<%BANNER%>

Permeabilities of Subduction Zone Sediments and Their Effect on Pore Pressure Generation


PAGE 1

PERMEABILITIES OF SUBDUCTION ZO NE SEDIMENTS AND THEIR EFFECT ON PORE PRESSURE GENERATION By KUSALI R. GAMAGE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Kusali R. Gamage

PAGE 3

To my parents and all my teachers.

PAGE 4

iv ACKNOWLEDGMENTS I would like to express my sincere gratitude to my advisor, Dr. Elizabeth Screaton, for her continuous support and encouragemen t throughout my graduate studies. She guided me through the dissertation process, ne ver accepting less than my best efforts. Her technical and editorial advice was essentia l to the completion of this dissertation and has taught me innumerable lessons and insights on the workings of ac ademic research in general. I cannot thank her enough for beari ng my complaints and procrastination with unbelievable patience. My thanks also go to the members of my committee, Drs. David Foster, Jonathan Martin, Louis Motz, and D ouglas Smith, who have patiently sat through committee meetings and taken time to unders tand my work. I also thank them for reading previous drafts of this dissertati on and providing many valuable comments that improved the presentation and contents of this dissertation. Many thanks go to Dr. Barbara Bekins of the USGS, Menlo Park, California, for providing core samples and suggesting work ing on the Peru permeability measurements and Dr. Ivano Aiello of the Moss Landing Mari ne Laboratories, California, for providing me with grain size data relating to this resear ch. I am also very gr ateful to Kevin Hartl for his invaluable technical support, Dr. J ohn Jaeger and William Vienne for helping me with grain size analyses and Dr. Jason Curtis for carbon analyses. My special thanks go to my colleagues Troy Hays, George Kame nov, Jennifer Martin, Victoria Mejia and many friends out in the real world for all their support a nd encouragement throughout my graduate studies.

PAGE 5

v Last, but not least, I would like to thank my husband, Sanjaya, for his understanding and love during th e past few years. His encouragement was in the end what made this dissertation possible. My parents, Punyasena and Kalyani Gamage, and my brother, Rachita Gamage, receive my deepest gratitude and love for their dedication and the many years of support during my undergraduate studies that provided the foundation for this work.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT......................................................................................................................x ii CHAPTER 1 GENERAL INTRODUCTION....................................................................................1 Deformation Processes of Subduction Zones........................................................2 Role of the Dcollement Zone...............................................................................3 Critical Taper Theory............................................................................................4 Role of Fluid Flow at Convergent Margins..................................................................6 Evidence for Fluid Flow and Pressures.................................................................6 Basics of Fluid Flow Re lated to Subduction Zone.....................................................10 Permeability.........................................................................................................11 Hydrogeologic Modeling.....................................................................................12 Statement of Problem.................................................................................................13 2 A COMPARATIVE STUDY OF PERMEABILITY MEASUREMENTS FROM THE SUBDUCTION ZONES OF NORT HERN BARBADOS, COSTA RICA, NANKAI, AND PERU...............................................................................................16 Introduction.................................................................................................................16 Background.................................................................................................................18 Barbados..............................................................................................................18 Costa Rica............................................................................................................19 Nankai..................................................................................................................20 Peru......................................................................................................................22 Laboratory Permeability Data.....................................................................................23 Barbados..............................................................................................................24 Nankai and Peru..................................................................................................25 Costa Rica............................................................................................................26 Permeability-Porosity Relationship............................................................................27 Description of Statistical Methods..............................................................................27

