneities and the finite length of the break, strain created by the earthquake includes substantial components of compression and dilatation, as well as shear (Nur and Booker, 1972). Therefore, for a given seismic event, stress is relieved along the fault plane, but is transferred to the near and far fields via the propagation of seismic waves. This propagation of energy provides stress to the surrounding regions, which undergo deformation or volumetric strain. Depending on the focal mechanism and kinematics of slip, these regions become zones of induced compression or dilatation. Pore 15 pressure will change by an amount proportional to the volumetric shear stress change, and pore-fluids will flow from areas of compression to those of dilatation. The focal mechanism for each earthquake provides the key constraints for the poroelastic response. Different kinematics of slip result in unique stress transfer conditions and resultant strain patterns. Generally associated with normal faulting is a dilatation of pore-spaces and cracks during the interseismic periods, followed by reduction in porosity and fracture size with the onset of an earthquake. This results in the expulsion of pore-fluids during the seismic event (Muir-Wood and King, 1993). Reverse faulting results in the opposite response. As stress builds up on the fault region during the interseismic periods, porosity and fractures become reduced. During the fault rupture, the pore spaces and fractures open up, serving to draw down the fluids towards the fault zone as a pump (Muir-Wood and King, 1993). Strike-slip and oblique slip faults have been shown to be very complex with regards to kinematics and thus may be difficult to be described by such an idealized coseismic strain model. Therefore, the key component to this hypothesis is what happens to the individual pore-spaces and existing fractures following the fault rupture. If the deformation due to rupture results in compression, then pore-fluids will be forced out of voids from the reduction in pore-space and an increase in pore pressure. If the region undergoes dilatation, the pore-space becomes expanded and thus fluid pressures drop with the onset of an earthquake. This resultant drop in fluid pressure theoretically corresponds also to a drop in water-table elevation. In summary, one should expect to see a net increase in water-table elevation with compression due to an expulsion of pore-fluids out of the pore- space and a decrease in water-table elevation with dilatation due to a decrease in pore- 16 pressure. This response is thus a viable mechanism for inducing hydrologic anomalies such as rapid changes in well monitor readings and variations in stream discharges following an earthquake. Fracture-Induced Permeability Enhancement Rojstaczer and Wolf (1994) and Wang et al. (2010) proposed an alternative mechanism to explain invoked hydrologic anomalies associated with seismic events. Both explored in detail the response of stream discharges to earthquakes. They found that stream discharges commonly increased post-earthquake, after a lag time of minutes up to a few weeks following the rupture. They argued that these pronounced increases in stream discharge are related to fracturing of rocks controlling the groundwater flow system. Following the 1989 Loma Prieta earthquake, data from stream-gauging stations indicated significant increases in stream discharge within 15 minutes of the rupture. Similarly, Rojstaczer and Wolf (1994) found ionic concentrations and the calcite saturation index of the stream water had also increased. Stream flow and solute concentration decreased significantly over a period of several months following the earthquake. An increase in solute concentrations and calcite saturation index in the surface waters provides compelling evidence that excess water must have interacted with the rock units along a flow path that originated from the nearby mountainous regions (Rojstaczer and Wolf, 1994). Furthermore, it was observed that in the highland portions of the basin, groundwater levels were lowered by as much as 21 m within weeks to months after the earthquake. From these data, they surmised that fracturing generated by 17 the earthquake had increased the rock permeability and enhanced groundwater flow and dewatering from the higher portion of the mountains and basins (i.e., recharge areas) towards lower elevations discharge areas through increased base-level contribution. Similar increases in stream discharge were observed following the 1999 Chi-Chi earthquake, Taiwan. Discharge values obtained from stream gauging stations near the Western Foothills displayed rapid coseismic increases. Base flow analysis of these streams, however, revealed that no significant changes in horizontal hydraulic diffusitivity resulted from the earthquake (Manga, 2001; Manga et al., 2003). Similarly, no significant precipitation events occurred during this time frame to cause the increased discharges. It was hypothesized by Wang et al. (2010) that vertical fracturing caused by the seismic event could be a mechanism for increase in stream discharge. Well-lithified shale and sandstone units of the Western Foothills provide the possibility of perched water-tables. Sandstone reservoirs could be confined by underlying relatively impermeable shale units. However, if these shale units become fractured by the tensile stresses generated from the drop of the hanging wall (Lee et al., 2002) or by strong seismic shaking, then new flow paths from higher elevations in the foothills towards lower elevations may develop. Evidence of decreased water-levels in perched water- tables could provide more scientific evidence that such processes may be occurring. Therefore, Wang et al. (2010) concluded that increased fracturing in the Western Foothills region of Taiwan created a connection between perched water-tables and streams at lower elevations. Fracturing-induced permeability by tensile stresses or extreme seismic shaking can be a viable hypothesis if the resultant fractures are well connected to permeable 18 geologic layers. If the fractures remain poorly connected, new pathways for fluid flow are not possible, and increases in base flow and stream discharge would not likely result. Seismic shaking can also lead to the removal of particles or gas bubbles that may have otherwise blocked fluid flow through these fractures, providing yet another way that fracturing can increase permeability (Brodsky et al., 2002). Compaction Manga (2001) studied stream flow response to the Chi-Chi earthquake in the Choshui River alluvial fan region of Taiwan as well as the response of several California streams to seismic events. He argued that a third plausible mechanism is responsible for increased water-tables and stream discharge rates. Manga (2001) postulated that for the studied stream systems, the coseismic increase of discharge required increased hydraulic head gradients resulting from the rapid release of water from some storage source in the shallow crust. He hypothesized that a change of fluid pressure in the matrix materials by transient dynamic strain created elevated stream discharges. If the seismic waves propagating through the layers contain enough energy to surpass a threshold resistance then the sediments generally will preferentially re-orient themselves into a new configuration. This agitation brought about by seismic shaking or transient dynamic strain generally results in a tighter packed sediment configuration reducing the pore- space and increasing grain-to-grain contacts. As a direct result, pore-waters would be expulsed, providing a source for increases in water-table elevations or stream discharge values. Following the Loma Prieta earthquake, water temperatures of streams were found to be lower than normal with excess flow from a shallow source (Manga, 2001). This is 19 somewhat contrary to the coseismic volumetric strain hypothesis in that Muir-Wood and King (1993) suggested that the source water most likely came from the mid-crust. Compaction is most commonly seen in alluvial fans or regions dominated by relatively unconsolidated sediments near the fault zone. Numerical Modeling Baird (2002) researched the interplay of coseismic strain and pore pressure for the 1999Chi-Chi earthquake. He created a one-way coseismic deformation fluid-pressure diffusion model. His results showed a promising correlation between induced strain fields andthe behavior of pore fluids, supporting the mechanism for coseismic volumetric strain (Muir-Wood and King, 1993) as the trigger for some hydrologic anomalies observed in Taiwan. Baird?s research shows that some coseismic drops in water-table were found to be located in dilatational zones. Pore-pressure modeling revealed that these zones showed cosesimic decreases in pore pressure and thus water-table drops would be consistent with a coseismic volumetric strain hypothesis. However, Baird?s (2002) finite fault rupture model was not calibrated against field data . Contrastingly, this study improved Baird?s model by making use of a published finite-fault rupture model (Ji et al., 2005) and thus pore-pressure models will be integrated with well calibrated spatially-dependent stress and strain values for the Chi-Chi earthquake. 20 Observed Hydrologic Anomalies Water-Table Changes A network of 58 monitoring wells documented the coseismic and post-seismic responses of the Choshui River alluvial fan aquifers to the 1999 Chi-Chi earthquake (Fig. 4). The observed hydrologic anomalies following the Chi-Chi earthquake varied widely. Coseismic variations in water-table elevations were found to range from a decrease of - 11.1 m to an increase of 7.42 m within the Choshui River alluvial fan and slope region. The magnitude of change corresponded relatively well to the proximity of well location to the Chelungpu fault. Generally, the largest variations occurred near the rupture zone, with diminishing values farther from the fault. Two types of hydrologic responses were observed in the well logs: (1) persistent changes and (2) oscillatory changes. Persistent changes were commonly found within the Choshui River alluvial fan and slope regions. These persistent changes are characterized by step-wise coseismic changes in the water levels and gradual diffusion towards a pre-seismic water-table elevation. Oscillatory changes occurred mainly in the shallow, unconfined aquifer and are the result of passing seismic waves through the water column. These changes lasted only while the shear waves originating from the Chi-Chi earthquake propagated through the well locations. 21 Figure 3. Map showing the locations of monitoring wells within the Choshui River alluvial fan and Western Foothills upper-slope region. Red dots indicate well locations. Black line represents the Chelunpgu fault. 22 Persistent Water-Table Variations Persistent water-table variations showed two general trends with regards to proximity to the seismogenic zone. Coseismic water-table falls were observed in regions nearest the fault zone, mainly on the edge of the Western Foothills and slope regions of the Choshui River alluvial fan. These drops were generally located in consolidated sedimentary rocks and located within 5 km distance from the Chelungpu fault. Farther away from the fault zone (<15 km), within the middle and lower portions of the Choshui River alluvial fan, strong coseismic water-table rises were observed. In between these zones, a mixture of rises and falls were observed within the monitoring well network. Of the 58 total monitoring wells recorded readings, the predominant response was a coseismic rise (50 wells) following the earthquake. Piezometric readings can be grouped into four distinct diffusion patterns within the wells: (1) a up-up response (Fig. 5a), (2) down-down response (Fig. 5b), (3) up-down response (Fig. 6a), and (4) down-up response (Fig. 6b) (Wang et al., 2004). These responses were characterized by Wang et al. (2004) to be related to the local geology of the well and also the redistribution of stress fields in the shallow subsurface due to fault rupture (Chia et al., 2008). This research focuses on two responses: (1) rise-fall and (2) fall-rise as these are the only responses that can be investigated by a coseismic volumetric strain mechanism. Rise-rise and fall-fall responses could be attributed to outside factors such as local irrigation or pumping or possibly attributed to be the response due to another mechanism altogether (e.g., fracturing). 23 Figure 5. (a) Well FY shows a up-up response following the Chi-Chi earthquake. (b) Well LY shows a down-down response following the earthquake. Black arrows indicate timing of the earthquake event. 24 Figure 6. (a) Well YL shows an up-down response to the Chi-Chi earthquake. (b) Well CS shows a down-up response to the earthquake event. Black arrow indicates timing of the earthquake. 25 Coseismic Response of Aquifer F2 Aquifer F2 is mostly confined by clay and silt layers above and below. It is composed mostly of medium-to-coarse grained sands in the central and western regions of the fan and gravel sized sediments (Fig. 2) to more consolidated sedimentary rock in the upper fan and slope regions (Lai et al., 2004). Careful investigation of all the major aquifers in the region revealed that aquifer F2 showed the largest magnitude changes and thus is the most useful aquifer to study for the purposes of this research. The fan is divided into three segments based upon distance from the fault zone (Fig. 7). The proximal zone extends roughly 15 km westward from the Chelungpu fault zone. This zone includes the slope region of the Western Foothills and upper portion of the Choshui River alluvial fan. Aquifer F2 in this region is generally regarded to be composed of gravel-to-sand sized sediments as well as consolidated sandstone (Fig. 2). The second zone, the middle fan, ranges from 15 km to 30 km west of the Chelungpu fault. Aquifer F2 in this zone reflects a gradation from more coarse sands and gravels in the east to medium grained sands in the west (Fig. 2). The final zone, the distal fan, extends from 30 km west of the fault to the coast. The sediments constituting the F2 aquifer in the distal zone grade from medium to fine-grained sands before pinching out beneath the Taiwan Strait (Fig. 2). 26 Figure 7. Well locations are divided into three categories based on proximity to fault. Proximal < 15 km. Middle Zone 15 to 30 km. Distal Zone > 30 km. Chelungpu fault is highlighted by solid dashed line (after Baird, 2002). 27 Distal Response In total 25 wells were located in the distal zone (Fig. 8). The dominant coseismic response (21 wells) was observed to be an increase in water-table elevation . The largest magnitude changes occurred in the northern portion of the distal zone (north of Choshui River), averaging an approximate increase of 3 m. The largest documented changes occurred in the far northern section of the zone in wells CC, HO, LT, and HS (Fig. 9a), where changes were 5.4 m, 5.28 m, 5.22 m, and 4.28 m, respectively. South of the Choshui River, the magnitude of coseismic changes diminished. A majority of the wells south of the river exhibited small coseismic increases, averaging less than 1 m. Three wells documented minute coseismic decreases, however, these decreases in water-level were found to average less than 0.1 m. Middle Fan Response A total of 24 wells are located within the middle portion of the Choshui River alluvial fan (Fig. 8). Water-table change maps illustrating the coseismic water-table elevation response at each well location were made with GMT?. Of these 24 wells, only two (TZ and KS) displayed coseismic decreases in water-table elevation. Wells TZ and KS are located within 5 km of the gproximal zone and responded with coseismic drops of -2.47 m and -0.07 m, respectively. The remaining 22 wells exhibited increases in water- table elevation, ranging from 0.02 to 7.42 m. The largest coseismic increases were observed within the central portion of the middle fan around 17 to 25 km away from the 28 fault zone. Wells HE, YL, and JL (Fig. 9b) showed the largest increases with values of 7.42 m, 6.46 m, and 4.21 m, respectively. Anomalously, many of the wells located in this zone showed no coseismic response, often located near wells exhibiting large magnitude changes. Proximal Response Eight wells are located within the proximal zone (Fig. 8). Five of these wells showed strong drops in water-table elevation following the Chelungpu fault rupture. Two wells (ES, KC) showed small coseismic increases of 0.08 m and 0.29 m respectively. One well (WT) showed no change following the Chi-Chi earthquake. The largest decreases in water-table elevation were found to occur within 5 km of the Chelungpu fault, with the largest observed decrease of -11.09 m (CS) (Fig. 9c) occurring within 2 km. The other decreases in water-table elevation were found to range from -0.8 m to -5.94 m within the proximal zone. 29 Figure 8. Coseismic well response of aquifer F2 to the Chi-Chi earthquake. Black dashed lines mark zone boundary. Solid black line represents Chelungpu fault. Red square is the epicenter. Black dots indicate coseismic decreases in water-level. Key wells discussed in text are labeled. 30 Figure 9. (a) Coseismic response of well HS, located in the distal zone of the Choshui River alluvial fan. (b) Coseismic response of well JL, located in the middle zone of the alluvial fan. (c) Coseismic response of well CS, located in the proximal zone. 31 Post-Seismic Responses After 50 days, the water-levels in the F2 aquifer show a different pattern of changes (Fig. 10). Wells found in the distal zone, primarily showed a decrease in water- level, below that of pre-seismic levels. In total, 16 of the 25 wells located in the distal fan illustrated this pattern. Geographically, the majority of the decreases were found in the southern portion of the distal zone, where coseismic increases were found to be on average only around 1 m. Seasonal variations (i.e., rainfall decrease, irrigation pumping, etc) may therefore be the cause for this response. In the middle zone, a total of 6 wells were found to have dropped below pre- seismic levels. Anomalously, all of these wells showed strong coseismic increases in water-level of 1 to 3 m. Geographically, these wells are generally located within 10 km of the proximal zone and near the southeastern corner of the zone. The two wells that exhibited coseismic decreases in water-level were found to have rebounded over the 50 day time-span, responding with increased levels above that of pre-seismic level. The remaining wells (16), illustrated average decreases of 1 to 2 m from strong coseismic increases,indicating a slow recovery back to pre-seismic levels. Seven of the nine wells located in the proximal zones were found to have water- levels below that of pre-seismic water-table. Two wells that exhibited coseismic increases were found to have dropped below pre-seismic levels. One well, that showed a coseismic decrease in water-level, was found to have increases above pre-seismic level after 50 days. 32 Figure 10. Post-seismic (50-day) water-level changes show that a majority of the wells located in the distal zone show drops below pre-seismic levels. Within the southeastern section of the middle zone, a group of wells that showed strong coseismic increases, responded with decreases over the next 50 days (red outline). Wells located in the proximal zone continue to show decreased water-levels. Black dots indicate negative levels. 