PAGE 7

vii Results........................................................................................................................ .28 Effects of Depositional Environment..................................................................29 Effects of Grain Size...........................................................................................34 Effects of Structural Domain...............................................................................38 Discussion...................................................................................................................40 Conclusions.................................................................................................................42 3 CHARACTERIZATION OF EXCESS PO RE PRESSURES AT THE TOE OF THE NANKAI ACCRETIONARY COMPLEX, OCEAN DRILLING PROGRAM SITES 1173, 1174, AND 808: RESULTS OF ONEDIMENSIONAL MODELING..................................................................................44 Introduction.................................................................................................................44 Background.................................................................................................................45 Geologic Setting..................................................................................................45 Previous Hydrologic Studies...............................................................................48 Laboratory Permeability Measurements.....................................................................52 Modeling Methods......................................................................................................55 Theoretical Background......................................................................................55 Model Implementation........................................................................................58 Model Dimensions, Boundary Condi tions, and Initial Conditions.....................59 Sedimentation and Prism Thickening Rates........................................................61 Results........................................................................................................................ .62 Model Results......................................................................................................62 Sensitivity to Bulk Permeability..........................................................................65 Sensitivity to Lateral Stress.................................................................................67 Sensitivity to a Low-Permeability Barrier...........................................................70 Implications................................................................................................................73 Conclusions.................................................................................................................74 4 EVOLUTION OF HIGH PORE PR ESSURES AND IMPLICATIONS FOR EPISODIC FLUID FLOW AT THE NORTHERN BARBADOS ACCRETIONARY COMPLEX.................................................................................76 Introduction.................................................................................................................76 Background.................................................................................................................79 Fluid Flow and Pore Pressures............................................................................82 Hydrofractures.....................................................................................................85 Modeling Methods......................................................................................................86 Model Implementation........................................................................................86 Model Equations.........................................................................................................87 Model Dimensions, Boundary Condi tions, and Initial Conditions.............................88 Sedimentation and Loading Rates.......................................................................91 Results and Discussion...............................................................................................93 Conclusions...............................................................................................................102 5 SUMMARY AND CONCLUSIONS.......................................................................104

PAGE 8

viii APPENDIX: LISTING OF PERMEABI LITY, POROSITY, GRAIN SIZE DATA WITH REFERENCES USED IN THIS STUDY.....................................................107 LIST OF REFERENCES.................................................................................................121 BIOGRAPHICAL SKETCH...........................................................................................133

PAGE 9

ix LIST OF TABLES Table page 2-1 Log linear permeability-porosity relationships predicted for varying lithologies at Barbados, Costa Rica, Nankai and Peru...............................................................33 2-1 Permeability-porosity relationships based on grain size analyses...........................35 3-1 A summary of laborator y measured permeabilities for samples from ODP Leg 190 Sites 1173 and 1174. .......................................................................................53 3-2 Fluid and solid matrix propertie s used for numerical simulations...........................61 3-3 Summary of sedimentation rates calcula ted from biostratigraphy at Sites 1173, 1174 and 808 and prism thickening rates calculated from prism geometry and convergence rate.......................................................................................................62 3-4 Summary of simulation r uns at Site 1174 and 808. ...............................................64 4-1 Fluid and solid matrix propertie s used for numerical simulations...........................91 4-2 Summary of sedimentation rates calcula ted from biostratigraphy at Sites 1173, 1174 and 808............................................................................................................92 4-3 Summary of simulation runs with the estimated values of maximum .................99 A-1 Listing of permeability, porosity, grain size data with references used in this study. .....................................................................................................................108

PAGE 10

x LIST OF FIGURES Figure page 1-1 Diagram of accretionary prism and the processes of frontal accretion and underplating. ..............................................................................................................3 1-2 Schematic of an ideal coulomb wedge......................................................................5 2-1 Map of large-scale regi onal setting location of the dr ill sites at the Barbados accretionary complex. .............................................................................................18 2-2 Map showing location of ODP drilli ng along the Costa Rica subduction zone. ....20 2-3 Location map of ODP Leg 190 and pr evious ODP/DSDP drill sites in the Nankai Trough. .......................................................................................................21 2-4 Map showing general locations of drill sites occupied during ODP Legs 138 and 112 at Peru subduction zone. ...........................................................................23 2-5 Plot of laboratory derived permeability measurements from Barbados, Costa Rica, Nankai, and Peru subduction zones................................................................29 2-6 Permeabilities classified based on depositional environment and location............33 2-7 Permeabilities classified based on grain size distribution. ....................................37 2-8 Permeabilities classified based on structural domain. ...........................................39 3-1 Location map of the study area in th e Nankai accretionary complex and sites used for this study....................................................................................................46 3-2 Schematic interpretation of the Muro to Transect showing tectonic domains and location of Leg 190 drill sites used in this study......................................................47 3-3 Porosity profiles of Sites 808, 1174, and 1173. ....................................................50 3-4 Permeability data measured for samples from ODP Leg 190.................................54 3-5 Simulated porosity and profiles for base run at Site 1173..................................65 3-6 Simulated porosity and profiles with varying bulk permeability at Sites 1174 and 808. ..................................................................................................................66

PAGE 11

xi 3-7 Simulated porosity and profiles with added lateral stress at Sites 1174 and 808. ......................................................................................................................... 68 3-8 Simulated porosity and profiles with added low-permeability barrier at Sites 1174 and 808. .........................................................................................................71 4-1 Location map and cross-section of the Barbados accretionary complex.................80 4-2 Grid and boundary conditions for the prism growth/flow model............................90 4-3 Simulated values for base run at 0.6, 1.3, 2.0, and 2.7 million years.................97 4-4 Estimated values in the dcollement at 0.6, 1.3, 2.0 and 2.7 million years of prism growth for all simulations............................................................................101

PAGE 12

xii Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PERMEABILITIES OF SUBDUCTION ZO NE SEDIMENTS AND THEIR EFFECT ON PORE PRESSURE GENERATION By Kusali R. Gamage December 2005 Chair: Elizabeth Screaton Major Department: Geological Sciences Permeability is a fundamental sediment prope rty influencing flui d flow, hence fluid pressures in the subsurface. Because elevated fluid pore pressures play a critical role in the development of accretionary comple xes, including the development of the dcollement zone, it is important to simu late pore pressures based on a systematic relationship of permeability and porosity. Based on laboratory permeability measurements of Northern Barbados, Co sta Rica, Nankai and Peru subduction zones sediments, a high correlation between permeability and porosity was found for argillaceous sediments while little corr elation was found for carbonate dominant sediments. Classification based on location and grain size distribu tion provided greater correlation between permeability and porosity than the depositional environment of the sediment alone. In the second part of th e research, a one-dimensional loading and fluid flow model near the toe of th e Nankai subduction zone was used to examine the effects of lower bulk permeability (sensitivity to a permeab ility-porosity relationship), lateral stress

PAGE 13

xiii in the prism, and addition of a low-permeability barrier to the dcollement. The results predicted significant increase in pore pressure s below the dcollement zone with lower bulk permeability, or when a low-permeability barrier is added at the dcollement. Both simulations with lateral stress and a low-perm eability barrier at the dcollement resulted in sharp increases in porosity at the dcoll ement, similar to that observed in measured porosities. In addition, these two scenarios predict maximum excess pore pressure ratios at the dcollement suggesting that either of these factors would contribute to stable sliding along the dcollement. In the third part of the research, results from a twodimensional prism growth and flow model i ndicate pore pressures close to lithostatic pressures at the dcollement when dcollem ent was given the same permeability as the surrounding sediments. However, these pore pr essures were unable to reach lithostatic pressures thus not allowing horizontal hydrofra cture in the dcollem ent zone. Addition of vertical hydrofractures in the underthrust sediments did not increase pore pressures to lithostatic pressures in the dcollement. When a bulk permeability-vertical stress relationship was assigned to the dcollement pore pressures reach values close to lithostatic pressures, suggesting that high pore pressures can be sustained at the base of the prism while fluid is expelled at the toe of the complex.

PAGE 14

1 CHAPTER 1 GENERAL INTRODUCTION Examining the fluid flow of the deep hydrosphere is extremely important because fluid flow alters the physical and chemical properties of the Earth’s crust, which in turn affects the ocean and the atmospheric chem istry that is vital for human existence (COMPLEX, 1999). At active plate margins, fluid flow can influence movement along faults and thus the nature of the earthquake cycle. Several re search projects have recently been focused on studying fluid flow along ac tive plate margins. The Ocean Drilling Program (ODP) has contributed valuable info rmation on fluid behavior by sampling the sediments and oceanic crust at sha llow ends of the subduction zones. At convergent margins, the incoming sediments and lithosphere are fed into the subduction factory where processes such as compaction and dewatering, diagenesis, dehydration, metamorphism, melting, melt migr ation and mantle convection result in hazardous seismicity, explosive volcanism as well as the formation of ore deposits and new continental crust (Moore, 1998). A large number of the world’s greatest earthquakes are associated with subduction zones. A sm all portion of the plate contact, known as the seismogenic zone, is responsible for genera ting these large earthquakes. Understanding the processes of the seism ogenic zone provides valuable information on earthquake generation, but requires studyi ng many aspects of geology. Shallowly dipping subduction zones provide a large fault surface that is accessible to study by allowing sampling of the incoming sedi ments. In such localities, accretionary complexes are formed if sediments are sc raped off the subducting oceanic plate and