33 The water-levels after 100 days (Fig. 11) show yet a different pattern than that of the coseismic and 50-day pattern. The number of wells displaying negative responses (below pre-seismic levels) decreased from 16 wells to 9 wells within the distal zone. The remaining wells (16) exhibited water-levels above that of pre-seismic conditions, but averaging only an increased head of 0 to 1 m. Wells located within the middle zone illustrated a similar pattern to that of the 50- day pattern. In total 7 wells were found to be negative. The remaining wells (16) had positive levels, but averaged an increased hydraulic head of 0 to 1 m. Only two wells had hydraulic head increases greater than 2 m. Wells located within the proximal zone continued to show decreased water-levels. In total 7 wells were found to have negative levels. The remaining two wells had positive levels; however, neither had levels that were 1 m above that of pre-seismic level. In general, the wells located in the Choshui River alluvial fan showed sporadic 100-day responses to the Chi-Chi earthquake (Table 1). It appears that wells located in the proximal zone show a more permanent trend of decreased water-level. Yet, many of the wells located farther out in the fan show an oscillating pattern, in which water-levels swing above and below those of pre-seismic conditions. 34 Figure 11. Post-seismic (100-day) response of wells located within the Choshui River alluvial fan shows that wells generally recover with a sporadic and oscillating pattern. Wells located in the proximal zone show a more permanent pattern of decreases water- level over the 100-day span. Black dots indicate negative water-levels. 35 Well Abbrev Coseismic (m) 50-Day (m) 100-Day (m) Well Abbrev Coseismic (m) 50-Day (m) 100-Day (m) CZ 4.75 0.16 -0.46 LH 0 5.04 0.37 CT 3.86 0.33 0.29 HH 2.57 -0.38 -0.91 TC 4.01 -0.17 -0.4 IW 0.21 0.38 0.42 TZ -2.47 -1.29 -3.21 KH 0.02 -0.25 1.92 FY 0.49 1.5 2.01 TY 0.72 -3.18 -1.3 CC 5.4 -0.09 0.2 HL 0.03 0.08 -0.1 YL 6.46 2.12 1.58 KH 0.02 0.8 1.28 CU 4.76 0.93 0.78 HA 0.48 -1.09 -1.5 HB 3.23 -1.2 -1.24 HG 0.74 -1.53 -2.17 HT 1 -2.29 -2.37 KC 0.02 -2.12 -1.17 HO 5.28 1.65 1.89 PK 0.97 -0.1 0.9 KS -0.07 0.7 -0.12 BT 0.56 -2 -1.86 CS -11.09 -3.63 -6.77 FG -0.09 -1.68 2.11 LY -5.94 -8.45 -8.6 WT 0 -1.19 -2.56 CK -2.23 -1.41 -1.84 SH 0 -1.55 0.69 CH 4.66 0.66 0.37 WR 0.32 -2.19 -0.55 LT 5.22 1.97 2.38 JL 4.11 0.46 0.39 TF 3.73 1.62 1.45 TG 0.5 0.02 1.09 WC 2.76 0.59 0.64 HU 1.13 -2.78 -2.23 ST 1.59 0.74 0.19 LZ -0.02 0.19 1.93 TW 3.95 2.23 1.33 TT 0.52 -1.26 0.56 HN 4.43 1.81 1.11 SN -0.01 -0.38 1.47 SO 0.24 -1.88 -3.99 YC 0.02 0.62 0.55 FT 5.8 0.14 1.07 MT -0.01 -1.33 0.52 AN 1.44 -1.57 -1.7 HR 4.55 1.31 1.17 HY 0.4 5.4 -1 TR 0.03 -0.83 2.97 HE 0.87 1.07 0.89 TK 0.99 -1.53 0.02 TH -2.96 2 0.01 GC 0.01 0.48 3.56 AH 0.09 -1.16 -0.18 TS 0.01 -0.92 0.6 Table 1. Co- to Post-seismic water-levels for monitoring wells located in the Choshui River alluvial fan. Levels are taken at three steps: (1) Coseismic, (2) 50-day, and (3) 100- day. Water-levels are plotted in Figures 8, 10, and 11. 36 Stream Response Stream discharge data gathered from the Ta?ang, Wu, and Choshui rivers (Fig. 12) show dynamic coseismic and post-seismic changes in stream flow within west-central Taiwan. Generally, the trend of these streams was a coseismic increase in discharge followed by a slow post-seismic recovery back to normal gauging levels. Precipitation is likewise measured at these stations, providing the data necessary to make distinctions between changes caused by rainfall and changes brought upon by earthquake. The Ta?ang River borders the northern portion of the Chelungpu fault zone, flowing from central Taiwan to the strait of Taiwan (Fig. 12). Along the Ta?ang River, three gauging stations monitored stream flow (SS, TL, YL). Gauging station SS (Fig. 13a), located near the border of the Western Foothills and east of the Chelunpgu fault, recorded a coseismic increase of 0.5 m. This coseismic increase took approximately 2 months to recover back to a pre-seismic level. Station TL (Fig. 13b) is located north of the fault tip, within the Western Foothills. Data retrieved from this station show a larger coseismic increase in stream level (~0.7 m). This increase took approximately 8 days to achieve maximum levels (Fig. 13b). Following its peak, the stream recovered nearly to pre-seismic level within 2 months. Station YL (Fig. 13c) is located northwest (~ 10 km) of the fault. This station shows a coseismic increase of approximately 0.5 m, followed by an oscillating recovery back to pre-seismic levels. Within a couple days of fault rupture, stations TL and YL, received around 50 mm of precipitation. Based upon plotted pre- seismic stream responses to rainfall, however, the rainfall does not appear to significantly affect the stage levels. 37 Figure 12. Stream gauging station location map for streams located in the fault zone. Chelungpu fault is highlighted with dashed black line. 38 Figure 13. Pre- to post-seismic surface (river) water level changes observed at monitoring stations SS (a), TL (b), and YL (c) along the Ta-ang River (as shown on Fig. 12). The arrow marks the Chi-Chi earthquake. Rainfall amounts are shown by the vertical bars (left Y-axis) (modified after Water Resource Bureau of Taiwan, 2000). (a) (b) (c) 39 The Wu River transverses the northern part of the Chelungpu fault (Fig. 12). Three gauging stations located to the east of the fault recorded data as well as three stations located to the west. One of these stations (WS) is located very near the fault zone. All of these stations recorded coseismic increases (1 or 2 days post-event). Generally, stations located farther than 10 km away from the fault zone displayed coseismic increases of less than 1 m and recovered back to pre-seismic levels within two months (Figs. 14 and 15). Station WS (Fig. 14c), is the exception to these observations. This station recorded an abrupt (Fig. 14) coseismic increase in stream level of over 4.5 m. In the two months following the earthquake, the station did not recover back to pre- seismic levels, but showed a continuous increased stream level. Precipitation data show that rainfall did not play a significant role in any coseismic or post-seismic changes in stream flow. 40 Figure 14. Pre- to post-seismic surface (river) water level changes observed at monitoring stations CN (a), TD (b), and WS (c) along the Wu River (as shown on Fig. 11). The arrow marks the Chi-Chi earthquake (modified after Water Resource Bureau of Taiwan, 2000). (a) (b) (c) 41 Figure 15. Pre- to post-seismic surface (river) water level changes observed at monitoring stations CF (a), NB (b), and KI (c) along the Wu River (as shown on Fig. 11). The arrow marks the Chi-Chi earthquake (modified after Water Resource Bureau of Taiwan, 2000). (a) (b) (c) . 42 The Choshui River transverses the southern portion of the Chelungpu fault (Fig. 12). Four gauging stations monitor stream levels in the area (Figs. 16 and 17). Two are located to the west of the fault (TC and CCY), one is located near the fault zone TT (Fig. 16a), and one to the east, NMP (Fig. 16b). Two stations show coseismic increases of less than 1 m, TC (Fig. 16a) and NMP (Fig. 16b), one shows no coseismic change, CCY (Fig. 17b), and one shows a coseismic decrease, TT (Fig. 17a). Stations that display a coseismic increase generally recover back to pre-seismic levels within a month. Station TT (Fig 17a), located along the fault zone illustrated a coseismic decrease of approximately 1 m. Similar to station WS (Fig. 14c) along the Wu River, the gauge showed no recovery and the stream remained at 1 m lower for at least two months. 43 Figure 16. Pre- to post-seismic surface (river) water level changes observed at monitoring stations TT (a) and NMP (b) along the Choshui River (as shown on Fig. 11). The arrow marks the Chi-Chi earthquake (modified after Water Resource Bureau of Taiwan, 2000). (a) (b) 44 Figure 17. Pre- to post-seismic surface (river) water level changes observed at monitoring stations TC (a) and CY (b) along the Choshui River (as shown on Fig. 11). The arrow marks the Chi-Chi earthquake (modified after Water Resource Bureau of Taiwan, 2000). (a) (b) 45 Theory of Poroelasticty Poroelasticity is a continuum theory for the analysis of a porous media consisting of an elastic matrix containing interconnected fluid-saturated pores (Wang and Anderson, 1982). The theory states that when a porous medium is subjected to an applied stress, the resulting deformation causes volumetric changes in the pores. Because the pores are fluid-filled, the presence of the fluid not only acts to make the material more rigid, but also leads to the flow of pore-fluids between regions of elevated and decreased pressure. Fluid flow is found to occur in order to dissipate the excess pore-pressure. Biot (1941) founded the theory of poroelasticity on two linear constitutive equations: (1) changes in applied stress and pore-pressure produce a fractional volume change and (2) changes in applied stress and pore-pressure require fluid to be added or removed from storage. Solid-to-fluid coupling occurs when a change in applied stress produces a change in volumetric strain and fluid pressure. In this research, crustal strain generated by Coulomb stress changes following large magnitude earthquakes forms the basis for inducing pore-pressure change. Pore-pressure is defined as the pressure of the fluid occupying the pore space. Generally, the pore-pressure initial condition for poroelastic problems is often based on an equilibrium applied boundary load (e.g., hydrostatic condition), which induces the starting distribution of pore-pressure. Due to the complex nature of fault kinematics and rupture, these applied stresses are non-uniform in distribution and thus lead to intrinsically complex spatial variations in pore-pressure distribution. Scientifically important to this concept is that a non-uniform pore-pressure distribution will lead to fluid flow through the geologic media (Wang and Anderson, 1982). 46 The magnitude of the solid-to-fluid coupling depends largely on the compressibility of the geologic framework, the fluid compressibility, and the porosity of the media. Compressibility (? in 1/Pa), in this sense, is the resistance of volume change in a geologic media to an applied stress while pore-pressure is held constant. When considering a poroelastic problem, the compressibility of the pore, the compressibility of the grains, and the compressibility of the fluid occupying the pore-spaces should be defined (Wang and Anderson, 1982). A highly compressible skeletal framework will lead to larger reduction in pore volume and thus impart more stress upon the fluids occupying the pores. More compressible fluids will result in more significant volumetric changes of the pore-fluids. Thus, the volumetric response of the framework and pore-fluids could instigate a migratory diffusion of fluids in or out of the pores. Specific storage, Ss ( L-1), also known as the elastic storage coefficient, is another important parameter in determining the pore-fluid response to applied stress. Specific storage is defined as the amount of water per unit volume of a saturated formation that is stored or expelled from storage owing to the compressibility of the mineral skeleton and the pore-water per unit change in head. This coefficient links dynamic changes of fluid content in an aquifer to compressibility and pressure change. With increased pressure or hydraulic head, water exerts more pressure on the pore walls and framework grains, causing a volumetric expansion of the pores. Decreases in pressure or hydraulic head will have the opposite effect, leading to a contraction of pore-spaces and an expansion of pore-fluids. The specific storage coefficient will determine how much fluid will be stored or expulsed from each aquifer in relation to the change in pore-pressure for a given 47 compressibility. Higher specific storage coefficients would indicate that the aquifer is more likely to undergo a drastic change in fluid volume. Skempton?s coefficient (B) relates the ratio of induced pore-pressure change to change in applied stress. Specifically, it is a measure of how the applied stress is distributed between the skeletal framework and the fluids. A value for pore-pressure can be obtained by multiplying the Skempton?s coefficient by the applied stress. Skempton?s coefficient varies between zero and one, with more compressible materials (e.g., clay) being closer to one and uncompressible materials estimated to be zero. Therefore, scientists generally regard a Skempton?s coefficient of one to correspond to a completely fluid supported load and Skempton?s coefficient of zero to correspond to a gas supported load. The hydraulic conductivity K (m/d) of the sediments plays a major role in determining the rate of pore-pressure diffusion and fluid flow. Hydraulic conductivity is directly proportional to the density of the fluid and inversely proportional the viscosity of the fluid. The conductivity of the material also depends upon the intrinsic permeability k (m2 or darcy) of the geologic media. Permeability is the measure of the geometry and size of the pore structure. The more well connected the pores, the greater the hydraulic conductivity. It can be assumed that higher values of hydraulic conductivity would correspond to faster diffusion rates, because the fluid flow can proceed with less restriction. Thus, groundwater should recover back to steady-state with higher values of conductivity. Generally speaking, the linear poroelastic response to coseismic volumetric deformation can be constrained to three material constants: (1) bulk modulus, (2) 48 poroelastic expansion coefficient, and (3) specific storage. The bulk modulus (Pa) for each geologic material is the resistance to uniform compression. It is defined as the pressure increase needed to cause a given relative decrease in volume. The poroelastic expansion coefficient (1/L) is opposite to the bulk modulus in that it is the materials susceptibility to undergoing expansion. Specific storage dictates the relationship between pore-fluid volume to induced pressure. The first two constants govern the amount of volumetric deformation a material will experience due to an applied stress. The last constant implies how much fluid will be stored or expulsed due to pressure changes. Together, these material constants can be used to derive other parameters such as compressibility. 49 Objectives and Research Significance The extensive hydrologic and ground-motion data gathered from the 1999 Chi- Chi earthquake presents an opportunity to examine the plausible mechanisms proposed to explain observed earthquake-induced hydrologic anomalies. The objectives of this research are: (1) to test the newly developed 3-D pore-pressure diffusion code (PFLOW), (2) to examine the mechanisms using Coulomb 3.1? and PFLOW in one-way coupling for exploring the observed anomalies created by the Chi-Chi earthquake, and (3) using PFLOW to examine the importance of hydraulic conductivity (K) in dissipating induced excess pore-pressures resulting from the earthquake. The results of this study could shed light and provide a better understanding of the dynamic relationship between volumetric strain and induced pore pressure diffusion in seismic zones and furthermore could help researchers understand: (1) predicting the effects induced fluid flow can have on solute and hazardous pollutant transport, (2) determining good locations for the storage of hazardous material, (3) the ability of pore pressure diffusion to trigger earthquakes or inhibit them, and (4) discovering the role of pore fluids in aftershock temporal and spatial distribution. In addition to providing valuable insight as to how coseismic deformation and resulting changes in pore pressure can account for hydrological variations observed 50 during strong earthquakes, this research will result in a new tool for exploring pore- pressure development and dissipation as a function of time. 51 Methodology Observed Field Data For this study, hydrologic data from 1999 to 2004 were acquired from the Water Resources Bureau of Taiwan. A dense network of 73 hydrologic stations, along with 188 individual monitoring wells, was installed within the Choshui River alluvial fan beginning in 1992 (Chia et al., 2001). Each monitoring station is equipped with one to five wells, ranging in depth from 14 to 300 meters. The monitoring stations provide good spatial coverage for investigating the hydrologic response of the alluvial fan to the crustal deformation generated from the Chi-Chi earthquake; the stations range from within 2 km of the fault zone to over 50 km from the fault zone near the coastal areas. These wells provide a complete hourly record of the coseismic and post-seismic responses of the major aquifers located in the Choshui river alluvial fan. Discharge data (stream level) gathered from gauging stations (Water Resources Bureau of Taiwan, 2002) along the Ta? ang, Wu, and Choshui Rivers provide added insight into coseismic and post-seismic hydrologic responses. 