PAGE 15

2 accreted on the upper plate. Many accretiona ry complexes grow above sea level and even until they form mountain belts. On e such example of a partially exposed accretionary complex is the Barbados Island located in the Caribbean. Examples of ancient complexes include the Shimanto Belt of southwest Japan, the Franciscan complex of California, and the Kodiak accretionary complex of Alaska. Even though more than half of the worl d’s convergent plate boundaries are forming accretionary prisms (von Huene and Scholl, 1991) there are margins where all of the sediments riding on the oceanic plate are underthrust beneath th e upper plate of the subduction zone (e.g., Costa Rica). Particularly in the west ern Pacific, trenches lack accretionary sediments because the terrigenous sediment supply reaching the trench floor is insufficient to accrete (von Huene and Sc holl, 1991). Thus, typi cally non-accreting margins are bordered by sediment-starved tren ches such as the Mariana and Tonga. Deformation Processes of Subduction Zones At convergent boundaries, sediments can be eith er offscraped as a series of thrust sheets at the frontal edge of the accretiona ry prism or underthrust with the subduction plate to great depths (Moore, 1989). Accreted sediments form a series of imbricate thrust sheets that extend from the surface to the ba sal detachment fault or the dcollement. Accretionary prisms grow volumetrically by two main processes (Figure 1-1). On the surface, accretionary prisms gr ow by frontal accretion while on the subsurface they grow by underplating (von Huene and Scholl, 1991). According to von Huene and Scholl (1991) the division of these two processes refe rs to the seaward position of the margin’s resistive rock structure (backstop), which al so acts as the mechanical backstop of the seaward part of the margin. Frontal accre tion takes place in front of the backstop by offscraping the upper part of the oceanic sedi ment while the lower parts of the oceanic

PAGE 16

3 sediment are underthrust. Sediment subduc ted beneath the backst op is subsequently accreted by underplating or transported to grea ter depths (von Huene and Scholl, 1991). During frontal accretion, thrust slices w ill detach the upper part of the incoming sediments. When sediments move from the oceanic plate to the accretionary prism, the state of stress changes from gravitational to that of a thrust belt (Moore, 1989). On the oceanic plate the maximum principal stress is oriented vertically and as sediments get accreted the maximum principal stress gradually inclines (Davis et al., 1983). With the stacking and upward rotation of these thrust sl ices, the accreted material will thicken and shorten (von Huene and Scholl, 1991). Figure 1-1. Diagram of accretionary prism a nd the processes of frontal accretion and underplating (modified from von Huene and Scholl, 1991). Role of the Dcollement Zone The dcollement zone is the principal boundary that separates the upper and lower converging plates (von Huene and Scholl, 1991 ). The sediments above the dcollement are highly deformed while the sediments belo w remain coherent (Moore et al., 1982). Change in structural style acr oss the dcollement suggests that this zone marks a major shift in the orientation of the stresses (Moore, 1989). Th is has been supported by the presence of extensional veins observed in th e underplated sediments that are interpreted as hydrofractures with near ve rtical orientation of the maxi mum principal stress (Fisher and Byrne, 1987). An important question that has not yet been answered is how the

PAGE 17

4 dcollement initiates. It has been proposed by many studies that elevated fluid pressures are necessary for the initiation and sliding of the dcollement (Davis et al., 1983; Westbrook and Smith, 1983). If the dcoll ement steps down within a packet of subducted sediments, then the material of the lower plate can be tran sferred to the upper plate, effecting underplating a nd volumetric growth of the prism (von Huene and Scholl, 1991). Furthermore, if the dcollement moves upward, accreted material will be transferred to the lower plate promo ting subduction erosion (Charlton, 1988). At some accretionary prisms such as Cas cadia, southern Chile and eastern Alaska, the dcollement lies at the base of the inco ming sedimentary section suggesting that all incoming sediments are frontally accreted ( von Huene and Scholl, 1991). Furthermore, Davis and Hyndman (1989) inferred that large accretionary pr isms such as Barbados or the Makran prism of southern Pakistan ha ve achieved their exceptional size due to efficient offscraping favored by slow c onvergence and thick incoming sediments, although the dcollement is located we ll above the igneous basement. Critical Taper Theory The critical taper theory has been widely used to explain the shape of accretionary prisms as well as to estimate excess pore pr essures. Davis et al. (1983) used Coulomb failure theory to demonstrate that homogeneous wedges reach a stable critical taper that remains constant as long as the controlling parameters do not change. Once it reaches the maximum thickening and shortening, the accre tionary prism will maintain its taper by adding material either by underplating or by ne w thrust faults that cut the accretionary prism at shallow angles, which are known as the out of sequence thrusts (von Huene and Scholl, 1991). The critical taper is defined by where and are the topographic