52 Water-Table Data Analysis This research project focuses solely on the F2 aquifer response because it showed the most pronounced response to the earthquake among four major aquifers in the alluvial fan. Due to its confined condition, the F2 aquifer maintained changes longer, providing the best opportunity to study the diffusion of excess pore pressures. The analysis followed the previously described divisions of the alluvial fan: (1) proximal, (2) middle, and (3) distal (Fig. 4). The proximal zone extends 10 km westward from the Chelungpu fault and encompasses the slope and upper fan regions. The middle fan extends westward from the proximal zone for 20 km and is comprised of upper fan and middle fan lithologies as described by Lai et al. (2004). The distal fan extends approximately another 20 km westward from the middle fan until it reaches the Taiwan Strait. Water-table data for this study were organized based upon distance from the Chelungpu fault in the alluvial fan (i.e. proximal, etc). Data were plotted as graphs to show the temporal evolution of the water column at each location. Coseismic piezometric well readings were used to test pore-pressure diffusion from coseismic volumetric strain as a mechanism for generating the observed hydrologic anomalies. In the coseismic volumetric strain hypothesis, the calculated resultant strain generated from the Chelungpu fault rupture should dictate the style of aquifer response immediately following the event. Specifically, if the wells are in areas of compressional strain, then the pore-fluids should be expulsed due to a reduction in net pore volume. This would lead to elevated piezometric readings following the earthquake. If the wells are in 53 areas of dilatation, then pore-fluids would be drawn in due to an increase in pore volume and correspond to lower piezometric readings. Post-Seismic Water-Table Analysis A 100-day time-series was prepared for each obtained monitoring well (Appendices A to C). These time-series plots were prepared in Microsoft? Excel? and grouped based upon proximity to the fault zone. Well data were then analyzed to determine whether or not proximity to the fault zone or local geology plays a role in fluid diffusion processes. Lastly, the observed water level recovery pattern for well JL was compared with the modeled diffusion pattern (assuming pre-earthquake hydraulic conductivity values) for well JL. The program, 3PFLOW (Lee and Wolf, 1998) was used to model the recovery of well JL with pre-seismic hydraulic conductivity values (K). This analysis tested whether the permeability of the sediments was permanently or temporarily altered by the earthquake. Themis-match between calculated and observed diffusion rates would suggest the possibility that the hydraulic properties of the sediments were changed permanently due to the propagation of seismic waves following the earthquake. The post-seismic response or diffusion patterns of the piezometric readings also provide a test of the pore-pressure diffusion model. Regardless of the well response, whether the water table rises or falls, there should be a return of the well reading back to a steady-state condition, if no permanent deformation has taken place. If coseismic volumetric strain is the only mechanism responsible for the observed hydrologic anomalies pore fluids should diffuse back to a pre-earthquake level over time. Thus, if the piezometric reading showed an increase, the water-table should show a relatively smooth 54 recovery back to a base-line, re-established the height of the water-table before fault rupture occurred. Conversely, if the piezometric reading showed a decrease in water- table, then there should be a smooth and continuous rise back to steady-state elevation. Stream Discharge Analysis Stream discharge data from the Choshui River alluvial fan, slope, and Western Foothills were used to test the volumetric strain hypothesis. Total stream discharge in most areas is the composite of contributions by surface flow and base-flow. Surface flow is the contribution made by surface runoff due to precipitation events or snowmelt at higher elevations through a drainage basin, whereas base flow is dominated by the contribution of the groundwater discharge into the streams. The mountainous regions in the foothills of Taiwan, where most streams have their headwaters, consist of mainly consolidated sedimentary rocks. As Muir-Wood and King (1993) proposed, the strain generated by fault rupture can initiate pore volume reduction or expansion dependent upon the strain regime. For consolidated regions, this strain could either open up fractures or close them. In thrust faulting, such as in the Chi-Chi earthquake, the pore spaces and fractures close up during interseismic periods due to buildup of stresses. During rupture these fractures can open up providing conduits for groundwater to flow. Therefore, if an induced hydraulic gradient is created by the strain regime, or permeability along flow path is enhanced by seismic fracturing, groundwater can flow from mountains towards lower elevations. This would result in a significant increase in base-level contribution and a resultant increase in stream discharge. 55 Stream gauge readings from three rivers in the alluvial fan, slope, and Western Foothills regions were used. The nature of the coseismic responses and post-seismic recoveries are used in conjunction with water-table data to shed light on whether or not coseismic strain is responsible for invoking the hydrologic change. Numerical Modeling Numerical modeling was used to investigate the poroelastic response of the Choshui River alluvial fan to crustal deformation generated by the Chi-Chi earthquake. This study makes use of a published finite fault rupture model (Ji et al., 2005) for 3D strain modeling, thus expanding on the model developed by Baird (2002). This finite fault rupture model has been carefully calibrated to match field observations and slip kinematics. Coulomb 3.1?, a Coulomb stress change/deformation modeling program (Toda et al., 2005), was used to perform the 3D strain modeling. The program writes an output file of resultant coseismic strain values and also provides graphic output for the calculated strain fields. The strain values calculated from the deformation code are used as an initial disturbance in the poroelastic medium. PFLOW, a 3D, finite-element pore- pressure diffusion model, was then used to calculate the resultant coseismic pore-pressure response and diffusion pattern over a specified time frame. Coulomb 3.1? Strain Modeling Coulomb 3.1? is a Matlab-based deformation modeling program that makes use of finite fault rupture models to calculate Coulomb stress changes or strain resulting from fault dislocation (Lin and Stein, 2004). The program converts elastic dislocation and 56 crustal deformation values to strain, and results can be compared with seismograms and GPS data. A carefully calibrated finite fault rupture model provides the basis for accurate strain modeling. Recent advances in technology have resulted in the development of geodetic observational tools. These tools range from satellite-derived GPS devices to regional and global scale seismic monitoring stations equipped with broadband digital strong motion stations. Researchers combine static GPS measurements with strong motion data sets and teleseismic P-waves to model the complexities inherent in fault dislocation. GPS stations provide the location and displacement measurements generated by the fault rupture, while seismic stations provide the arrival times for the passing seismic waves. The seismic data can be used to help define the temporal evolution of the fault rupture (Ji et al., 2005). Teleseismic P-waves are often used to test the validity of the model in replicating the release in energy required to produce the observed dislocation fields. Explanation and Application of Finite Fault Models Elastic dislocation theory present faults as discontinuities or dislocations in an otherwise perfectly continuous elastic medium (Okada, 1992). Specifically, faults are represented as surfaces across which there is defined to be a discontinuity in the elastic displacement field. The surrounding medium can be modeled as a uniform elastic half- space with boundary conditions of zero normal and shear tractions at the free surface and zero displacement at an infinite distance from any dislocation or as a more detailed layered earth half-space model which incorporates more realistic geologic heterogeneities 57 into the surrounding elastic medium. Okada?s (1992) formulation expresses the displacement field generated by rupture at any given point as a function of fault locations, dimension, geometry, slip and the elastic constraints (Healy et al., 2004). Fault Parameters and Input File Coulomb input files are constructed from parameters grouped into three different sections. The first section includes (1) source data, (2) elastic parameters, (3) friction coefficient, and (4) regional stress tensors. Specifically, this section provides parameters and constraints for an elastic half-space medium. Two elastic parameters are added to the input file: (1) Poisson?s ratio and (2) Young?s Modulus. Poisson?s ratio (v) is an elastic constant that gives the relationship between contraction and extension of a geologic material and is given by v = -?1/?3, where ?1 is the strain in the principle stress direction and ?3 are the strains in mutually orthogonal directions. Young?s Modulus (E) relates the ratio of tensional stress to the resulting tensional strain. It is given by E = ?/?, where ? is the applied stress and ? is the resulting strain. These two values show how the elastic medium will deform when rupture occurs. The coefficient of friction, along with compressive normal stress, determines how easily fault motion will be achieved. Higher coefficients of friction require more frictional force and result in lower displacements. Regional stress tensor information gives the directions of principle stresses acting upon the source faults. The second section of the Coulomb input file provides: (1) fault geometry, (2) source fault positions and slip, and (3) grid information. This section contains the necessary source or receiver fault parameters for modeling. The model geometry consists 58 of the x and y locations at which calculations are made and the grid information in which the calculations proceed. Also included are dip angles and slip sense for each point listed in the input file. The third section of the input file provides: (1) graphical representation parameters and (2) plotting parameters. Grid parameters constrain the size of the modeled area with regards to two dimensions and also control the increment of size for each grid. Size parameters control the size of the graphic output. Shade and color parameters control the increment for dilatation and stress values, and exaggeration parameters control the values for displacement, distorted grid, and slip line. Cross-section parameters control the size and increments for any cross-sections run in Coulomb. Earthquakes occur as a result of localized stress concentrations at suitably oriented anisotropies (Lin and Stein, 2004). Fault rupture occurs when the accumulation of stress exceeds the yield point. Within the modeling process there are two important types of faults mentioned: (1) source fault(s) and (2) receiver fault(s). Source faults are those faults that have slip components and impart stress on the surrounding region; in this case the Chelungpu fault serves as the source. Receiver faults are those faults that do not slip and receive stress from the source rupture. They can be moved closer to or further away from failure by the changes in Coulomb stress. Fault Model and Geometry This study uses the fault model developed by Ji et al. (2005). This fault model uses three planar fault segments to approximate the rupture plane during the Chi-Chi mainshock (Fig 18). Fault 1 is located along the north-south trending Chelungpu fault and 59 has a strike of N3?E. Fault 2 is connected to fault 1 in the north and follows along the eastward trending rupture for approximately 13 km with a strike of N80?E. Fault 3 matches the bend in the surface break at the southern end and has a strike of N45?E. All of the fault segments have a dip angle of 29? and all fault segments were extended down to a dip width of 17 km (Ma et al., 2005). To simulate the slip distribution accurately along the fault segments, Ma et al. (2005) divided each fault into smaller rectangular regions of equal area, or subfaults. All of the subfaults have the same dimension of 3.8 x 3.7 km. There are a total of 360 subfaults (Fig. 18); however, not all of them are used to generate the synthetic fault rupture. Because most of the slip attributed to rupture was located to the surface of the ?wedge-shaped? block formed at the intersection of fault 1 and 2. Ji et al. (2003) found it a plausible assumption to set the slip amplitudes of the subfaults below the surface of the ?wedge-shaped? block to zero. Thus, the actual number of contributing subfaults used in the generation of the synthetic fault rupture is only 324. Within the modeling process each subfault was allowed to have a slip amplitude ranging from 0 to 24 m. For fault segments 1 and 3 the rake angle was fixed between 0? and 180? to suppress downward slip not observed during the rupture. Fault 2 was allowed to have a rake angle ranging from 0? to 360? because there were some observations of normal faulting and associated downward slip along this segment of the rupture. Average rupture velocity as calculated from the GPS measurements and waveform data, was allowed to vary between 1.5 to 3.0 km s-1. 60 Figure 18. The Chelungpu fault is divided into three segments (1 to 3) to account for the geometry and the slip pattern logged by GPS stations. Slip varied from 0 to 18 m. The largest amounts of slip occurred near the intersection of fault segment 1 and 2. Fault model from Ji et al. (2005). 61 PFLOW Modeling Output values of strain calculated from the deformation models provide the initial conditions to relate strain to pore-pressure. Pore-pressure diffusion evolves according to spatial characteristics of the elastic medium and the governing equations. PFLOW fluid pressure modeling results can then be compared to the observed water-level change and stream discharge data to test the coseismic volumetric strain hypothesis. Unique to the PFLOW algorithm is the ability to spatially vary the hydraulic conductivity of the modeled area in three dimensions. This application allows users to explore more realistic heterogeneous geological models. Input parameters for PFLOW include calculation area and depth, diffusion time, number of time steps, Skempton?s coefficient, compressibility, boundary conditions, and hydraulic conductivities of the modeled space. PFLOW calculates the flux gradient and pore-pressure at the time steps specified. Poroelastic Governing Equations Systems of partial differential and algebraic equations can be used to describe the relationship between stress (?), strain (?), pore pressure (P), and increment of fluid content (?) in a deformed porous media (refer to Wang and Anderson, 1982). The following governing equation describes the diffusion of induced excess pore pressure, QPktPS s ??????? )( ? (1) 62 where Ss is the hydrogeologic specific storage of rock, k is the permeability, ? is the viscosity, and Q is a fluid source term, as induced by seismic faulting. PFLOW was developed to investigate the first-order evolution of pore-pressure changes induced by changes in Coulomb stress and their possible relation to water-level changes in wells or changes in stream discharge. It utilizes linear or quadratic finite elements for spatial discretization and first or second order, explicit or implicit finite difference discretization in time. PFLOW uses finite difference discretization to evaluate the pressure at each node with respect to time. After temporal discretization is performed the equation which solves for pressure becomes a semi-discrete (discrete in time, continuous in space) formulation ?? ? ?+1 ? ?? ?? ? ?? ? ? ??? ?+1 = ???+1 + 1 ? ? Qn + 1 ?? ? ? (k ??P n) (2) where the superscript n represents the discrete time level at which the function is evaluated and ?t is the specified time step. The parameter ? determines the type of discretization. For ? = 0, the discretization is an explicit first-order (known as a forward Euler). Explicit discretization relies on previous time steps to calculate future ones. Specifically, use of the explicit scheme implies that the values of the space derivatives at the old time level (n) is the best approximation for the future time step. For ? = 1, the discretization is an implicit first order (known as a backward Euler). Implicit discretization is not expressed explicitly in terms of known quantities and requires that pressure at the future time level be obtained as the solution of a system of linear equation. The space derivatives are thus approximated at the advanced time level (n+1) and it is assumed that value of the space derivative at the future time is the best approximation. 63 The use of ? = ?, it is presumed that the best value lies halfway between time levels (n) and (n+1). This is termed the Crank-Nicolson method and it is an implicit second-order discretization. After spatial (finite element) discretization, equation (1) becomes a system of linear algebraic equations, with very sparse matrices. Once the initial (coseismic) pressure P0 caused by faulting is obtained from a deformation modeling program, such as Coulomb, the system can be solved for Pn for n?1. At each time, the system is solved using a sparse direct solver, a built-in function of MatlabTM. PFLOW offers a tool to explore how changes in Coulomb stress can influence pore-pressure through time following an earthquake. Coulomb stress transfer (?CFS) can be caused by fault rupture and can have significant impact on the fault zone region. It is defined as ???? = ??? + ?(?? ? ???) (3) where ??s is the change in shear stress, m is the coefficient of friction, ?P is the change in pore pressure, and ??n is the change in normal stress. Positive Coulomb stress changes promote failure, whereas negative changes in Coulomb stress discourage failure along a fault. A positive increase in normal stress or a decrease in pore-pressure therefore will lead to negative Coulomb stress change and discourage failure along specified faults. Stress (?) can then be related to strain (?) by Young?s modulus. Strain generated by stress leads to net changes in volume of the media. The fractional volume change generated by these fault ruptures can be expressed by the equation ?? ? = ? 1 ? 2? = ???? (4) 64 where v is Poisson?s ratio and ? is the bulk compressibility. The strain derived from Coulomb stress will initiate the development of excess pore pressure (overpressure development), pressure gradients, and corresponding flux. In this research (one-way coupling), the pore pressure change is calculated in the main PFLOW driver script by multiplying the calculated stress change by the Skempton?s coefficient (B) ?? = ??? (5) Since variations in pore-pressure will lead to a net gradient, pore-fluids will flow to alleviate overpressure and restore a steady-state equilibrium condition. The equation that describes fluid flow in a 3D heterogeneous media is described by ?? ???? = ??? ?? ???? + ??? ?? ???? + ??? ?? ???? (6) where Kx, Ky, and Kz are the hydraulic conductivities in the x, y, and z directions, h represents the hydraulic head, and Ss is the specific storage for the aquifer. Therefore, since induced pressure changes will lead to the development of hydraulic gradients and expulsion of fluids out of storage, flow will take in place in three dimensions and be dictated by the hydraulic conductivity in the respective directions. The net pore pressure change over time depends on the magnitudes of the volumetric strain and also the hydraulic conductivity (higher conductivity equals faster diffusion). 65 Numerical Modeling Results Introduction This chapter provides the results of three individual case studies for investigation into the relationship between crustal deformation and pore-pressure changes. The first case study uses a simplified fault model to explore the basic relationships between strain and induced pore-pressure and evolution. Hydraulic conductivity (K) values are then changed for each simplified model to determine the effect of the type of geology and hydraulic conductivity (K) on pore-pressure evolution with time. Building off these simplified fault models, a second case study is performed to illustrate the pore-pressure response of the pore-fluids within the island of Taiwan in response to the crustal deformation generated by the 1999 Chi-Chi earthquake. In this case study, the finite-fault model of Ji et al. (2005) is used to replicate the dynamic strains and rupture kinematics associated with the large magnitude earthquake. This Chi-Chi case study is divided into two distinct models. The first model is fixed with a homogeneous geology or uniform hydraulic conductivity (K), whereas the second model contains a more complex layered geologic medium with spatially varying values of conductivity. A third case study solely targets the poroelastic response of the Choshui River alluvial fan to the crustal deformation generated by the earthquake. This study involves the most complexly layered geologic media and was modeled off of the stratigraphy presented by Lai et al. 66 (2004). The modeled results for the Choshui River alluvial fan and the island of Taiwan as a whole form the basis for comparison with the observed hydrologic changes. Case Study 1: Simplified Right-Lateral Strike-Slip Fault The simplified strike-slip fault was constructed using Coulomb 3.1?. The fault plane trends due north for 30 km and has a homogenous slip of 3 m (i.e., strain = 1.0 x 10-4). Rupture occurs with a pure strike-slip motion. The fault plane dips 90? and downwards to a depth of 10 km. Poisson?s ratio (?) was specified to be 0.25 and the Young?s Modulus set to 8 x 105 bars. The coefficient of friction was set to 0.4, meaning that the fault plane contains some fluid lubrication. Modeling of this simplified right-lateral fault rupture shows a distinct four-lobe strain pattern (Fig. 19). Generally, compressional strain (blue zones) occurred in the direction of fault slip, whereas opposite to the sense of slip, positive dilatational strain occurred (red zones). Near the northern fault tip, maximum compressional strain develops (-1.0 x 10-4). Similarly, at the southern fault tip maximum dilatational strain occurs (1.0 x 10-4). Away from these localized zones, the strain field diminishes. Two cross-sections constructed for the modeled area perpendicular to the fault plane down to a depth of 10 km are shown in Figure 19. The first cross-section transverses 3 km north of the northern fault tip and extends west to east for 32 km. This section shows that west of the fault plane, compressional strain occurs. Conversely to the east of the fault plane, dilatational strain occurs. At the center of the cross-section, a zone of compressional strain is found. This zone correlates well to the compressional zone near the fault tip in the plane view strain map. The second cross-section transverses 3 km 67 south of the southern fault tip and extends west to east for 32 km. This section shows that to the west of the fault plane, dilatational strain occurs. To the east of the fault plane, compressional strain occurs. At the center of the cross-section, a zone of dilatational strain occurs. This zone corresponds to the zone of maximum dilatational strain, found on the plane view strain map. 68 Figure 19. (top) Plane view strain map shows that zones of compressional (blue) and dilatational (red) strain develop along a right-lateral strike-slip fault (green). (bottom) Cross-sections show the resultant strain field calculated at the northern and southern fault tips. Black arrows indicate sense of motion. 69 Pore-Pressure Response to Simplified Right-lateral strike-slip faultt Coupled coseismic pore-pressure modeling mimics the calculated strain pattern observed for this simplified right-lateral strike-slip fault, but opposite in sign (Fig. 20). Two coseismic pressure figures were generated for the right-lateral strike-slip model (Figs. 20 and 21). Maximum induced excess and decreased pore-pressure changes reached 5.0 x 107 Pa and -5.0 x 107 Pa respectively. These values correlate well with localized zones of maximum strain (+/- 1.0 x 104) located near the fault tips (Fig. 19). In contrast, Figure 20 shows that the four-lobed coseismic pressure field generated farther away from the fault tips reaches maximum magnitudes of 105 Pa and -105 Pa. Thus, there is a contrast of two orders of magnitude between the induced pressure fields located near the fault tips, and those within the four lobes away from the fault. Two pore-pressure diffusion models were generated for the right-lateral strike-slip fault. The first model has a homogeneous hydraulic conductivity value of 3 x 10-4 m/d and boundary conditions were set so that flow could only be achieved through the surface (elevation = 0 m). By not spatially varying hydraulic conductivity, the model space assumes a constant geology. In this case, a hydraulic conductivity of 3 x 10-4 m/d is consistent with that of a sandstone. 70 Figure 20. Coseismic pore-pressure field for simplified right-lateral strike-slip fault. Excess pore-pressure developed in the direction of fault slip (red). Under-pressure developed opposite the sense of fault motion (blue). Black arrow indicates fault motion. 71 Figure 21. Coseismic pore-pressure map illustrating only? zones where maximum pore- pressure developed. Near the northern fault tip, maximum excess pore-pressure developed. Near the southern fault tip, maximum under-pressure developed. These zones are two orders of magnitude greater the rest of the pressure field (Fig. 15). Black arrow indicates sense of motion. 72 After a time window of 1 day, the diffusion model changes distinctly. Instead of the expected gradual diffusion of the four-lobes back towards pre-seismic conditions, the modeled pattern shows that the northern half of the grid is dominated by induced excess pressure. Conversely, the southern half of the grid is dominated by underpressure (Fig. 22). Modeling shows that maximum overpressure and underpressure zones remain localized near the fault tips (+/-1.0 x 106 Pa). In the northern half of the grid, excess pore- pressure zones surrounding the fault tip reaches a maximum of 5.0 x 105. Similarly, in the southern half, maximum underpressure surrounding the southern fault tip is maintained around -5.0 x 105 Pa. In this case-study, modeling results suggest that the zones of maximum pressure change near the fault tips dictate the corresponding diffusion pattern. Large differences of excess pressure establish high pressure gradients in the northern and southern segments, which control the direction of flow and resulting pore pressure pattern (Fig. 23). These strong gradients reduce diffusion from taking place in a north to south direction in large areas away from the fault. The pressure gradient established between the northern maximum excess pressure zone and the adjacent zone of under-pressure (to the east) is 3.33 x 106 Pa/km. Comparing this to the pressure gradient calculated between the zones of excess pressure and under-pressure east of the fault (3.33 x 103 Pa/km), it can be deduced that groundwater will flow in the direction of maximum gradient. Similarly, the pore-pressure front will migrate in the direction of maximum gradient. Therefore, the modeled pattern to the north is linked to the node of maximum overpressure near the fault tip. Initially the northeast quadrant is dominated by underpressure or decreased pore pressure (Fig. 20). Very quickly over-pressure spreads to the northeast quadrant (Fig. 22) 73 as fluid flow and the migrating pressure-front move down gradient. In the south, groundwater flow and the pressure gradient is dictated by the maximum underpressure near the southern fault tip. Overpressure in the southeast quadrant dissipates quickly and drops below pre-seismic condition as water there moves quickly toward the underpressure center. Figure 22. 1-day pressure map shows that the northern section of the grid is dominated by excess pressure and the southern section is dominated by under-pressure. Zones of maximum pressure dominate the diffusion pattern. 74 Figure 23. Coseismic pressure gradient field shows that the zones of maximum pressure (near fault tips) dictate the diffusion pattern. Gradients are orders of magnitude greater in an east to west direction than in a north to south direction. Thus the pressure-front and pore fluids migrate in the direction of maximum gradient. Dotted black-line indicates no- flux. Large gradients also exist between the zones of maximum overpressure and underpressure near the fault tips. Arrows indicate gradient. 75 After 10 days (Fig. 24), the pore-pressure pattern and distribution remain similar to that of the 1-Day pattern, but are smaller in magnitude. Maximum zones of overpressure exist to the north at depth, but have been dissipated to 5.0 x 103 Pa. Similarly, maximum zones of underpressure exist to the south with depth, but have been reduced to 5.0 x 10-3 Pa. In the upper 2 km, overpressure has further diminished to 1.0 x 103 Pa, while under-pressure had dissipated to -1.0 x 103 Pa. The east-west pressure gradient established by the induced coseismic pressure field has reduced pressure changes by two orders of magnitude. Large north-south gradients exist between the zones of maximum overpressure and underpressure near the fault tips. Flow from north fault tip toward south fault tip allows continuous dissipation of excess and decreased pore pressure. After 20 days (Fig. 25), the pore-pressure pattern still resembles that of the Day 1 pattern. However, maximum values of pore-pressure change have been reduced by over 99%. At depth, overpressure is present, but only reaches a maximum of 5 Pa. Under- pressured zones show a similar pattern,existing to a greater extent at depth, but only to a maximum of -5 Pa. In the upper 2 km, overpressure and underpressure has been reduced to 2 Pa and -2 Pa, respectively. Thus, a homogeneous hydraulic conductivity (K) of 3.0 x 10-4 m/d dissipates the maximum pressure changes (+/- 5.0 x 107 Pa) associated with the right-lateral strike-slip rupture within 20 days. 76 Figure 24. 10-day pore-pressure map shows a similar pore-pressure pattern to that of the 1-day map (Fig. 17). However, maximum pressure changes have dissipated by 2 orders of magnitude. The greatest values of overpressure and underpressure exist at depth. 77 . Figure 25. Pore-pressure model shows that a hydraulic conductivity value of 3.0 x 10-4 m/d is enough to dissipate the induced pressure-change field in 20 days. Overpressure and underpressure remain; however, they have been reduced to below +/- 5 Pa. 78 The second model features a reduced hydraulic conductivity (3 x 10-5 m/d). Similar to the first model, hydraulic conductivity (K) was not allowed to vary within the modeling space and boundary conditions only allowed flow through the top surface. Because hydraulic conductivity does not affect the initial pore-pressure distribution, the induced coseismic pressure field is the same as that of Model 1 (Figs. 20 to 21). Similarly, the induced pressure gradient remains the same as Model 1 (Fig. 23). After 1 day (Fig. 26), the pore-pressure pattern and distribution is more complex than that of the first model (see Fig. 22 for comparison). The northern section of the grid is still dominated by induced excess pressure; however, localized zones of underpressure are found. Excess pressure is mostly on the order of 105 Pa. Maximum overpressure occurs near the northern fault tip and is on the order of 106 Pa (not shown). Underpressure in the north, exists on the order of -105 Pa. To the south, the pressure field is symmetrical to that of the north, but opposite in magnitude. Maximum underpressure is found near the southern fault tip and is on the order of -106 Pa (not shown). The south is dominated by underpressured zones, yet a few localized zones of overpressure exist. Generally, under- pressure zones are around -105 Pa and excess pressure is on the order of 105 Pa. In comparison to the Day 1 results from Model 1 (Fig. 17), this model (Fig. 26) shows that the excess pore-pressure field has not dissipated as rapidly. Reducing the hydraulic conductivity one order of magnitude has retained the coseismic pore-pressure distribution better over a 1-day period of time, despite the same induced pressure gradient. Lower conductivity reduces the ease of fluid-flow and thus the pressure-front cannot migrate as quickly through the model space. 79 Figure26. 1-day pressure map for the low-K model shows that coseismic pore-pressure changes are retained better over a 1-day time span. Overpressure dominates the northern section (red) and under-pressure dominates the southern section (blue). 80 After 10 days (Fig. 27), the calculated pore-pressure resembles that of the 1-day map from model 1 (Fig. 22). The northern section is dominated by excess-pressure, centered around the northern fault tip. Maximum over-pressure reaches 5.0 x 105 Pa near the fault tip and grades laterally to approximately 3.0 x 105 Pa within 15 km. Small zones (Fig. 21) of underpressure still persist after 10 days in the north. These zones are localized and generally do not exceed -0.5 x 10-5 Pa. To the south, the pressure field shows a similar pattern. However, it is dominated by under-pressure. Localized zones of overpressure are observed and show a similar distribution to localized under-pressure zones in the north. Maximum under-pressure reaches -5.0 x 105 Pa. Also evident from this map is that pressure changes are maintained better at depth. In the next 10 days, the pressure field dissipates very little (Fig. 28). The only major difference between 10 days (Fig. 27) and 20 days (Fig. 28) is that the isolated zones of under-pressure and over-pressure have dissipated. Maximum overpressure is centered around the northern fault tip and remains to be 5.0 x 105 Pa. This zone diminishes radially to around 0.5 x 105 Pa after 30 km. In the south, maximum under- pressure is centered around the southern fault tip and diminishes radially, values are similar to that of the northern section, but opposite invalue. With depth, zones of induced pressure change are better maintained and occupy a larger area. 81 Figure 27. 10-day pressure map shows a similar pattern to that observed in Figure 17. Maximum over-pressure is centered around the northern fault tip and maximum under- pressure is centered around the southern fault tip. Isolated patches of under-pressure and overpressure remain in the northern and southern sections, respectively. Induced pressure changes are better maintained at depth. 82 Figure 28. 20-day pressure map shows that very little dissipation has occurred since the 10-day map. The only noticeable difference is that isolated zones of under-pressure and overpressure have dissipated completely. Pressure changes are maintained better at depth. 83 After 100 days (Fig. 29), the pressure field resembles that of the 10-day map from Model 1 (Fig. 24). Excess-pressure in the upper 2 km of the northern section remains, but only on the order of 1.0 x 103 Pa. Similarly, underpressure exists in the upper 2 km of the southern section, but reaches only to -1.0 x 103 Pa. Pressure changes at depth are better maintained; however, these zones have dissipated to +/- 3.0 x 103 Pa. There is a strong correlation between Model 1 (K = 3.0 x 10-4 m/d) and Model 2 (K = 3.0 x 10-5 m/d). By decreasing the conductivity by exactly one order of magnitude, the dissipation time increases by about ten-fold. Therefore, with a hydraulic conductivity of 3.0 x 10-5 m/d, the recovery time for the same initial pore-pressure disturbance (from Model 1), increases from 20 days to 200 days. This case study highlights the importance of hydraulic conductivity as the most important variable for determining the recovery rate of pore-pressure in response to strain generated from earthquakes. The result also indicates that the induced coseismic pressure gradient dictates the direction of fluid-flow and thus the distribution of excess and decreased pore pressure, which could potentially affect the spatial distribution of aftershock activities in the fault zone. 84 Figure 29. 100-day pressure map resembles that of the 10-day map from model 1. Overpressure remains to the north on the order of 105 Pa. Under-pressure remains in the south on the order of 105. Decreasing the hydraulic conductivity by one order of magnitude increases the recovery time by ten-fold. 85 Case Study 2: Chi-Chi Earthquake The second case study numerically models the strain fields generated by the Chi- Chi earthquake (Ma et al., 2005) and corresponding pore-pressure fields that resulted. The Chelungpu fault rupture is much more complex than the example described previously. The fault contains several compressional bends where thrusting motion transforms into more oblique slip (Ji et al., 2005). Also, different from the simplified strike-slip fault model is differential slip on the fault plane. Specifically, different parts of the fault slip more than others, leading to eccentricities in the overall strain pattern. Using the fault model of Ji et al. (2005) for the Chi-Chi earthquake, the calculated strain pattern is consistent with what was observed in the field (Ji et al., 2005). Like the first case study , this study examines the role hydraulic conductivity (K) plays in dissipating pore-pressure. The models examined are (1) a homogeneous geologic model (conductivity does not vary spatially) and (2) a more realistic heterogeneous geology in which the hydraulic conductivity (K) is allowed to vary spatially. Strain modeling for the Chi-Chi earthquake reveals a complex strain pattern (Fig. 30). The hanging-wall block (east of fault trace) initially slipped near the fault southern terminus. As rupture progressed northward, fault slip gained momentum and at fault bends the kinematics of rupture, switching to a more left-lateral strike-slip (Ji et al., 2005). These factors led to increased deformation along northern sections of the fault and also larger horizontal displacements. 86 The rupture generated large zones of dilatational strain (red) parallel to fault segment 1 (Fig. 30). Near the rupture zone, positive dilatation reaches a maximum value of 1 x 10-5. Positive dilatational strain gradually decreases to the east and west away from the fault plane, diminishing to 0.2 x 10-5 approximately 50 km from the rupture zone (Fig 23). Near the northern (segment 2) and southern terminus (segment 3), compression occurs (blue) (Fig 30). Compressional strain reaches a maximum of 1 x 10-5 nearest the fault tips, but diminishes quickly to 0.2 x 105 approximately 25 km from the tips. Strong zones of compression (negative dilatation) occur on the leading edge of the hanging-wall (near surface). As the hanging-wall move upwards, the top edge of the fault block is compressed. Local nodes of either compression or dilatation can be found throughout the fault plane; these anomalous zones are most likely the result of differing amounts of frictional resistance. (slip) as well as differential stress build-up on the fault plane before rupture (i.e., different principle directions) (Ji et al., 2005). 87 Figure 30. Plane view of the strain field calculated for the Chi-Chi earthquake (top) from rupture model of Ji et al. (2005). The model shows that zones of dilatational strain (red) developed mostly parallel to fault segment 1 (middle). Zones of compression develop near the northern (segment 2) and southern fault tips (segment 3) (blue). Cross-section (bottom) shows compression along the footwall (blue) and dilatation along the hanging- wall (red). Black arrows indicate sense of motion. Green (top) and black (bottom) lines illustrate the fault. 1 2 3 88 Pore-Pressure modeling of Chi-Chi earthquake For this study, pore-pressure models for the Chi-Chi earthquake were constructed using two distinct spatial distributions of hydraulic conductivity (K). The first pore- pressure model used a uniform hydraulic conductivity of K = 3 x 10-4 m/d throughout the grid space. Boundary conditions allow for flow to be achieved out of the top of the model space (surface). The coseismic pore-pressure response (Fig. 31) mimicked the strain field calculated from the deformation modeling program (Fig. 30), but opposite in convention. Parallel to fault segment 1, zones of under-pressure developed (blue) in response to the dilatational strain field. Coseismic decreases in pore-pressure reached -6 x 104 Pa, while zones of overpressure exceeded 2 x 104 Pa. Increased magnitudes of pore-pressure with depth are in good agreement with strain modeling, as deformation was calculated to be more extensive near the hypocenter (Ji et al., 2005). After 4 days (Fig. 32), the pressure field had decreased by approximately 3 orders of magnitude. Induced zones of overpressure are located parallel to the fault trace (blue), but have diminished to around 6 x 103 Pa. Figure 32 illustrates that at depth, these zones are maintained longer, but could be a function of Neumann?s boundary conditions. Localized zones of underpressure remain near the fault tips at depth, but likewise had diminished to around -3 x 103 Pa. At the surface, a state of zero pressure had already been achieved within 4 days, most likely by fluid-flow through the surface. After 25 days (Fig. 33), the four lobed pore-pressure patterns had changed. Most of the excess pore-pressure has dissipated to below 2 x 10-2 Pa by this time. Zones of underpressure remain more prevalent to the east of the fault and with increasing depth. 