PAGE 18

5 slope and the dcollement dip respectively (Fig ure 1-2). The parameters that control the taper angle are the internal ( ) and basal friction angles ( b) and the pore fluid pressure ratios of the wedge ( ) and the base ( b). The is given by seafloor litho seafloor fP P P P where Pf is the pore pressure in the sediments, Pseafloor is that pressure in the water column above the seafloor, and Plitho is the total pressure of the overlying water column and sediments. The internal () and basal friction (b) coefficients are given by tanandbtanbAccording to critical wedge theory (D avis et al., 1983; Dahlen, 1984) the wedge taper gives an indication of eith er the material properties of the wedge or the friction at the base of the wedge. A large critical an gle indicates either a weak material, which needs to deform before stable sliding could occur, or high basal friction (Davis et al., 1983). In contrast a small critical angle indicat es either a strong mate rial, which need not deform for stable sliding to occur, or very little basal friction (Davis et al., 1983). Figure 1-2. Schematic of an ideal coulom b wedge (modified from Hatcher, 1995). Using the concepts presented by Davis et al (1983), Bernstein-Ta ylor et al. (1992) interpreted that a large change in basal friction could result from a change in fluid

PAGE 19

6 pressure from an overpressured dcollement zone to hydrostatic fluid pressure beneath the toe. If the material is different from the toe to arcward, then the material at the toe can be relatively weak and can deform until a large critical taper is achieved. Arcward of the toe, stronger material need not deform inte rnally for stable sliding and thus, will have a small critical angle (Davis et al., 1983). Role of Fluid Flow at Convergent Margins Approximately 40% of the sediment sect ion entering the world’s subduction zones is composed of water in pore spaces (Moore et al., 2001). Fluid flow at accretionary complexes is due to a number of different driving forces, including gravitational loading, tectonic compression, fluid de nsity gradients, and dehydrat ion reactions. During both gravitational loading and tectonic compre ssion, sediments will generally compact. However, if the rate of lo ading or compression is suffici ently high, then fluids cannot escape fast enough, which causes pore pressure s to rise. These localized excess pore pressures generate a pressure gradient and he nce fluid flow. Density variations can result from differences in solute concentrations or due to the introducti on of heat sources. Mineral dehydration is another method wher e fluids are released during the reaction, hence increasing the fluid pore pressures. A common example of a dehydration reaction found in accretionary complexes is th e smectite to illite transition. Evidence for Fluid Flow and Pressures At many active accretionary prisms (e .g., Japan, Barbados, Oregon-Washington, Marianas) evidence for fluid expulsion has been observed, including the presence of biological communities, heat flow and geoc hemical anomalies, and mud volcanoes and diapirs (Peacock, 1990). Each of the evid ence types is summarized in the following paragraphs.

PAGE 20

7 One of the most interesting indicators of fluid expulsion from accretionary prisms is the presence of biological communities (L e Pichon et al., 1987). Studies have shown that the range of biomass found in colonies is related to the fluid chemistry and flow rate (Sibuet et al., 1990). Thus, studying the nature and distribution of the colonies provide valuable information on qualitative fluid flow. Such examples of ancient seep communities have been identified in accreti onary prisms. These biological communities were formed as concentrations of macrofo ssils in deepwater rock s (Moore and Vrolijk, 1992). If flow continues at seeps for a longer time the biomass can grow beyond the limits of the calcite compensation depth (depth below which calcite is dissolved in the deep-sea) developing into reef like structur es that eventually could transform into hydrocarbon reservoirs (Hovland, 1990). Fluid transport has been inferred from fl uid inclusions from veins that show anomalously high temperatures at shallow de pths in ancient accretionary complexes (Vrolijk et al., 1988). At modern accretionary complexes, the age of the oceanic crust can be used to predict conductive heat flux a nd therefore, anomalie s from the conductive heat flux that might be due to fluid flow can be identified. The be st evidence for deeply derived warm fluids comes from the Northern Barbados accretionary prism (Davis and Hussong, 1984). It is interprete d that the observed heat flow anomalies were caused by advection of heat during cha nnelized fluid flow along faults (Moore and Vrolijk, 1992). At the Barbados accretionary complex, both bor ehole temperatures and marine heat flow measurements demonstrate thermal gradients approximately twice that expected for the age of the subducting crust (Fisher and Houns low, 1990). Heat flow values that are lower than predicted by conductive cooling of the oceanic crust have been reported

PAGE 21

8 offshore of Peru (Langseth a nd Silver, 1996). One possible e xplanation for this low heat flow is the rapid sedimentation at Peru, which may prevent equilibrium (Moore and Silver, 2004). Similarly, heat flow meas urements at Cocos Plate off the Nicoya Peninsula have revealed significantly low heat flow valu es (Langseth and Silver, 1996). It has been suggested that hydrothermal cooli ng of the oceanic basement occurs from the flow of seawater into the uppe r crust of the subducting plate (Langseth and Silver, 1996). Furthermore, evidence for possible fluid migration comes from observed low chloride anomalies (concentrations less th an seawater), found in many modern prism sediments (Kastner et al., 1991). One of the widely used explanations for the observed low chloride anomalies is the dehydration reac tion of smectite to illite. Smectite is a common and abundant type of clay found in subduction zones (Moore and Vrolijk, 1992). The dehydration of smectite to illite is a kinetic reaction that depends on temperature and time (Elliott et al., 1991). The reaction takes place in temperatures above 60C and the amount of water released during the reaction is estimated to be 20% by weight (Bekins et al., 1995). Smectite is replaced by fluid plus illite, which has greater volume than smectite, thus resulting in an increase in pore pressures (Bekins et al., 1995). At Barbados, low-chloride anom alies were observed along the dcollement (Kastner et al., 1993) and it is inferred th at pore fluids generated from smectite dehydration migrate toward the toe of the prism mainly through faults or fractures lowering the chloride concentration at the to e of the prism (Bekins et al., 1995). A broad low chloride anomaly was also observed above and below the dcollement at the Nankai accretionary complex. Because the smectite contents at the Nankai sites are low, it has been inferred that freshening of pore fluids may be related to the in situ dehydration

PAGE 22

9 caused by high temperature regimes (Brown et al., 2001). The high temperature regimes are believed to be related to the fossil sp reading ridge (Kinan Seamount) that ceased spreading ~ 15 Ma ago located on the Philippine Sea Plate (Shipboard Scientific Party, 2001). Mud volcanoes and diapirs are also co mmon features found in accretionary complexes. They form due to increased pr essure at depth and tr ansport overpressured mud to the surface. In areas where serpentine diapirs or volcanoes are present, it requires fluids to hydrate basalt in order to form serpentine and these structures become conduits of fluid flow from depth (Fryer et al., 1990). During diapirism and mud volcanism rock sequences are disrupted and mlanges are formed (Brown and Westbrook, 1988). Mud diapirs and mud volcanoes are of ten observed along thrust faults, suggesting thrusting as a mechanism that triggers mud diapirism (Behrmann, 1992). All the evidence for fluid flow from accre tionary complexes implies fluid pressure gradients in excess of hydrostatic because pres sures gradients drive fl uid flow. Indirect evidence for elevated fluid pressures includes the presence of exte nsional veins (Moore and Vrolijk, 1992). Crack-seal textures in veins indicate re peated pulses of high fluid pressure (Vrolijk, 1987; Fisher and Byrne, 1990). Repeated events of vein growth suggest that fluid pressure evolve throughout the growth of the accretionary complex (Labaume et al., 1997). Fluid pressure in a ccretionary prisms can vary from hydrostatic (equivalent to the weight of the overlying column of water) to nearly lithostatic (equivalent to the weight of the overlying co lumn of sediments). Variation in fluid pressures from hydrostatic to lithostatic has been observed in wells at the eastern Aleutian Trench where the fluid pressures are at hydrostatic at 2-3 km from the surface