89 These zones occur because the fault dips towards the east and regions proximal to the fault plane underwent the most extensive deformation. However, modeling shows that these zones of underpressure correspond to only about -8 x 10-2 Pa. Induced pressure changes in areas farther away from the fault zone (10-15 km) had diminished to below +/- 1 x 10-2 Pa (Fig. 33). Therefore, after 25 days a constant hydraulic conductivity of K = 3 x 10-4 m/d was enough to dissipate the induced pressure fields back to near zero. After 50 days (Fig. 34), pressure changes are reduced back to zero. The model shows that no zones of underpressure or overpressure remain. With this hydraulic conductivity (K = 3 x 10-4 m/d), pore-pressure modeling shows that all anomalous hydrologic changes associated with changes in pore pressure should have ceased by 50 days. 90 Figure 31. Coseismic pore-pressure calculated for the Chi-Chi earthquake. Model shows that the resultant pore-pressure field mimics that of the strain field, but opposite in sign. Zones of induced under-pressure developed parallel to fault segment 2, while zones of overpressure develop near the northern and southern tips of the fault. Underpressure was calculated to reach -6 x 104 Pa and overpressure around 2 x 104 Pa. Red line indicates fault trace. Hanging-wall is located to the east of the fault plane. 91 Figure 32. 4-day pressure map shows that all induced coseismic pressure changes had been reduced by three orders of magnitude. Zones of overpressure remain parallel to the fault trace and are maintained better at depth (blue). Localized zones of underpressure remain near the fault tips, especially with depth (red). 92 Figure 33. Resultant pore-pressure fields after 25 days. Zones of over-pressure are highly localized and reduced to less than 2 x 10-2 Pa. This suggests that a hydraulic conductivity of K = 3x10-4 m/d was enough to dissipate the induced over-pressure over this time-span. Circular zones of under-pressure remain throughout the sub-surface, however these zones correspond to less than -6 x 10-2 Pa. Red line indicates the fault trace. Hanging-wall is located to the east of the fault plane. 93 Figure 34. Resultant pore-pressure field 50 days after the Chi-Chi earthquake. All induced zones of pore-pressure caused by deformation had completely dissipated by this time. Thus, a hydraulic conductivity (K) of 3x10-4 m/d was enough to dissipate these zones via fluid-flow. Red line indicates surface fault trace. Hanging-wall is to the east of the fault plane. 94 The complex geology of the island of Taiwan cannot be represented in a numerical model. However, a simplified fault model can provide insight into pore- pressure changes. Thus, a second, three-layer model was created to represent a generalized geology of the island of Taiwan (Fig. 35). The hydraulic conductivity (K) decreases with depth and each successive layer. A hydraulic conductivity of 2.74 x 10-4 m/d was assigned for strata in the first 4 km. This value is the average appropriate for the lithologies and sediments that dominate the Choshui River alluvial fan and Western Foothills region. The second layer extends down from 8 km and has a hydraulic conductivity value of 3.0 x 10-5 m/d. This layer is consistent with a silt- to fine-grained sandstone. The third layer extends to 10 km depth and was assigned a hydraulic conductivity value of 3 x 10-7 m/d. This value is consistent with a well-consolidated mudrock or bedrock. The last layer extends down to 10 km and was given a conductivity of 1.0 x 10-8 m/d (consistent with bedrock). The layered Chi-Chi model illustrates a slower recovery pattern. The more conductive upper layers dissipate induced pore-pressure more rapidly than those of the lower layers. Coseismic induced excess pressure (refer to Fig. 31) reached 2.0 x 104 Pa, while underpressure zones reached -6.0 x 104 Pa, as in the homogeneous model. After 25 days (Fig. 36), the excess pore-pressure field was reduced to below 500 Pa near the surface and to around 1.0 x 103 Pa with increasing depth. Underpressured zones remained and generally increased with depth, reaching -2.5 x 103 Pa at 8 km. This model shows that lower values of conductivity retain induced pressure-changes longer. Fluid flow is not as easily achieved with low conductivities and thus the excess cannot be relieved as rapidly. 95 Figure 35. Schematic diagram of the hydraulic conductivity used for the layered Chi-Chi pore-pressure model. 96 Figure 36. Resultant strain field after 25 days. Near the surface, induced pressure changes are reduced to less than -100 Pa. Coseismic overpressured zones near the surface had effectively dissipated. With depth, induced pore-pressure changes persist, ranging from 1.0 x 103 Pa to -2.5 x 103 Pa. Increasing depth also corresponds to decreased hydraulic conductivity in this model and thus, lower conductivity values trap induced pressure changes and make fluid flow more to difficult to achieve dissipation. Red line indicates fault trace. Hanging-wall is to the east of the fault plane. 97 After 50 days, overpressured zones near the surface had effectively decayed to zero (Fig. 37). However, at depth, pressure remained around 1.0 x 102 Pa in the less conductive layers. Underpressure zones near the surface returned to zero pressure within this time span as well, suggesting that fluid flow from zones of high pressure to zone of lower pressure dissipated the induced pressure changes. With depth (4 to 8 km), under- pressured zones are sustained and ranged from approximately -100 to -500 Pa. In less conductive layers, fluid flow and the migration of the pressure front could not be achieved to re-equilibrate the pressure. In the next 50 days pore-pressure continued to dissipate steadily. After 100 days (Fig. 38), the overpressured zones remaining at depth (4 to 8 km) had recovered to about 10 Pa. Zones of under-pressure at these same depths had dissipated to -50 Pa. Two hundred days after the Chi-Chi earthquake (Fig. 39), pressure had normalized back to pre-seismic conditions (zero excess pressure). In the more realistic layered geology model, dissipation took between 100 and 200 days. Also, of special interest, the regions pertaining to the Choshui fan and Western foothills had completely dissipated all induced pressure changes in the upper 2 km by 50 days. Therefore, any hydrologic anomalies generated by coseismic strain, in general, should not persist past 50 days in these regions although excess pore pressure may maintain within localized low-conductivity zones. 98 Figure 37. Resultant pore-pressure field after 50 days shows that zones of excess pore- pressure remain intact from 4 to 8 km depth, but does not exceed 100 Pa. Zones of decreased pore-pressure remain to be around -500 Pa at 8 km depth. Only small amounts of decreased pore-pressure remain at 2 km depth. Red line indicates fault plane. Hanging- wall is to the east of the fault plane. Boundary conditions were set so that flow could be achieved out of the top of the model. 99 Figure 38. Resultant pore-pressure field after 100 days. Pressure changes are only maintained at depth in less conductive layers. Overpressure exists at 8 km depth and is calculated to be around 10 Pa. Underpressured zones range from -25 to -50 Pa within the fault zone at 8 km depth. Red line indicates fault trace. Hanging-wall is to the east of the fault trace. Boundary conditions only allow for flow out of the surface. 100 Figure 39. Resultant pore-pressure field after 200 days shows that all induced pressure changes dissipated to pre-seismic conditions. Modeling suggests that complete recovery took between 100 and 200 days. Red line indicates fault trace. Hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow to be achieved through the surface of the box. 101 Case Study 3: Choshui River Alluvial Fan Choshui Strain Model The final case study focuses solely on the Choshui River alluvial fan region located directly east of the Western Foothills. This region is of the utmost importance as the bulk of the observed hydrological changes were monitored therein. Thus, this model is a very diagnostic test to determine whether or not coseismic volumetric strain could generate the observed hydrologic anomalies. The model covers an area directly to the west of the intersection between fault segments 1 and 3 (Fig. 40) (Ji et al., 2005). The strain pattern calculated for the Choshui River alluvial fan is relatively simple (Fig. 40). The far southeastern portion of the fan, located near the Chelungpu fault?s southern terminus, corresponds to a zone of compressional strain. Compressional strain reaches a maximum of 1 x 105 in the areas surrounding the southern fault tip and grades to 0.5 x 105 Pa approximately 15 km west. The remaining portions of the fan, bordering fault segment 1, are zones of dilatation. Dilatational strain grades from 1.0 x 10-5 Pa near fault segment 1 to 0.1 x 10-5 Pa near the coastline (Figs. 40 and 41). A cross-section through the alluvial fan (Fig. 41) illustrates that the entire hanging-wall block of fault segment 3 was dilated (1 x 10-5) by the rupture event. However, the footwall block is shown to have been compressed by the rupture (-1 x 10-5). West of the footwall, the alluvial fan underwent positive dilatation, ranging from a maximum of 1 x 10-5 near the surface trace to 0.2 x 10-5 approximately 20 km west (Fig. 41). 102 Figure 40. The Choshui River alluvial fan resides in west-central Taiwan and borders the southern terminus of the Chelungpu fault. Fault model of Ji et al. (2005) produces compressional strain (blue) in the southeastern section of the fan. The remaining portion of the fan underwent dilatation (red). The fault is highlighted by a green line. The footwall is located west of the fault plane. Black line indicates coastline. 103 Figure 41. Coseismic strain map (top) and cross-section (bottom) for the Choshui River alluvial fan shows that the hanging-wall is dilated by rupture. The footwall is subjected to compressional strain near the fault plane. West of the footwall, the fan is in the dilatational field. Black line represents fault plane at depth. Black arrow indicates sense of motion. 104 Choshui Pore-Pressure Modeling To represent the interfingering aquifers and aquitards of the Choshui River alluvial fan, the model incorporates eight distinct layers (Table 2). The upper 50 m was assigned a hydraulic conductivity value of 2.74 x 10-4 m/d (consistent with a sandy layer). The second layer extends down to 75 m and represents a clay aquitard (K = 1 x 10-6 m/d). The third layer represents a more conductive sand layer and extends to a depth of 150 m. (K = 2.5 x 10-3 m/d). The fourth layer extends down to 180 m and once again represents a clay aquitard (K = 1 x 10-7 m/d). The fifth layer represents a sand layer extending to a depth of 230 m (K = 2 x 10-4 m/d). The sixth layer has a thickness of 20 m, extending down to 250 m (K = 2 x 10-7 m/d). The seventh layer represents an average of sand and clay sediments and extends down to 9 km. The last layer extends down to 10 km and represents a more consolidated bedrock (K = 1 x 10-8 m/d). Part of this model was synthesized from the work of Lai et al. (2004). The complex layering of the Choshui River alluvial fan creates a complex pore pressure recovery pattern. Overall, the coseismic pore-pressure response of the alluvial fan is consistent with the calculated strain field (Fig. 42). The southeastern portion of the fan shows elevated fluid-pressure, while the rest of the fan became underpressured (Fig 42). Peak induced excess fluid-pressure reaches approximately 1.0 x 104 Pa, while zones of underpressure reach -6.0 x 104 Pa. 105 Layer ID Depth (m) Unit Hydraulic Conductivity (m/d) 1 0 to 50 Aquifer 2.74 x 10-4 2 50 to 75 Aquitard 1.0 x 10-6 3 75 to 150 Aquifer 2.5 x 10-3 4 150 to 180 Aquitard 1.0 x 10-7 5 180 to 230 Aquifer 2.0 x 10-4 6 230 to 250 Aquitard 2.0 x 10-7 7 250 to 9000 Mix of sand and clay 2 x 10-5 8 9000 to 10000 Bedrock 1.0 x 10-8 Table 2. Geologic layering assigned to the Choshui River alluvial fan model. Depth is in meters below the surface. 106 Figure 42. Coseismic pore-pressure reveals that the majority of the Choshui River alluvial fan was under-pressured (blue) by dilatational strain. The southeastern and southwestern portions underwent compression (red). Chelunpgu fault is marked by red line. Hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow out of the surface. 107 After 25 days (Fig. 43), most of the induced excess pressure had been dissipated near the surface. At 2 km depth some excess pressure existed directly beneath fault segments 1 and 3. These zones correspond to the compression of the footwall at depth (Fig. 41). Zones of under-pressure persist with increasing depth. These zones have a maximum value of -2.5 x 104 Pa. Pressure in these zones are consistent greater strain at depth and less conductive layers above and below that prevent equilibration of the pressure field to pre-seismic conditions (Fig. 43). Over the next 25 days, pore-pressure continues to dissipate. After 50 days (Fig. 44), excess pressure has been completely relieved. Zones of decreased pore-pressure do not equilibrate as fast as those in excess pressure zones, possibly because zones of coseismic underpressure were greater in size and extent than overpressured zones. However, dissipation has reduced under-pressured zones to -2.5 x 103 Pa at 8 km depth. Pore-pressure fields after 100 days (Fig. 45) show the same trend. With increasing depth, zones of underpressure remain, but only reaching a maximum of -50 Pa. After 200 days (Fig. 46), all remaining zones of underpressure have dissipated back to pre-seismic conditions (zero pressure). 108 Figure 43. Pore-pressure field after 25 days shows that most zones of over-pressure have dissipated. However, zones of dilatation continue to show underpressure with increasing depth. Red line marks the Chelunpgu fault. The hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow out of the surface. 109 Figure 44. Pore-pressure field after 50 days shows that most zones of over-pressure have dissipated. However, zones of dilatation continue to show underpressure with increasing depth. Red line marks the Chelunpgu fault. The hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow out of the surface. 110 Figure 45. Pore-pressure field after 100 days shows that most zones of overpressure have dissipated. However, zones of dilatation continue to show underpressure with increasing depth. Red line marks the Chelunpgu fault. The hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow out of the surface. 111 Figure 46. Pore-pressure field after 200 days shows all induced pressure changes have dissipated completely. Red line marks the Chelunpgu fault. The hanging-wall is located to the east of the fault plane. Boundary conditions allow for flow out of the surface. 112 Stra in Field Fig. 15 Fig. 23 Dip 30 ? W 29 ? E Strike 0? Se g.1 = 3? Se g. 2= 80 ? Se g. 3= 45 ? Coe fficie nt of f ric tion 0.4 0.4 Young? s Modul us (E ) 8 x 10 -5 8 x 10 -5 Poiss on?s Ra tio ( ?) 0.25 0.25 Sli p Spe cifice d 3 m Allowe d to var y from 0 t o 24 m Fa ult Type Pur e Strike -Sli p Most ly pure reve rse , some le ft- late ral strike sli p ne ar be nds Ca se I D Sim pli fed Right - late ral Chi -C hi earthqua ke Tab le 3 . Input par amete rs for stra in m ode ls us ed i n thi s rese arc h. 113 Discussion Factors Influencing Well Response The most important factors influencing well responses during earthquakes events are thought to be proximity of the well to the fault zone, local geology, and mechanisms controlling hydrologic changes (e.g., Muir-Wood and King, 1993; Rojstaczer and Wolf, 1994; Manga et al., 2003). Secondary factors contributing to well responses are tidal or barometric pressure variations as well as local irrigation or pumping conditions. In this research, responses to tidal influences or barometric pressure deviations are not examined. These variations are typically very small and thus should not play a large role in quantitatively determining the response of each individual well. Local pumping and irrigation in the Choshui river alluvial fan is quite extensive. Thus, accounting for these variations is complex and requires data beyond that available for this study. Therefore, anthropogenic changes are not regarded in this research. The proximity of the well to the seismic zone can have a large impact on the coseismic response. Regions closest to a fault rupture zone undergo the most extensive crustal deformation and are subjected to the greatest peak ground acceleration. If ground acceleration is significant for producing widespread hydrologic anomalies, some of the largest magnitude anomalies should be witnessed in the proximal regions. Lai et al. (2004) noted that some of the largest magnitude changes occurred near the rupture zone. In comparing magnitude of change and distance to the Chelungpu fault, however, they 114 found that this correlation broke down after a distance of approximately 10 km. They offered no specific explanation for this observation. The local geology can play an important role in how the water-table responds to stress generated by large magnitude earthquakes. Unconfined or partly confined aquifers typically show less change and shorter lag times in recovery than do more confined aquifers. This is due to the ease of groundwater flow and diffusion in these unconfined aquifers. Unconfined aquifers in the Choshui River alluvial fan (F1) generally recovered to their steady-state conditions within a few hours of disturbance (Chia et al., 2008). Confined aquifers (F2, F3, and F4), however, showed a step-wise change in water-table elevation following the Chi-Chi earthquake and lag times of weeks to months for recovery. The nature of unconsolidated sediment in the Choshui River alluvial fan plays an important role in influencing coseismic and postseismic responses. Variations in particle size influence the hydraulic properties of the aquifers. As a general rule, aquifers consisting mainly of gravel would be more conductive and permeable than aquifers consisting mainly of sand and silt. Overall the sediment type tends to grade from coarse gravels and sands nearest the Western Foothills to silt and mud-sized particles in distal zones, with interfingering aquifers and confining units down to a few hundred meters depth. Sediment sizes and permeability generally decrease from north to south within the fan and thus could also play a role in governing differing magnitudes of hydrologic response. Field observations appear to quantitatively show that zones of higher hydraulic conductivity correspond to zones of larger coseismic changes. In general, slightly larger 115 changes in well levels are associated with the northern and central portions (?h= 3 to 5 m) of the Choshui river alluvial fan (K = 1.0 x 10-3 m/d) than the southern fan (K = 10-4 m/d) where coseismic changes in hydraulic head averaged only around 1 to 2 m. Several other factors, however, may have influenced the change in well levels. These changes may also be attributed to small differences in compressibility (?). The higher values of hydraulic conductivity also correspond to faster diffusion rates for well recovery as fluid flow is more easily permitted. Towards the south, hydraulic conductivity values decrease and water-table changes recover more slowly. Coseismic Response Of the 58 total wells analyzed from the Choshui River alluvial fan and Western Foothills slope region, only nine wells demonstrate a decrease in water-table elevation in response to the Chelungpu fault rupture (Figs. 47 and 48). The majority of these wells (7) occurred within 10 km of the fault zone. The proximal zone (<10 km) is dominated by consolidated sedimentary sequences, drastically differing from the rest of the alluvial fan, which is comprised of thick unconsolidated Holocene fluvial deposits (Lai et al., 2004). The remaining two wells reside within 15 km of the fault zone. The rest of the wells analyzed (49) illustrate coseismic increases in response to the Chi-Chi earthquake, with the largest positive shifts corresponding to areas within the middle zone of the Choshui River alluvial fan. The one-way coupled models generated in this study indicate that most of the Western Foothills slope region and alluvial fan were subjected to tensional strain by the rupturing of the Chelunpgu fault. In response, these areas should have seen a widespread 116 reduction in pore-pressure (Figs. 47 and 48), consistent with the coseismic volumetric strain hypothesis (Muir-wood and King, 1993). Wells within the slope and alluvial fan regions should likewise show a decrease in water-table elevation. These predicted results agree fairly well with observations from Figure 48 and suggest that coseismic strain may be responsible for the hydraulic anomalies observed in wells near the Chelungpu fault, at least in the F2 aquifer. Pore pressure decreased (-2 to -10 Bar) beneath hydrostatic pressure in these regions, indicating that large decreases in pressure head could lead to a significant coseismic lowering of the water table. An alternate explanation for the observed water-table responses in the Western Foothills and upper slope regions is that fracturing generated by seismic shaking could have led to a downward gradient for fluid-flow. Rojstaczer and Wolf (1994) proposed a causal inducing of hydrologic change by fracturing created by the passing of strong seismic waves through a geologic media. Prevalent vertical to sub-vertical fracturing was observed in the Western Foothills following the Chi-Chi mainshock (Lee et al., 2002). Furthermore, a significant amount of water was found pouring down into tunnels running beneath the Western Foothills following the earthquake (Lin, 2000). Both observations indicate that fractures created by tensile stresses or intense seismic shaking may have facilitated an induced coseismic flow downwards from perched aquifers located in the upper aquifers of the Western Foothills. This hypothesis could explain the reduction in water-table elevation within the consolidated sedimentary rocks of the slope and Western Foothills region. 117 Similarly, most other areas in the Choshui River alluvial fan are associated with dilatational strain (Fig. 47). Pore-pressure modeling reveals that these areas should correspond with decreases in water-level (Fig. 48). However, field observations show that water-levels generally displayed a coseismic increase (Fig. 48). Therefore, a coseismic volumetric strain mechanism (Muir-Wood and King, 1993) cannot be used to explain a majority of the observed coseismic water-table responses in the Choshui River alluvial fan. Seismic fracturing (Rojstaczer and Wolf, 1994) requires the presence of consolidated rock, which is inconsistent with the unconsolidated sediments found in the rest of the Choshui River alluvial fan. Therefore, this mechanism can be eliminated as a possible explanation for the widespread increase in hydraulic head throughout the middle and distal zones of the alluvial fan. 118 Figure 47. Widespread coseismic increases (49 wells) in water-table elevation were observed to occur in zones that underwent tensional strain (grey to red colored zones) in contrast to the coseismic strain hypothesis. Decreases in water-table elevation (9 wells) were observed within 15 km of the fault plane and correspond to areas of tensional strain (grey to red colored zones in the model). Well locations are represented by round dots; black dots correspond to decreases in water-table elevation. The Chelungpu fault is demarcated by a black line. Black star indicates focal point for fault rupture. Calculation depth was set to 0 m. 119 Figure 48. Widespread increases (49 wells) in water-table elevation were observed to occur in zones that were calculated to have decreases in pore-pressure (grey to blue colored zones), contrasting with the coseismic strain hypothesis. Decreases in water-table elevation (9 wells) were observed to occur within 15 km of the fault plane and correspond to areas of decreased pore-pressure (grey to blue colored zones in the model). Well locations are represented by round dots; black dots correspond to decreases in water-table elevation. The Chelungpu fault is demarcated by a black line. White star indicates focal point for fault rupture. Calculation depth was set to 0 m. 120 Manga (2001) postulated a third mechanism for invoking hydrologic change. This hypothesis states that strong seismic shaking could preferentially re-arrange unconsolidated sediments into a tighter packing scheme (compaction). Although, this mechanism cannot be applied to more consolidated sedimentary sequences in the proximal zone, it could be applied to the more unconsolidated Alluvial Plain sediments. The compaction of grains coincident with the passing of seismic waves leads to a sudden reduction of primary porosity. A sudden reduction in primary porosity by seismic shaking could therefore produce an expulsion of pore-fluids out the pore-spaces and lead to the elevated water-tables that were observed (Wang et al., 2004). Although this hypothesis was not tested directly by this study, moderate to severe amounts of liquefaction (75 sites) within the fan and small degrees of hill slope failure were observed (Wang et al., 2003). Both of these observations could lend evidence to coseismic compaction as the dominant mechanism for invoking hydrologic anomalies within the fan. Post-seismic Pore Pressure Diffusion Generally, the wells studied recovered sporadically over the 100 days following the Chi-Chi earthquake ( Appendices A to C). Only 16 of the 41 wells with complete data illustrated a relatively smooth and continuous dissipation pattern over the 100 day time- span. One (KC) of the six wells analyzed in the proximal zone showed a smooth and continuous recovery (Appendix A); 10 out of 19 wells analyzed in the middle zone displayed a continuous recovery back to pre-seismic levels (Appendix B), and 5 of 16 wells in the distal zone displayed a smooth recovery (Appendix C). The majority of the other wells showed an oscillating style of recovery. A few of the wells (Fig 49) that 121 showed coseismic drops in water-level continued to drop with time, perhaps indicating a permanent change in hydraulic conductivity (K) in the aquifer, or permeability enhancement generated by seismic fracturing (Rojstaczer and Wolf, 1994). A significant amount of the oscillating recovery patterns took place in wells that showed minimal coseismic changes. These small variations could be caused by seasonal or daily factors (i.e., local pumping, tidal response, or barometric pressure). It is difficult to properly analyze the diffusion patterns observed over the 100 days following the Chi-Chi earthquake. Unlike the coseismic response of the wells, in which there appears to be correlations among the piezometric response, proximity to the fault, and the type of geology, the well recoveries show no strong correlation with proximity to the fault zone or type of geology. Of the 5 monitoring wells located near the Chelungpu fault that registered coseismic drops (Fig. 49), only one well (LY) showed a continuous decline in water-level over the 100-day period. One well (TH) followed with an extreme rise in water-table in the 50 days following the earthquake (Fig. 50). The other three wells registered an oscillating style pattern in which there was a recovery back towards pre-earthquake levels followed by another drop in water-table. This irregular recovery pattern indicates that the local groundwater system may have been permanently altered by the deformation generated by the earthquake and perhaps a new steady-state condition will be established slowly in concordance with new aquifer properties. Another possible explanation for the variances seen in the water-level data is that aftershocks generated by the transfer of stress and also possibly changes in pore-pressure could have resulted in another series of deformation induced water-table changes. Chi 122 and Dreger (2004) show that two Mw = 6.2 or greater aftershocks occurred in the week following the Chi-Chi mainshock (Table 4). In total there were a total of 5 Mw = 5.8 or greater aftershocks following the Chi-Chi earthquake. Of these five, four were observed to be reverse-slip earthquakes. These findings make a simple analysis of water-table recoveries even more difficult. Regardless of possible permanent changes to the aquifers or impacts created by aftershocks, simple pore-pressure diffusion and resulting groundwater flow does not restore all water-tables to previous levels and thus is probably not the logical mechanism governing the post-seismic response. Table 4. Aftershocks may play a factor in the sporadic water-table patterns following the Chi-Chi mainshock. Two large magnitude aftershocks occur in the week following the Chi-Chi earthquake Aftershock ID Date Mw 0014 9/22/99 6.2 2352 9/25/99 6.3 123 Figure 49. Observed post-seismic recovery of proximal wells (<15 km) shows a general coseismic decrease in water-level. The recovery pattern plotted over 100 days is sporadic. Wells do not generally recover back to pre-seismic levels (0 m). This suggests that volumetric strain and subsequent diffusion are not the dominant mechanism for inducing these changes. 124 The remaining wells were examined with regard to regional context (Fig. 50 to 53) and 100-day responses. The monitoring wells located in the northern portion of the middle and distal fans showed very large coseismic rises and similar declines in the time period following the earthquake. Prevalent oscillating recovery patterns were found especially within wells that showed small coseismic changes. These oscillations can most likely be attributed to daily and seasonal variations. Several prominent up-up responses, in which coseismic increases were followed with continued water-level increase, were found along the coastal region (Fig. 52 to 53). However, it is unknown as to why these responses occurred mainly along the coastal region. Increasing water-table elevations indicate these regions could be induced groundwater sinks or perhaps local irrigation may play a large role in the 100-day response. 125 Figure 50. Observed post-seismic recovery patterns for wells located within the middle of the Choshui River alluvial plain (15-30 km from fault).Strong coseismic increases in water-table are found, eliminating coseismic volumetric strain as an inducing mechanism. Some of the recoveries show a step-wise decrease back to a pre-seismic levels; however, others show very irregular recoveries. 126 Figure 51. Observed post-seismic recovery patterns for wells located within the middle of the Choshui River alluvial plain (15-30 km from fault).Strong coseismic increases in water-table are found, eliminating coseismic volumetric strain as an inducing mechanism. One of the recoveries show a step-wise decrease back to a pre-seismic levels (JL); however, most show very irregular recoveries. 127 Figure 52. Observed post-seismic recovery patterns for wells located within the distal portion of the Choshui River alluvial plain (< 30 km).Strong coseismic increases in water-table are found, eliminating coseismic volumetric strain as an inducing mechanism. One of the recoveries (ST) shows a step-wise decrease back to pre-seismic level. 128 Figure 53. Observed post-seismic recovery patterns for wells located within the distal portion of the Choshui River alluvial plane (< 35 km). Weak coseismic increases in water-table are found, eliminating coseismic volumetric strain as an inducing mechanism. The recovery pattern for every well is sporadic. 129 Rainfall events in the Changhua and Yunlin counties over the 100 days (Fig. 54) following the Chi-Chi earthquake were found to produce relatively small and localized amounts of precipitation. From the 0 to 50 day time span, both counties reported approximately 25 millimeters of rainfall. In the time span from 50 days to 100 days, the counties reported precipitation totaling slightly more than 25 millimeters. Only incremental changes in water-table could be attributed to these events. Figure 54. Rainfall for the cities of Douliou and Changhua show marginal rainfall (around 2 inches total or 55 mm). These amounts are not significant enough to produce widespread variations in the water tables. 130 The chaotic well recovery pattern observed within the Western Foothills slope region and Choshui River alluvial fan may help to shed light on the mechanisms responsible for invoking anomalous hydrologic changes following the Chi-Chi earthquake. PFLOW modeling of the Choshui River alluvial fan and slope show that regions of induced excess or decreased pore-pressure change gradually returned to a pre- seismic steady state condition following the earthquake. Furthermore, these results show that after 100 days, most anomalous fluid-pressure had dissipated. The time period in which the pressure returned back to steady state was dependent upon the magnitude and spatial variation of hydraulic conductivity (K) for the slope and fan regions. However, relatively few to no oscillations in pressure where found to occur over the course of diffusion back to steady-state in the respective PFLOW models. This predicted diffusion pattern is in contrast with to the observed sporadic recovery patterns. In addition, only 16 of the 41 wells recovered back to a pre-existing state, and oddly, one of the wells continued to show decreases in water-level over a 100-day span. If coseismic volumetric strain and elastic deformation was the only mechanism controlling the hydrologic response of these wells, then there should be a relatively smooth, continuous recovery of hydraulic head through pore-pressure diffusion with time. Results from this study essentially point to two likely mechanisms for hydrologic changes observed in the alluvial fan: fracture-induced permeability enhancement and coseismic compaction. Fracture-induced permeability could significantly alter the dissipation patterns. Strong ground motion generated by the earthquake led to wide-scale fracturing within the Western Foothills (Lin et al., 2004). In this region, the passing of 131 seismic waves could have also cleared sediments from the pre-existing fractures and cracks. Both of these factors could create new conduits or close others for fluid flow and thus, could explain why some of the wells continued to show decreases in hydraulic head after the seismic waves had passed, while others showed a sporadic response with time. Rojstaczer and Wolf (1994) suggested that fracturing of perched aquifers associated with the Loma Prieta earthquake could have contributed to increases in stream discharge at lower elevations. Fluids within perched aquifers located in the upper slope and Western Foothills could have been breached downslope, as confining beds located beneath the aquifer became fractured. Since groundwater located at higher elevations also has a higher potential (hydraulic head) than the surrounding regions, water would flow in the direction of the induced gradient (downhill). This coseismic and post-seismic release of water from higher elevations to zones down-gradient could explain the delayed discharges at lower elevations. Similarly, the flow of the fluids down gradient could also be used to explain abrupt coseismic decreases in the discharge of the Choshui River near the fault zone. Prior discussion shows that a coseismic compaction mechanism is most likely the dominant factor for invoking wide-spread increases in water-level within the Choshui River alluvial fan, but can this model be used to explain the observed dissipation patterns? The consequent coseismic re-packing of sediments alters the permeability of the sediments and results in changed hydraulic conductivity. This could serve to significantly alter the diffusion rates. Instead of seeing a smooth and continuous diffusion back to a steady-state, changes in permeability and conductivity could also alter pre-existing flow paths or lead to new migratory pathways altogether. Thus dissipation would occur, but 132 would be dictated by the new hydraulic properties of the fan. The observed recovery rate for well JL was modeled against the theoretical (calculated) recovery rates using different values of hydraulic conductivity (Fig. 55). Modeling results for this well show that recovery took place much slower than with its pre-seismic hydraulic conductivity value, providing strong evidence of decreased conductivity caused by coseismic compaction. Furthermore, coseismic compaction may also contribute to the observed stream discharge values in the alluvial fan (Baird, 2002; Manga et al., 3003). The expulsion of pore-fluids from the upper aquifers could increase the amount of base-flow to the streams and explain observed increases (Fig. 12). 133 K= 6.91x10-1 m/d Actual Recovery K= 1.04x10-1 m/d K= 8.64 m/d (Pre- earthquake) T i m e ( d a y s ) W a t e r - L e v e l C h a n g e ( m ) 0 2 0 4 0 6 0 8 0 1 0 0 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5 Figure 55. Observed water-table dissipation for well JL. The observed recovery (dashed- line) does not match the recovery modeled with pre-seismic hydraulic conductivity values. This suggest that seismic waves decreased the hydraulic conductivity value of the alluvial fan sediments in this region from 8.64 m/d to between 6.9 x 10-1 m/d and 1.04 x 10-1 m/d. Black dashed-line indicates observed recovery. Blue line indicates recovery with pre-seismic hydraulic conductivity. Green and red lines are modeled recoveries. 134 Importance of Hydraulic Conductivity The importance of hydraulic conductivity as a dominant factor for controlling diffusion patterns was evident from the pore-pressure models. In general, rocks and sediments with intrinsically higher hydraulic conductivity values were found to accommodate the diffusion of excess pressure with much greater ease. Formations that are comprised mainly of gravels, coarse-sands, or coarse-grained sandstones will allow for faster diffusion rates when compared to those dominated by silt, mud, or similar lithologies. Similarly, rocks and sediments with intrinsically low hydraulic conductivities (i.e., shale, crystalline bedrock) will maintain pore-pressure changes longer, because pressure does not propagate quickly into adjacent formations. Results from this study show that by decreasing the magnitude of hydraulic conductivity (K) by one order of magnitude, the corresponding time to diffuse excess pressure on the order of 1 x 106 Pa by 95% increases from 30 days to almost 200 days. Such strong a correlation between diffusion time and hydraulic conductivity highlights the type of geologic media (i.e., geologic layering) as the dominant variable for determining diffusion rates following a large magnitude earthquake. 