PAGE 23

10 and increasing to more than 80% of lithos tatic pressures at to tal depth (Moore and Vrolijk, 1992). Basics of Fluid Flow Re lated to Subduction Zone Although sediments found in accretionary prisms are highly deformed during subduction, it is assumed that the high density and interconnectedness of the fractures in accretionary prisms approximate Darcian fl ow (Moore and Vrolijk, 1992). The fluid produced during the accretionary process can be evaluated usi ng the two principles that govern fluid flow in the subsurface. They are the principle of conservation of mass and Darcy’s law. The principle of conservation of mass states that for an arbitrary control volume, the rate of mass accumulation within the volume plus the net mass flux out of the volume must equal the rate of mass genera tion within the volume (Bird et. al., 1960). If we consider a very sma ll volume of the aquifer known as a control volume, we can approximate the flow through the matrix usi ng Darcy’s law. The most basic form of Darcy’s law states Q/A=-K dh/dl where Q/A is flow per area or linear velocity [LT-1], K is hydraulic conductivity [LT-1], and dh/dl is hydraulic gradient. The hydraulic conductivity (K), which is a proportionality constant, represents both propert ies of the fluid and the porous media. It is given by K= kg/ where k is intrinsic permeability [L2], which is representative of the properties of the porous medium, is the fluid density [ML-3], is the fluid viscosity [L-1 T-1] and g is the gravitational constant [L T-2]. Intrinsic permeability depe nds on variables such as grain

PAGE 24

11 size, sorting, and roundness of the sediment through which fluid flowing. Hereafter, intrinsic permeability will be re ferred to as permeability. Permeability Sediment permeability is the most important factor that controls pore pressures as permeability can vary by many orders of ma gnitude (e.g., Bruckmann et al., 1997; Saffer and Bekins, 1998). Therefore, the use of a systematic relationship between porosity and permeability is valuable for approximating th e permeability structure in an accretionary wedge fluid flow model (Saffer and Bekins 1998). Processes in subduction zones such as loading, compaction and cementation of sedi ments lead to reductions in permeability. When permeability is reduced and reaches some critical value, rocks can no longer transmit significant amounts of fluid, but new permeability can be created in the form of a fracture or a fault. If fractures are form ed then they can significantly affect the permeability of the accretionary complex and, thus, large-scale field measurements of permeability would be more appropriate than core scale permeability measurements to determine large-scale fluid flow (Moore a nd Vrolijk, 1992). Unfortunately only a few large-scale field measurements of permeability have been made in accretionary settings. These include shipboard packer tests and s ubmersible-based tests conducted at a sealed borehole at the Oregon accretionary complex (Screaton et al., 1995) and shipboard packer tests (Fisher and Zwart, 1996) and submersible slug tests and discharge tests (Screaton et al., 1997) in the dcollement of the Barbados accretiona ry complex. Because of difficulties in conducting large-scale field measurements, permeability measurements are primarily from core samples that are retrieved from the frontal part or shallow depths of the accretionary complex. Even though these core sample measurements do not represent the large-scale variations in permeability due to faults, they provide valuable estimates

PAGE 25

12 for the matrix permeability that is critical in approximating permeability structures in the accretionary complex for modeling studies. Hydrogeologic Modeling Mathematical models used in hydrologic modeling are derived from the governing principles of fluid flow and specificati ons such as formation geometry, boundary conditions and initial conditions. These models help quantify conceptual models of subseafloor hydrogeologic flow system. These models can be extremely useful and cost effective in providing possible explanations for known or hypothesized conditions. They can also be used to assess whether or not a conceptual model is feasible. As a starting point, with limited data it is best to use one or two-dimensional analytical solutions derived from simple well-defined boundary problems (Anderson and Woessner, 1992). However, in sub-seafloor settings, numerical models are often n ecessary in order to account for parameters such as complex ge ometry, variable density fluid flow, and variations in heat flow. Due to limited access to convergent margins, models are essen tial for integrating the field observations with laboratory results. It has also been recognized that numerical models are required in order to extend obs ervations made at shallow parts of the subduction system to greater depths such to the seismogenic zone (COMPLEX, 1999). Most previous modeling studies have focu sed on coupled compaction-fluid flow and diffusion-advection models of pore fluid ch emistry and heat for Barbados, Nankai and Cascadia accretionary prisms (e.g., Bekins et al., 1995; Saffer a nd Bekins, 2002; Screaton and Ge, 1997). However, recent data co llection allows significantly improved characterization of permeability, which is a major component that affects modeled fluid pressures in accretionary complexes. Furthe rmore, previous modeling studies at Nankai