135 Summary and Conclusions This study integrates numerical modeling with the analysis of hydrogeologic field data to explore the hydrodynamic response of the Choshui River alluvial fan and upper slope region to crustal deformation generated by the 1999 Chi-Chi earthquake. Piezometric data were gathered from the Water Resources Bureau (WRB) of Taiwan to study the observed co- and post-seismic response of aquifers in the alluvial plain. Strain modeling was integrated with a pore-pressure diffusion model to create a one-way coupled model that investigates coseismic volumetric strain (Muir-Wood and King, 1993) as a tenable mechanism for inducing the observed hydrologic anomalies and subsequent diffusion. Strong declines in water-level were found to occur within the calculated dilatational zone near the southern edge of the Chelungpu fault, specifically located within the upper slope region of the Western Foothills. These drops were also modeled to have occurred within a corresponding zone of decreased pore-pressure and thus, the coseismic hydrologic response can be explained by coseismic volumetric strain. However, widespread increases in water-table were found to have occurred within the Choshui River alluvial fan. These increases occurred in zones of dilatational strain and decreased pore-pressure. Therefore, coseismic volumetric strain cannot explain the hydrodynamic response of the Choshui River alluvial fan. A coseismic compaction 136 mechanism generated by strong seismic-shaking, provides the best explanation for the hydrologic response of the wells in the alluvial fan. The post-seismic recovery of the 41 wells analyzed generally displayed a sporadic and chaotic recovery pattern, suggesting post-seismic changes in the sediment or rock properties. Wells showing coseismic declines were found to occur within 15 km of the Chelungpu fault can be explained by either coseismic strain or fracture-induced permeability. However, the post-seismic recovery of the proximal region did not recover smoothly, as would be the case if coseismic strain and elastic deformation was the only controlling factor. This favors fracture induced permeability enhancement (Rojstaczer and Wolf, 1994) as the dominant mechanism for controlling the post-seismic hydrodynamic response in this region. Fracture induced permeability enhancement may also explain the coseismic decrease of stream discharge for streams located in the Western Foothills and the increases in stream discharge down-slope in the Choshui River alluvial fan. At best it can be surmised that a combination of coseismic volumetric strain and fracture induced permeability could be used to explain the coseismic response of fluids contained within the upper slope near the fault rupture. Coseismic compaction may explain the chaotic post-seismic response of fluids contained within the Choshui River alluvial fan. Compaction leading to a reduction in primary porosity could alter the hydraulic conductivities (K) of the aquifers. Modeling suggests that in some areas, conductivity may have been reduced by the effects of the earthquake and thus, supports a coseismic compaction mechanism. The resultant fluid expulsion generated by compaction could also be used to explain the coseismic increases of discharge within streams of the alluvial fan. 137 Pore-pressure modeling did however highlight the importance of hydraulic conductivity as the dominant factor contributing to the speed of diffusion of induced excess pore-pressure. Layers with low conductivity values maintained induced excess pressure longer and delayed re-equilibration of the pressure field. Contrastingly, higher conductivity accommodates groundwater flow and pressure dissipation; allowing faster recovery. 138 References Baird, P., Hydrologic changes with the 1999 Chi-Chi earthquake, Taiwan: A Numerical study. M.S. Thesis. Dept of Geol. and Geog., Auburn University, Auburn, 2002. Biot, M. (1941), General Theory of Three-Dimensional Consolidation, J. of Applied Physics, 12(2), 155-164 Bosl, W.J., and Nur, A. (2000), Crustal Fluids and Earthquakes, in GeoComplexity and the Physics of Earthquakes, Geophysical monograph 120, American Geophysical Union, 267-284. Briggs, R.O., 1991, Effects of Loma Prieta earthquake on surface waters in Waddell Valley: Water Resources Bulletin, 27, 991-999 Central Geological Survey (2000), Survey report of earthquake geology in the 9/21 Chi- Chi earthquake: Taiwan, Central Geological Survey, 315 (in Chinese). Chai, B.H.T. (1972), Structure and tectonic evolution of Taiwan. American Journal of Science, 272, 389-422 Chen, C.-H., Wang, W.-H., Teng, T-1. (2001), 3D velocity structure around the source area of the 1999 Chi-Chi, Taiwan, Earthquake: Before and after the main shock, Bull. Seis. Soc. Am., 91(5), 1013-1027. Chen, Y.-G., Chen, W.-S., Less, J.-C., Lee, Y.-H., Lee, C.-T., Chang,H.-C., and Lo, C.- H. (2001), Surface rupture of the 1999 Chi-Chi earthquake yields insight on active tectonics of central Taiwan, Bull. Seis. Soc. Am., 91(5), 977-985. Chi, W.-C., and Dreger, D., 2004, Crustal deformation in Taiwan: Results from finite source inversions of six Mw > 5.8 Chi-Chi aftershocks, J. Geophys. Res., 109, 1- 19 Chia, Y., Wang, Y.-S., Chui, J.J., and Liu, C.-W., 2001, Changes of groundwater level due to the 1999 Chi- Chi earthquake in the Choshui River Alluvial fan in Taiwan, Bull. Seis. Soc. Am., 91(5), 1062-1068. 139 Chia, Y., J. J. Chiu, Y. H. Jiang, T. P. Lee, Y. M. Wu, and M. J. Horng (2008), Implications of coseismic groundwater level changes observed at multiple-well monitoring stations, Geophys. J. Int., 172, 293? 301. Chiang, C.-S., Yu, H.-S. and Chou, Y.-W. (2004), Characteristics of the wedge-top depozone of the southern Taiwan foreland basin system. Basin Research, 16, 65? 78. doi: 10.1111/j.1365-2117.2004.00222.x Chou , J. T. (1973), Sedimentology and paleogeography of the upper Cenozoic System of western Taiwan: Geol. Soc. China Proc.,16, 111-143 Covey, M. (1986), The evolution of foreland basins to steady state: evidence from the western Taiwan foreland basin, In: Allen, P. A. and Homewood, P. (Eds), Foreland Basins, International Association of Sedimentologists, Special Publication 8, 77-90 Healy, J., H., Rubey, W., W., Griggs, D., T., and Raleigh, C., B. (1968), the Denver Earthquakes, Science, 161, 1301-1310. Ho, C.-S. (1975), Taiwan-di-zhi-gai-lun, Ministry of Economic Affairs, R.O.C., p. 118. Ho, C.-S. (1986), Taiwan-di-zhi-gai-lun, Central Geological Survey, p. 164 Hsu, S. K., C. V. Chin, and L. H. Cheng (2000), The result of first stage of ?Groundwater Monitoring Network Plan in Taiwan?. Proceedings of water resources management in 21st century, 177?197. Huang, H.-C. and Tseng, Y.-S. (2002), Characteristics of soul liquefaction using H/V of Microtremors in Yuan-Lin area, Taiwan, TAO, 13(3), 325-338. Haung. C.-Y., Wu, W.-Y., Chang, C.-P., Tsa, S., Yuan, P.-B., Lin, C.-W., Xia, K.-Y. (1997), Tectonic evolution of accretionary prism in the arc-continent collision terrane of Taiwan, Tectonophysics, 281, 31-51. Katsumata, M. and Sykes, L.R. (1969), Seismicity and tectonics of the Western Pacific: Izu-Mariana-Caroline and Ryukyu-Taiwan regions. Journal of Geophysical Research, 74, 5923-5948 Lai, W.-C., Koizumi, N., Matsumoto, N., Kitagawa, Y., Lin, C.-W., Shieh, C.-L., and Lee, Y.-P. (2004), Effects of seismic ground motion and geological setting on the coseismic groundwater level changes caused by the 1999 Chi-Chi earthquake, Taiwan. Earth Planets Space, 56, 873-880 Lee, M., Liu, T.-K., Mai, K.-F., and Chang, Y.-M. (2002), Coseismic hydrological changes associate with dislocation of the September 21, 1999 ChiChi earthquake, Taiwan, Geophysical Research Letter, 29(17), 1824. 140 Lee, M.-K., Wang, C., Kao, H., and Wolf, L., W. (2001), Numerical analysis of coseismic and postseismic changes of 1999 Chi-Chi Earthquake, Taiwan, AGU Annual Meeting, San Francisco. Lee, M.-K. and Wolf, L., W. (1998), Analysis of fluid pressure propagation in heterogeneous rocks: Implications of hydrologically-induced earthquakes, Geophysical Research Letters, 25(13), 2329-2332. Lin. A., Ouchi, T., Chen, A., and Maruyama, T. (2001), Nature of the fault jog inferred from a deformed well in the northern Chelungpu surface rupture zone, related to the 1999 Chi-Chi, Taiwan, ML 7.3 earthquake, Bull. Seis. Soc. Am., 91(5), 959- 965. Lin, A.T., A.B. Watts, and S.P. Hesselbo (2003), Cenozoic stratigraphy and subsidence history of the South China Sea margin in the Taiwan region., Basin Research, 15, 453-478. Lin, J. and R.S. Stein (2004), Stress triggering in thrust and subduction earthquakes, and stress interaction between the southern San Andreas and nearby thrust and strike-slip faults, J. Geophys. Res., 109, B02303, doi:10.1029/2003JB002607. Liu, K.-S., Shin, T.-C., and Tsai, Y.-B. (1999), A free-field strong motion network in Taiwan: TSMIP, TAO, 10(2), 377-396. Lo, C.-H., and 24 other Co-authors (1999), A comprehensive summary of the hazards caused by the 9/21 Chi-Chi earthquake: Taipei, Taiwan National Center for Research on Earthquake Engineering, NRCEE ? 99 ? 033 (In Chinese). Ma, K.-F., Lee, C.-T., and Tsai,Y.-B. (1999), The Chi-Chi, Taiwan earthquake: largest surface displacements on an inland thrust fault, EOS Transactions, American Geophysical Union, 80, 605-606. Ma, K.-F., C.-H. Chan, and R. S. Stein (2005), Response of seismicity to Coulomb stress triggers and shadows of the 1999 Mw = 7.6 Chi-Chi, Taiwan, earthquake, Journal of Geophysical Research, 110, B05S19, doi:10.1029/2004JB003389. Manga, M. (2001), Origin of postseismic streamflow changes inferred from baseflow Recession and magnitude-distance relation: Geophysical Research Letters, 28, 2133-2136. Manga, M., Brodsky, E.E., and Boone, M., 2003, Response of streamflow to multiple Earthquakes and implications for the origin of postseismic dischanges: Geophysical Research Letters, 30(5), 1214, doi: 10.1029/2002Gl016618 141 Montgomery, D. and Manga, M. (2003), Stream flow and Water Well Responses to Earthquakes, Science, 300, 2047-2049 Muir- Wood, R., and King, G.C.P. (1993), Hydrological signature of earthquake strain, J. Geophys. Res., 98, 22035-22068. Nur, A., and Booker, J.R. (1972), Aftershocks Caused by Pore Fluid Flow? Science, 175 (4024), 885-887 Page, B. and Suppe, J. (1981), The Pliocene Lichi Melange of Taiwan: its olistrostromal and plate-tectonic origin. Amer. Jour. Sci., 281, 193-227. Roeloffs, E. (1988), Hydrologic precursors to earthquakes: A review, Pure Appl. Geophys., 126, 177?209. Okada, Y. (1992), Internal deformation due to shear and tensile faults in half-space, Bull. Seis. Soc. Am., 28, 1018-1040. Rojstaczer, S. and Wolf, S. (1994), Hydrologic changes associated with the earthquakes in the San Lorenzo and Pescadero drainage basins, in the Loma Prieta, California, Earthquake of October 17, 1989- Hydrologic disturbance, Rojstaczer S. (ed.), U.S. Geological Survey Prof. Paper, 1551-E, E51-E64. Seed, H.B., and Lee, K.L. (1966), Liquefaction of Saturated Sands during cyclic loading: Journal of Soil and Mechanic Foundation Division, American Society of Civil Engineers, 92, 105-134. Simpson, D.W. (1986), Triggered Earthquakes, Annual Review of Earth and Planetary Sciences, 14, 21-42. Shih, R.-C., Ho, S.-J., Chang, H.-C., and Lin, C.-W. (2000), Preliminary subsurface images of the Chelungpu fault by using shallow seismic reflection, Proceedings of International Workshop of Annual Commemoration of Chi-Chi Earthquake, September 18-20, 2000, Taipei, Loh, C.-H., Liao, W.-I., 123-132. Shin, T.-C. and Teng, T.-L. (2001), An Overview of the 1999 Chi-Chi, Taiwan, earthquake, Bull. Seis. Soc. Am., 91(5), 895-913. Sibson, H.H. (1994), Crustal stress, faulting, and fluid flow, Geofluids: Origin, Migration, and Evolution of Fluids in Sedimentary Basins, Parnell, J. (ed.), Geophysical Society Special Publication, 78, 69-84. Skempton, A. W. (1954), The pore-pressure coefficients A and B, Geotechnique, 4, 143- 147. 142 Teng, L. S. (1987a), Stratigraphic records of the late Cenozoic Penglai Orogeny of Taiwan: Acta Geol. Taiwan., 25, 205-224. Teng, L.S. (1990), Geotectonic evolution of late Cenozoic arc-continent collision in Taiwan, Tectonophysics, 183, 57-79 Toda, S., R. S. Stein, K. Richards-Dinger and S. Bozkurt (2005), Forecasting the evolution of seismicity in southern California: Animations built on earthquake stress transfer, J. Geophys. Res., B05S16, doi:10.1029/2004JB003415. earthquakes, Wang, H.,F. and Anderson, M., P. (1982), Introduction to groundwater modeling: Finite difference and Finite element methods, W.H. Freeman and Company, New York. Wang, C,-Y., Li, C.-L., Su, F.-C., Leu, M.-T., Wu, M.-S., Lai, S.-H., and Chern, C.-C. (2002), Structural mapping of the 1999 Chi-Chi earthquake fault, Taiwan by seismic reflection methods, Tao, 3, 211-226 Wang, C.-H., Peng, T.-R., Chen, W.-F., Lee, M.-K., and Chiang, C.-J. (2001), Hydrogen and oxygen isotopic variations of Choshuichi ground water before and after the Cho-Cho earthquake, Proceeding of the Third Symposium on Ground water Resources and water Protection, Pintug Technology University, Pintug, Taiwan Wang, C.-Y., Cheng, L.-H., Chin, C.-V., and Yu, S.-B. (2001), Coseismic hydrologic response of an alluvial fan to the 1999 Chi-Chi earthquake, Taiwan, Geology, 29, 831-834. Wang, C.-Y., Dreger, D.S., Wang, C.-H., Mayeri, G., and Berryman, J.G. (2003), Field Relations among coseismic ground motion, water level change and liquefaction for the 1999 Chi-Chi (Mw=7.5) earthquake, Taiwan, Geophysical Research Letters, 30. Wang, C.-Y., Wang, C.-H, Manga, M., 2004, Coseismic release of water from the mountains: Evidence from the 1999 (Mw = 7.5) Chi-Chi, Taiwan, earthquake, Geological Society of America, 32. Wong, A. and Wang, C.-Y., 2007, Field relations between the spectral composition of Ground motion and hydrological effects during the 1999 Chi-Chi (Taiwan)Earthquake, J. of Geophys. Res., 112, B10305, doi: 10.1029/2006JB004516. Water Resources Bureau, 2000, Survey report of surface water and groundwater level fluctuations during the 1999 Chi-Chi earthquake, Taiwan, Water Resource Bureau, 37. ( in Chinese). 143 Wolf, L.W., Rowe, C.A., and Horner, R.B., 1997 Periodic Seismicity near Mt. Ogden on the Alaska-British Columbia border: A case for Hydrologically-triggered earthquakes? Bull. Seis. Soc. Am, 87, 1473-1483. Yu, S.-B., and 12 other Co-Authors, 2001, preseismic deformation and coseismic displacements associated with the Chi-Chi, Taiwan, Earthquake, Bull. Seis. Soc. Am., 91(5), 995-1012. 144 Appendix A: Proximal Well Time-Series Plots A 100 day time-series was plotted for each well within the proximal zone (< 15 km) using daily averages. The dominant coseismic response was down, however here more interest is expressed in the post-seismic response. Generally, the wells show a sporadic post-seismic response. Outside influences such as local pumping or aftershocks may play major roles in the variations observed within these diagrams. Clearly, simple pore-pressure diffusion modeling cannot explain these patterns (as all curves would show a steady and continuous curve back to pre-seismic levels and maintain this level with some regularity). The timing of the Chi-Chi earthquake is marked by the black arrow. Possible responses caused by aftershocks are marked by red arrows. 145 146 147 148 Appendix B: Middle-Fan Time-Series A 100 day time-series was plotted for each well within the middle-fan zone (15- 35 km) using daily averages. The dominant coseismic response was up, however here more interest is expressed in the post-seismic response. Generally, the wells show a sporadic post-seismic response. Outside influences such as local pumping or aftershocks may play major roles in the variations observed within these diagrams. Clearly, simple pore-pressure diffusion modeling cannot explain these patterns (as all curves would show a steady and continuous curve back to pre-seismic levels and maintain this level with some regularity). The timing of the Chi-Chi earthquake is marked by the black arrow. Possible responses caused by aftershocks are marked by red arrows. 149 150 151 152 153 154 155 156 157 158 Appendix C: Distal Fan Time-Series A 100 day time-series was plotted for each well within the distal zone (> 35 km) using daily averages. The dominant coseismic response was down, however here more interest is expressed in the post-seismic response. Generally, the wells show a sporadic post-seismic response. Outside influences such as local pumping or aftershocks may play major roles in the variations observed within these diagrams. Clearly, simple pore- pressure diffusion modeling cannot explain these patterns (as all curves would show a steady and continuous curve back to pre-seismic levels and maintain this level with some regularity). The timing of the Chi-Chi earthquake is marked by the black arrow. Possible responses caused by aftershocks are marked by red arrows. 159 160 161 162 163 164 165 166 167 168 Appendix D: Strain Input File (For use: copy into text file and save within Coulomb directory as .inp file) Example for stress changes on specified faults and splitting fault into some elements. This example is very simple model for Kobe earthquake like. #reg1= 0 #reg2= 0 #fixed= 261 sym= 1 PR1= .250 PR2= .250 DEPTH= 1.0 E1= 0.800000E+06 E2= 0.800000E+06 XSYM= .000 YSYM= .000 FRIC= .400 S1DR= 122.0001 S1DP= 0.0001 S1IN= 100.000 S1GD= .000000 S2DR= 122.0001 S3DP= 90.0001 S3IN= 30.000 S3GD= .000000 S3DR= 32.0001 S2DP= 0.0001 S2IN= 0.000 S2GD= .000000 # X-start Y-start X-fin Y-fin Kode shear normal dip angle top bot xxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx 1 9.2324 13.9867 9.4364 17.6398 100 1.3413 1.8644 29.0000 0.0000 1.4780 1 1 9.4363 17.6399 9.6403 21.2930 100 0.2155 1.3846 29.0000 0.0000 1.4780 2 1 9.6402 21.2929 9.8442 24.9460 100 0.1563 1.0117 29.0000 0.0000 1.4780 3 1 9.8449 24.9461 10.0489 28.5992 100 -1.5493 2.2211 29.0000 0.0000 1.4780 4 1 9.9468 28.5991 10.1508 32.2522 100 2.4107 3.4542 29.0000 0.0000 1.4780 5 1 10.1507 32.3629 10.3547 36.0160 100 -1.2833 5.7722 29.0000 0.0000 1.4780 6 1 10.3546 36.0159 10.5586 39.6690 100 0.0722 4.3565 29.0000 0.0000 1.4780 7 1 10.5585 39.6691 10.7625 43.3222 100 -2.4080 7.9490 29.0000 0.0000 1.4780 8 1 10.7624 43.3221 10.9664 46.9752 100 -6.2396 7.0745 29.0000 0.0000 1.4780 9 1 10.9662 47.0859 11.1702 50.7390 100 0.0076 4.8641 29.0000 0.0000 1.4780 10 1 11.1701 50.7392 11.3741 54.3923 100 -3.0204 7.7489 29.0000 0.0000 1.4780 11 1 11.3748 54.3921 11.5788 58.0452 100 -3.8384 3.7829 29.0000 0.0000 1.4780 12 1 11.5787 58.0453 11.7827 61.6984 100 -6.7943 6.7744 29.0000 0.0000 1.4780 13 1 11.7826 61.8092 11.9866 65.4623 100 -4.3821 8.6429 29.0000 0.0000 1.4780 14 1 11.9865 65.4622 12.1905 69.1153 100 -2.7525 8.8808 29.0000 0.0000 1.4780 15 1 12.0884 69.1154 12.2924 72.7685 100 -9.1105 5.0744 29.0000 0.0000 1.4780 16 169 1 12.2923 72.7684 12.4963 76.4215 100 -5.6919 2.5499 29.0000 0.0000 1.4780 17 1 12.4962 76.5322 12.7002 80.1853 100 -9.5533 7.7450 29.0000 0.0000 1.4780 18 1 12.7008 80.1852 12.9048 83.8383 100 -10.4357 5.9100 29.0000 0.0000 1.4780 19 1 12.9047 83.8384 13.1087 87.4915 100 1.4356 5.7954 29.0000 0.0000 1.4780 20 1 12.4962 13.7654 12.7002 17.4185 100 2.3133 2.8858 29.0000 1.9000 3.3780 21 1 12.7008 17.4184 12.9048 21.0715 100 0.8395 2.9583 29.0000 1.9000 3.3780 22 1 12.9047 21.0716 13.1087 24.7247 100 -1.3508 2.2288 29.0000 1.9000 3.3780 23 1 13.1086 24.8352 13.3126 28.4883 100 -4.1872 1.6450 29.0000 1.9000 3.3780 24 1 13.3125 28.4884 13.5165 32.1415 100 0.9943 1.3276 29.0000 1.9000 3.3780 25 1 13.5164 32.1414 13.7204 35.7945 100 -1.2645 5.9615 29.0000 1.9000 3.3780 26 1 13.7203 35.7946 13.9243 39.4477 100 -3.4834 4.5251 29.0000 1.9000 3.3780 27 1 13.9242 39.5585 14.1282 43.2116 100 -2.5622 3.8976 29.0000 1.9000 3.3780 28 1 14.0261 43.2115 14.2301 46.8646 100 -4.4989 3.6352 29.0000 1.9000 3.3780 29 1 14.2308 46.8647 14.4348 50.5178 100 -2.1145 5.0992 29.0000 1.9000 3.3780 30 1 14.4347 50.5177 14.6387 54.1708 100 -8.1787 7.4813 29.0000 1.9000 3.3780 31 1 14.6386 54.2815 14.8426 57.9346 100 -4.8359 5.4503 29.0000 1.9000 3.3780 32 1 14.8424 57.9345 15.0464 61.5876 100 -8.1645 7.2145 29.0000 1.9000 3.3780 33 1 15.0463 61.5877 15.2503 65.2408 100 -6.0305 6.1019 29.0000 1.9000 3.3780 34 1 15.2502 65.2407 15.4542 68.8938 100 -5.0800 7.6646 29.0000 1.9000 3.3780 35 1 15.4549 69.0045 15.6589 72.6576 100 -5.7988 8.9853 29.0000 1.9000 3.3780 36 1 15.6588 72.6577 15.8628 76.3108 100 -3.8806 4.0930 29.0000 1.9000 3.3780 37 1 15.8627 76.3107 16.0667 79.9638 100 -13.4257 7.7167 29.0000 1.9000 3.3780 38 1 15.9646 79.9639 16.1686 83.6170 100 -14.2425 4.8546 29.0000 1.9000 3.3780 39 1 15.8627 13.5439 16.0667 17.1970 100 0.8195 1.0496 29.0000 3.8000 5.2780 41 1 15.9646 17.3077 16.1686 20.9608 100 0.0719 0.2005 29.0000 3.8000 5.2780 42 1 16.1685 20.9607 16.3725 24.6138 100 -0.0731 0.7725 29.0000 3.8000 5.2780 43 1 16.3724 24.6139 16.5764 28.2670 100 -1.0523 2.1639 29.0000 3.8000 5.2780 44 1 16.5763 28.2669 16.7803 31.9200 100 -1.9108 2.2028 29.0000 3.8000 5.2780 45 170 1 16.7802 32.0308 16.9842 35.6839 100 -1.9886 7.3496 29.0000 3.8000 5.2780 46 1 16.9848 35.6840 17.1888 39.3371 100 -3.7391 6.1171 29.0000 3.8000 5.2780 47 1 17.1887 39.3370 17.3927 42.9901 100 -0.0762 2.6781 29.0000 3.8000 5.2780 48 1 17.3926 42.9902 17.5966 46.6433 100 -2.0022 4.9448 29.0000 3.8000 5.2780 49 1 17.5965 46.6432 17.8005 50.2963 100 -0.3480 2.7397 29.0000 3.8000 5.2780 50 1 17.8004 50.4070 18.0044 54.0601 100 -4.8561 8.1461 29.0000 3.8000 5.2780 51 1 17.9023 54.0600 18.1063 57.7131 100 -4.3371 8.4900 29.0000 3.8000 5.2780 52 1 18.1062 57.7132 18.3102 61.3663 100 -6.4030 6.3129 29.0000 3.8000 5.2780 53 1 18.3109 61.4771 18.5149 65.1302 100 -3.5777 2.5984 29.0000 3.8000 5.2780 54 1 18.5148 65.1301 18.7188 68.7832 100 -7.1563 5.9950 29.0000 3.8000 5.2780 55 1 18.7186 68.7833 18.9226 72.4364 100 -6.9885 7.4263 29.0000 3.8000 5.2780 56 1 18.9225 72.4363 19.1265 76.0894 100 -2.3128 2.9629 29.0000 3.8000 5.2780 57 1 19.1264 76.2001 19.3304 79.8532 100 -7.6631 4.8123 29.0000 3.8000 5.2780 58 1 19.3303 79.8533 19.5343 83.5064 100 -7.1948 3.1282 29.0000 3.8000 5.2780 59 1 19.5342 83.5063 19.7382 87.1594 100 0.0000 0.0000 29.0000 3.8000 5.2780 60 1 19.1264 13.4333 19.3304 17.0864 100 -0.1864 0.1745 29.0000 5.7000 7.1780 61 1 19.3303 17.0863 19.5343 20.7394 100 -0.7790 2.0168 29.0000 5.7000 7.1780 62 1 19.5342 20.7395 19.7382 24.3926 100 1.7535 2.2674 29.0000 5.7000 7.1780 63 1 19.7389 24.3925 19.9429 28.0456 100 1.2065 1.7967 29.0000 5.7000 7.1780 64 1 19.8408 28.1563 20.0448 31.8094 100 -1.8346 1.7149 29.0000 5.7000 7.1780 65 1 20.0447 31.8095 20.2487 35.4626 100 -0.1392 6.1150 29.0000 5.7000 7.1780 66 1 20.2486 35.4625 20.4526 39.1156 100 -2.9538 4.2606 29.0000 5.7000 7.1780 67 1 20.4525 39.2263 20.6565 42.8794 100 -0.3162 1.3221 29.0000 5.7000 7.1780 68 1 20.6564 42.8793 20.8604 46.5324 100 -1.2044 2.0538 29.0000 5.7000 7.1780 69 1 20.8603 46.5325 21.0643 50.1856 100 0.3843 1.0348 29.0000 5.7000 7.1780 70 1 21.0641 50.1855 21.2681 53.8386 100 -3.9704 0.0721 29.0000 5.7000 7.1780 71 1 21.2688 53.9494 21.4728 57.6025 100 -3.9603 5.2540 29.0000 5.7000 7.1780 72 1 21.4727 57.6026 21.6767 61.2557 100 -6.1404 7.5931 29.0000 5.7000 7.1780 73 171 1 21.6766 61.2556 21.8806 64.9087 100 -3.8671 2.7360 29.0000 5.7000 7.1780 74 1 21.8805 64.9088 22.0845 68.5619 100 -5.4887 5.6068 29.0000 5.7000 7.1780 75 1 21.9824 68.5618 22.1864 72.2149 100 -7.7812 7.4434 29.0000 5.7000 7.1780 76 1 22.1863 72.3256 22.3903 75.9787 100 -3.7464 5.7804 29.0000 5.7000 7.1780 77 1 22.3902 75.9786 22.5942 79.6317 100 -2.3879 2.5455 29.0000 5.7000 7.1780 78 1 22.3902 13.2118 22.5942 16.8649 100 -0.8181 0.4993 29.0000 7.6000 9.0780 81 1 22.5949 16.8650 22.7989 20.5181 100 -0.4261 2.8136 29.0000 7.6000 9.0780 82 1 22.7987 20.6286 23.0027 24.2817 100 1.9106 2.5133 29.0000 7.6000 9.0780 83 1 23.0026 24.2818 23.2066 27.9349 100 1.7419 2.4929 29.0000 7.6000 9.0780 84 1 23.2065 27.9348 23.4105 31.5879 100 -1.1005 1.1636 29.0000 7.6000 9.0780 85 1 23.4104 31.5880 23.6144 35.2411 100 0.8354 1.1561 29.0000 7.6000 9.0780 86 1 23.6143 35.3519 23.8183 39.0050 100 -0.9538 0.6436 29.0000 7.6000 9.0780 87 1 23.8182 39.0049 24.0222 42.6580 100 0.1810 0.5646 29.0000 7.6000 9.0780 88 1 23.9201 42.6581 24.1241 46.3112 100 -1.8690 0.9655 29.0000 7.6000 9.0780 89 1 24.1248 46.3111 24.3288 49.9641 100 -0.6225 1.0974 29.0000 7.6000 9.0780 90 1 24.3287 50.0749 24.5327 53.7280 100 -0.6953 1.1402 29.0000 7.6000 9.0780 91 1 24.5326 53.7281 24.7366 57.3812 100 0.4077 7.6938 29.0000 7.6000 9.0780 92 1 24.7365 57.3811 24.9405 61.0342 100 0.4046 11.7388 29.0000 7.6000 9.0780 93 1 24.9403 61.0343 25.1443 64.6874 100 -0.1907 6.8862 29.0000 7.6000 9.0780 94 1 25.1442 64.7979 25.3482 68.4510 100 -4.8420 4.4077 29.0000 7.6000 9.0780 95 1 25.3481 68.4511 25.5521 72.1042 100 -8.4452 5.7905 29.0000 7.6000 9.0780 96 1 25.5528 72.1041 25.7568 75.7572 100 -2.3707 3.6889 29.0000 7.6000 9.0780 97 1 25.7567 13.1011 25.9607 16.7542 100 -0.0707 0.1438 29.0000 9.5000 10.9780 101 1 25.8586 16.7541 26.0626 20.4072 100 -1.2899 1.7043 29.0000 9.5000 10.9780 102 1 26.0625 20.4073 26.2665 24.0604 100 0.8894 1.9634 29.0000 9.5000 10.9780 103 1 26.2664 24.0603 26.4704 27.7134 100 0.7986 1.6212 29.0000 9.5000 10.9780 104 1 26.4703 27.8242 26.6743 31.4773 100 0.2934 0.5488 29.0000 9.5000 10.9780 105 1 26.6742 31.4774 26.8782 35.1305 100 -0.0180 0.0037 29.0000 9.5000 10.9780 106 172 1 26.8788 35.1304 27.0828 38.7835 100 0.2065 0.4977 29.0000 9.5000 10.9780 107 1 27.0827 38.7836 27.2867 42.4367 100 -0.0188 0.0445 29.0000 9.5000 10.9780 108 1 27.2866 42.5472 27.4906 46.2003 100 -0.6852 0.0856 29.0000 9.5000 10.9780 109 1 27.4905 46.2004 27.6945 49.8535 100 -2.3484 0.7970 29.0000 9.5000 10.9780 110 1 27.6944 49.8534 27.8984 53.5065 100 -0.2946 1.0616 29.0000 9.5000 10.9780 111 1 27.7963 53.5066 28.0003 57.1597 100 -0.6094 2.3330 29.0000 9.5000 10.9780 112 1 28.0002 57.2705 28.2042 60.9236 100 -0.8848 6.3717 29.0000 9.5000 10.9780 113 1 28.2049 60.9234 28.4089 64.5765 100 -2.5618 3.7693 29.0000 9.5000 10.9780 114 1 28.4088 64.5767 28.6128 68.2298 100 -5.0420 0.6951 29.0000 9.5000 10.9780 115 1 28.6127 68.2296 28.8167 71.8827 100 -7.5802 2.7648 29.0000 9.5000 10.9780 116 1 29.0204 12.8796 29.2244 16.5327 100 -0.0182 0.1196 29.0000 11.4000 12.8780 121 1 29.2243 16.5329 29.4283 20.1859 100 -0.5781 0.0581 29.0000 11.4000 12.8780 122 1 29.4282 20.2967 29.6322 23.9498 100 0.4607 0.7399 29.0000 11.4000 12.8780 123 1 29.6329 23.9497 29.8369 27.6028 100 -0.5637 0.3756 29.0000 11.4000 12.8780 124 1 29.7348 27.6029 29.9388 31.2560 100 0.0893 0.2763 29.0000 11.4000 12.8780 125 1 29.9387 31.2559 30.1427 34.9090 100 0.0411 0.1338 29.0000 11.4000 12.8780 126 1 30.1426 35.0197 30.3466 38.6728 100 0.2027 0.2985 29.0000 11.4000 12.8780 127 1 30.3465 38.6727 30.5505 42.3258 100 -0.2493 0.0861 29.0000 11.4000 12.8780 128 1 30.5504 42.3259 30.7544 45.9790 100 0.0208 0.0274 29.0000 11.4000 12.8780 129 1 30.7543 45.9789 30.9583 49.6320 100 -0.6949 0.5215 29.0000 11.4000 12.8780 130 1 30.9582 49.7428 31.1622 53.3959 100 -0.1658 0.0269 29.0000 11.4000 12.8780 131 1 31.1628 53.3960 31.3668 57.0491 100 -0.3909 0.1531 29.0000 11.4000 12.8780 132 1 31.3667 57.0490 31.5707 60.7021 100 -3.1725 1.0051 29.0000 11.4000 12.8780 133 1 31.5706 60.7022 31.7746 64.3553 100 -7.0166 2.4241 29.0000 11.4000 12.8780 134 1 31.7745 64.4658 31.9785 68.1189 100 -6.0071 2.7163 29.0000 11.4000 12.8780 135 1 31.8764 68.1190 32.0804 71.7721 100 -5.3549 3.3256 29.0000 11.4000 12.8780 136 1 32.2842 12.7690 32.4882 16.4221 100 -0.2242 0.3228 29.0000 13.3000 14.7780 141 1 32.4889 16.4220 32.6929 20.0751 100 0.6854 0.8873 29.0000 13.3000 14.7780 142 173 1 32.6928 20.0752 32.8968 23.7283 100 1.5415 2.4744 29.0000 13.3000 14.7780 143 1 32.8966 23.7282 33.1006 27.3813 100 -1.2909 1.4906 29.0000 13.3000 14.7780 144 1 33.1005 27.3814 33.3045 31.0345 100 -0.2003 0.8281 29.0000 13.3000 14.7780 145 1 33.3044 31.1452 33.5084 34.7983 100 -0.0583 0.2112 29.0000 13.3000 14.7780 146 1 33.5083 34.7982 33.7123 38.4513 100 -1.0880 0.0030 29.0000 13.3000 14.7780 147 1 33.7122 38.4514 33.9162 42.1045 100 -1.7469 0.0781 29.0000 13.3000 14.7780 148 1 33.8141 42.1044 34.0181 45.7575 100 0.0135 0.0194 29.0000 13.3000 14.7780 149 1 34.0188 45.8683 34.2228 49.5214 100 0.7337 1.3146 29.0000 13.3000 14.7780 150 1 34.2227 49.5215 34.4267 53.1746 100 0.3842 0.8418 29.0000 13.3000 14.7780 151 1 34.4266 53.1745 34.6306 56.8276 100 0.5745 1.1867 29.0000 13.3000 14.7780 152 1 34.6305 56.9383 34.8345 60.5914 100 -1.3403 1.0754 29.0000 13.3000 14.7780 153 1 34.8344 60.5913 35.0384 64.2444 100 -4.5599 2.0394 29.0000 13.3000 14.7780 154 1 35.6507 12.5475 35.8547 16.2006 100 0.0049 0.0343 29.0000 15.2000 16.6780 161 1 35.7526 16.2007 35.9566 19.8538 100 0.5617 1.3183 29.0000 15.2000 16.6780 162 1 35.9565 19.9646 36.1605 23.6177 100 1.1132 1.6378 29.0000 15.2000 16.6780 163 1 36.1604 23.6176 36.3644 27.2707 100 -1.2532 1.4355 29.0000 15.2000 16.6780 164 1 36.3643 27.2708 36.5683 30.9239 100 -0.4535 0.3373 29.0000 15.2000 16.6780 165 1 36.5682 30.9238 36.7722 34.5769 100 -0.8759 0.9496 29.0000 15.2000 16.6780 166 1 36.7729 34.6876 36.9768 38.3407 100 -2.4920 1.4279 29.0000 15.2000 16.6780 167 1 36.9767 38.3408 37.1807 41.9939 100 -1.7855 2.9120 29.0000 15.2000 16.6780 168 1 37.1806 41.9938 37.3846 45.6469 100 1.1842 4.9724 29.0000 15.2000 16.6780 169 1 37.3845 45.6470 37.5885 49.3001 100 3.2175 4.1147 29.0000 15.2000 16.6780 170 1 37.5884 49.3000 37.7924 52.9531 100 2.2544 2.9747 29.0000 15.2000 16.6780 171 1 37.7923 53.0638 37.9963 56.7169 100 0.6771 2.4577 29.0000 15.2000 16.6780 172 1 37.8942 56.7168 38.0982 60.3699 100 -0.6265 1.7035 29.0000 15.2000 16.6780 173 1 38.0981 60.3700 38.3021 64.0231 100 -4.8504 0.2992 29.0000 15.2000 16.6780 174 1 13.0575 88.1833 16.7295 88.8475 100 5.8241 3.2205 29.0000 0.0000 1.4780 1 1 16.7298 88.8476 20.4018 89.5118 100 3.4808 10.4071 29.0000 0.0000 1.4780 2 174 1 20.4013 89.5118 24.0733 90.1760 100 14.9460 13.2669 29.0000 0.0000 1.4780 3 1 24.0736 90.0652 27.7456 90.7294 100 0.4583 10.7601 29.0000 0.0000 1.4780 4 1 27.7452 90.7295 31.4172 91.3937 100 0.0684 - 0.8480 29.0000 0.0000 1.4780 5 1 31.4175 91.3937 35.0895 92.0579 100 2.3190 - 10.3626 29.0000 0.0000 1.4780 6 1 35.0898 92.0578 38.7618 92.7220 100 1.4169 - 6.6687 29.0000 0.0000 1.4780 7 1 38.7613 92.6114 42.4333 93.2756 100 1.1699 - 0.2321 29.0000 0.0000 1.4780 8 1 42.4337 93.2757 46.1057 93.9399 100 -1.0634 3.2834 29.0000 0.0000 1.4780 9 1 17.2395 85.6373 20.9115 86.3015 100 1.1250 2.6583 29.0000 1.9000 3.3780 11 1 20.9118 86.1907 24.5838 86.8549 100 6.2128 5.8789 29.0000 1.9000 3.3780 12 1 24.5833 86.8550 28.2553 87.5192 100 -2.3763 7.4124 29.0000 1.9000 3.3780 13 1 28.2557 87.5193 31.9277 88.1835 100 -0.6227 2.4030 29.0000 1.9000 3.3780 14 1 31.9272 88.1833 35.5992 88.8475 100 3.0039 - 5.9571 29.0000 1.9000 3.3780 15 1 35.5995 88.8476 39.2715 89.5118 100 1.4206 - 2.7480 29.0000 1.9000 3.3780 16 1 39.2718 89.4012 42.9438 90.0654 100 -0.8335 2.4951 29.0000 1.9000 3.3780 17 1 42.9434 90.0652 46.6154 90.7294 100 -0.3629 1.1915 29.0000 1.9000 3.3780 18 1 17.8512 82.3163 21.5232 82.9805 100 0.8612 - 0.0678 29.0000 3.8000 5.2780 20 1 21.5235 82.9805 25.1955 83.6447 100 4.6433 3.3160 29.0000 3.8000 5.2780 21 1 25.1958 83.6448 28.8678 84.3090 100 0.8444 6.5789 29.0000 3.8000 5.2780 22 1 28.8673 84.3088 32.5393 84.9730 100 -1.8847 6.2606 29.0000 3.8000 5.2780 23 1 32.5396 84.9731 36.2116 85.6373 100 3.1054 2.2248 29.0000 3.8000 5.2780 24 1 36.2112 85.5265 39.8832 86.1907 100 2.1115 3.5110 29.0000 3.8000 5.2780 25 1 39.8835 86.1907 43.5555 86.8549 100 2.6453 3.7115 29.0000 3.8000 5.2780 26 1 43.5558 86.8550 47.2278 87.5192 100 0.8992 - 1.6320 29.0000 3.8000 5.2780 27 1 22.1351 79.7701 25.8071 80.4343 100 2.8938 0.4897 29.0000 5.7000 7.1780 30 1 25.8074 80.4343 29.4795 81.0985 100 1.8543 3.8615 29.0000 5.7000 7.1780 31 1 29.4798 81.0986 33.1518 81.7628 100 0.8610 6.0668 29.0000 5.7000 7.1780 32 1 33.1513 81.6520 36.8233 82.3162 100 2.1464 5.5105 29.0000 5.7000 7.1780 33 1 36.8236 82.3163 40.4956 82.9805 100 2.0459 5.4223 29.0000 5.7000 7.1780 34 175 1 40.4952 82.9805 44.1672 83.6447 100 3.2273 4.9382 29.0000 5.7000 7.1780 35 1 44.1675 83.6448 47.8395 84.3090 100 2.1154 0.3790 29.0000 5.7000 7.1780 36 1 26.3172 77.2241 29.9892 77.8883 100 1.6052 1.8468 29.0000 7.6000 9.0780 40 1 29.9895 77.7775 33.6615 78.4417 100 2.3305 6.0755 29.0000 7.6000 9.0780 41 1 33.6618 78.4418 37.3338 79.1060 100 2.2507 6.9517 29.0000 7.6000 9.0780 42 1 37.3333 79.1060 41.0053 79.7702 100 2.0305 5.6839 29.0000 7.6000 9.0780 43 1 41.0057 79.7701 44.6777 80.4343 100 1.4382 4.1089 29.0000 7.6000 9.0780 44 1 44.6772 80.4343 48.3492 81.0985 100 1.6736 - 0.3024 29.0000 7.6000 9.0780 45 1 30.6012 74.5673 34.2732 75.2315 100 0.8433 3.6015 29.0000 9.5000 10.9780 50 1 34.2735 75.2316 37.9455 75.8958 100 4.3688 5.9138 29.0000 9.5000 10.9780 51 1 37.9458 75.8956 41.6178 76.5598 100 3.1951 5.9409 29.0000 9.5000 10.9780 52 1 41.6173 76.5599 45.2893 77.2241 100 3.3909 3.3225 29.0000 9.5000 10.9780 53 1 45.2896 77.1135 48.9616 77.7777 100 3.8704 - 1.5355 29.0000 9.5000 10.9780 54 1 34.8851 72.0211 38.5571 72.6853 100 3.5601 2.8825 29.0000 11.4000 12.8780 60 1 38.5574 72.6854 42.2295 73.3496 100 5.9234 6.2636 29.0000 11.4000 12.8780 61 1 42.1278 73.3496 45.7998 74.0138 100 3.6884 6.0104 29.0000 11.4000 12.8780 62 1 45.7994 73.9030 49.4714 74.5672 100 4.0120 1.8142 29.0000 11.4000 12.8780 63 1 39.0672 69.4752 42.7392 70.1394 100 4.9147 3.8167 29.0000 13.3000 14.7780 70 1 42.7395 70.0286 46.4115 70.6928 100 2.7376 3.5571 29.0000 13.3000 14.7780 71 1 46.4118 70.6928 50.0838 71.3570 100 2.9124 2.9724 29.0000 13.3000 14.7780 72 1 39.6796 66.1541 43.3516 66.8183 100 4.1555 1.6523 29.0000 15.2000 16.6780 79 1 43.3512 66.8183 47.0232 67.4825 100 3.6841 1.3920 29.0000 15.2000 16.6780 80 1 47.0235 67.4824 50.6955 68.1466 100 4.1546 2.1555 29.0000 15.2000 16.6780 81 1 -1.0953 11.1639 1.5567 13.7654 100 0.8599 0.2678 29.0000 0.0000 1.4780 1 1 1.5568 13.8207 4.2088 16.4222 100 1.3243 0.4899 29.0000 0.0000 1.4780 2 1 4.2089 16.3667 6.8609 18.9682 100 2.1507 0.7350 29.0000 0.0000 1.4780 3 1 6.8610 19.0235 9.5130 21.6250 100 -0.2609 1.3538 29.0000 0.0000 1.4780 4 1 9.4112 21.6804 12.0632 24.2818 100 0.4833 0.4405 29.0000 0.0000 1.4780 5 176 1 12.0633 24.2265 14.7153 26.8280 100 1.9984 0.9457 29.0000 0.0000 1.4780 6 1 1.2510 8.8392 3.9030 11.4407 100 0.7608 0.7723 29.0000 1.9000 3.3780 7 1 3.9031 11.4958 6.5551 14.0973 100 2.3114 0.6838 29.0000 1.9000 3.3780 8 1 6.5552 14.0420 9.2072 16.6435 100 0.4506 1.0849 29.0000 1.9000 3.3780 9 1 9.1054 16.6988 11.7574 19.3003 100 -0.6185 3.1578 29.0000 1.9000 3.3780 10 1 11.7567 19.3557 14.4087 21.9571 100 1.3282 4.0819 29.0000 1.9000 3.3780 11 1 14.4088 21.9018 17.0608 24.5033 100 0.2074 0.9030 29.0000 1.9000 3.3780 12 1 3.5973 6.6252 6.2493 9.2266 100 0.2249 0.0723 29.0000 3.8000 5.2780 13 1 6.2494 9.1712 8.9014 11.7726 100 1.9550 0.9830 29.0000 3.8000 5.2780 14 1 8.9007 11.8280 11.5527 14.4294 100 1.6076 0.4720 29.0000 3.8000 5.2780 15 1 11.4508 14.3742 14.1028 16.9756 100 1.3026 0.6923 29.0000 3.8000 5.2780 16 1 14.1029 17.0310 16.7549 19.6324 100 0.7295 2.1167 29.0000 3.8000 5.2780 17 1 16.7550 19.5772 19.4070 22.1786 100 1.5614 2.4159 29.0000 3.8000 5.2780 18 1 5.9428 4.3005 8.5948 6.9020 100 0.2307 0.7439 29.0000 5.7000 7.1780 19 1 8.5949 6.8465 11.2469 9.4479 100 0.2587 0.2500 29.0000 5.7000 7.1780 20 1 11.1450 9.5033 13.7970 12.1047 100 0.2172 0.6420 29.0000 5.7000 7.1780 21 1 13.7971 12.0495 16.4491 14.6509 100 0.6006 1.3243 29.0000 5.7000 7.1780 22 1 16.4492 14.7063 19.1012 17.3077 100 1.6601 1.2185 29.0000 5.7000 7.1780 23 1 19.1013 17.2523 21.7533 19.8537 100 1.3233 2.2221 29.0000 5.7000 7.1780 24 1 8.2890 1.9758 10.9410 4.5773 100 0.0626 1.4456 29.0000 7.6000 9.0780 25 1 10.9411 4.5218 13.5931 7.1232 100 0.5337 0.4546 29.0000 7.6000 9.0780 26 1 13.4913 7.1786 16.1433 9.7800 100 0.8918 0.3744 29.0000 7.6000 9.0780 27 1 16.1434 9.7248 18.7954 12.3262 100 -0.2080 1.0387 29.0000 7.6000 9.0780 28 1 18.7947 12.3816 21.4467 14.9830 100 1.2601 0.9080 29.0000 7.6000 9.0780 29 1 21.3449 14.9276 23.9969 17.5290 100 0.5614 0.9906 29.0000 7.6000 9.0780 30 1 10.6353 -0.3489 13.2873 2.2526 100 0.5173 0.2618 29.0000 9.5000 10.9780 31 1 13.1847 2.1971 15.8367 4.7985 100 0.0685 1.0385 29.0000 9.5000 10.9780 32 1 15.8368 4.8539 18.4888 7.4554 100 0.0159 0.0316 29.0000 9.5000 10.9780 33 177 1 18.4889 7.4001 21.1409 10.0015 100 0.4298 0.3264 29.0000 9.5000 10.9780 34 1 21.1410 10.0569 23.7930 12.6584 100 0.2501 0.4694 29.0000 9.5000 10.9780 35 1 23.6911 12.6029 26.3431 15.2043 100 0.3142 0.1870 29.0000 9.5000 10.9780 36 1 12.9808 -2.6736 15.6328 -0.0721 100 1.8746 0.8497 29.0000 11.4000 12.8780 37 1 15.5309 -0.1276 18.1829 2.4738 100 1.0630 1.3299 29.0000 11.4000 12.8780 38 1 18.1830 2.5292 20.8350 5.1307 100 -0.4306 1.6400 29.0000 11.4000 12.8780 39 1 20.8351 5.0754 23.4871 7.6768 100 0.4587 0.6433 29.0000 11.4000 12.8780 40 1 23.3853 7.7322 26.0373 10.3337 100 0.8180 1.1324 29.0000 11.4000 12.8780 41 1 26.0366 10.2782 28.6886 12.8796 100 0.3933 0.7954 29.0000 11.4000 12.8780 42 1 15.2251 -4.9983 17.8771 -2.3968 100 2.8672 1.5363 29.0000 13.3000 14.7780 43 1 17.8772 -2.4523 20.5292 0.1492 100 1.1866 1.7584 29.0000 13.3000 14.7780 44 1 20.5293 0.2045 23.1813 2.8060 100 0.1964 0.5458 29.0000 13.3000 14.7780 45 1 23.1806 2.7507 25.8326 5.3522 100 -0.2569 1.3407 29.0000 13.3000 14.7780 46 1 25.7308 5.4075 28.3828 8.0090 100 0.0314 0.4687 29.0000 13.3000 14.7780 47 1 28.3829 7.9535 31.0349 10.5549 100 0.0279 0.0140 29.0000 13.3000 14.7780 48 1 17.5714 -7.3229 20.2234 -4.7215 100 3.1591 1.6352 29.0000 15.2000 16.6780 49 1 20.2227 -4.7770 22.8747 -2.1755 100 1.7745 1.8217 29.0000 15.2000 16.6780 50 1 22.8748 -2.1202 25.5268 0.4813 100 0.0382 0.1195 29.0000 15.2000 16.6780 51 1 25.4249 0.5367 28.0770 3.1381 100 -0.3367 1.3104 29.0000 15.2000 16.6780 52 1 28.0770 3.0828 30.7290 5.6843 100 -0.0380 1.8403 29.0000 15.2000 16.6780 53 1 30.7291 5.7397 33.3811 8.3411 100 -0.4672 1.9956 29.0000 15.2000 16.6780 54 Grid Parameters 1 ---------------------------- Start-x = -111.0000 2 ---------------------------- Start-y = -220.0000 3 -------------------------- Finish-x = 195.0000 4 -------------------------- Finish-y = 278.0000 5 ------------------------ x-increment = 5.0000 6 ------------------------ y-increment = 5.0000 Size Parameters 1 -------------------------- Plot size = 1.000000 2 -------------- Shade/Color increment = 0.100000 3 ------ Exaggeration for disp.& dist. = 10000.00 Cross section default 178 1 ---------------------------- Start-x = -6.70000 2 ---------------------------- Start-y = 34.66000 3 -------------------------- Finish-x = 33.58000 4 -------------------------- Finish-y = 34.66000 5 ------------------ Distant-increment = 0.500000 6 ---------------------------- Z-depth = 11.00000 7 ------------------------ Z-increment = 0.500000 Map info 1 ---------------------------- min. lon = 119.5000000 2 ---------------------------- max. lon = 122.5000000 3 ---------------------------- zero lon = 120.5882000 4 ---------------------------- min. lat = 21.5000000 5 ---------------------------- max. lat = 26.5000000 6 ---------------------------- zero lat = 23.4869000