PAGE 26

13 have largely focused on estimating pore pres sures with little focus on examining the causes of excess pore pressu res (Le Pichon and Henry, 1992; Screaton et al., 2002; Saffer, 2003). At Barbados, modeling studies were focused on pore pressure generation either at the toe of prism (Henry a nd Wang, 1991; Shi and Wang, 1994; Stauffer and Bekins, 2001) or in a steady-state appro ach instead of examining pore pressure generation through the subduction process. Statement of Problem The objective of this investigation was to expand the knowledge of fluid flow and the development of pore pressures based on both laboratory measured permeability values and numerical models at selected accre tionary complexes. This study benefits our current understanding of fluid flow in accreti onary complexes in several ways. One of the primary benefits of this research is the contribution and synthesis of permeability measurement of subduction zone sediments, which provides new insight to flow simulations in convergent margins. This study also provides valuable information on fluid flow paths, areas of excess pore pre ssures, degree of impor tance of lithology and sediment thickness in fluid flow, the initiation of dcollement, and factors that contribute to the initiation of the dcollement. These results further provide valuable information for future drilling projects such as the Se ismogenic Zone Experiment (SEIZE) that is focused on understanding the relationship be tween earthquakes, deformation, and fluid flow. The following were the speci fic objectives of this study: To synthesize permeability data and pred ict permeability-porosity relationships at four major convergent margins. The four locations are the Northern Barbados, Costa Rica, Nankai and Peru subduction zones, which represent a variety of marine sediments. To investigate the effects and magnitudes of parameters such as bulk permeability, lateral stress, and the presence of a low-permeability barrier at the dcollement on

PAGE 27

14 the generation of excess por e pressures at the toe of the Nankai accretionary complex. To investigate the evolution of pore pressu res and implications for episodic fluid flow at the Barbados accr etionary complex using a tw o-dimensional growth and flow model. The following chapters include detailed methodology and discussions of the results of the proposed research. Chapter 2 is titled “A comparative study of permeability measurements from the subduction zones of northern Barbados, Costa Rica, Nankai and Peru” and a modified version of this chapte r will be submitted for publication to Marine Geology. This chapter is a contribution to the permeability data of marine sediments at subduction zones. The major results of this chapter include relationships among permeability and porosity for different types of marine sediments. The constraints provided by these relationships will allow re alistic estimation of fluid flow and pore pressures in marine settings. Furthermore, the laboratory measur ed permeability data from Nankai and Peru contributes to the gene ral knowledge of mari ne sediments. The permeability data of this work has been published as two data reports (Gamage and Screaton, 2003; Gamage et al., 2005). A modified version of Chapter 3, titled “Characterization of excess pore pressures at the toe of the Nankai accretionary complex, Ocean Drilling Program sites 1173, 1174, and 808: Results of one-dimensional modeling” has been accepted for publicati on by the Journal of Geophysical Research, authored by Gamage and Screaton. This chap ter contributes to th e understanding of the development of pore pressure at the toe of the Nankai accretionary complex. This study was based a one-dimensional model and uses the permeability-porosity relationship developed for Nankai hemipelagic sediments in the previous study. The sensitivity of pore pressures to bulk-permeability, lateral stresses within the prism, and a low-

PAGE 28

15 permeability barrier at the dcollement was also tested. The results of this simplified model assess parameters that significantly affect pore pressures in subduction zones. Furthermore, the results of th is study provide insight on the initiation of the dcollement zone. Chapter 4 contributes to the understa nding of the developm ent of pore pressures through space and time at the Barbados accretionary complex. This study is based on a two-dimensional model that allows traci ng the development of pore pressures as sediments subduct beneath the prism. The mo del allows examination of the effect of hydrofracture and a dcollement with vary ing permeability based on a relationship of bulk permeability -vertical effective stress. Results indicate the spatial and temporal variations of excess pore pressures and provide insight to possible mechanics for episodic fluid flow. This chapter will be adapted for submittal to the Earth Planetary Science Letters. Chapter 5 summarizes the principal findings of chapters 2 through 4.

PAGE 29

16 CHAPTER 2 A COMPARATIVE STUDY OF PERMEABILITY MEASUREMENTS FROM THE SUBDUCTION ZONES OF NORTHERN BARBADOS, COSTA RICA, NANKAI, AND PERU Introduction Marine sediments have been widely studi ed for their physical properties both in academic and industrial research. With the in troduction of the Deep Sea Drilling Project (DSDP) and the Ocean Drilling Program (ODP ), a new level of understanding has been added to the knowledge of marine sediments during the past few decades. Physical properties of submarine sediments have been studied largely thr ough recovered cores, down-hole logging, and also by in situ inst rumentation. Permeability is one such physical property that has been closely studi ed for its importance in fluid flow and pore pressures in the subsurface. Previous st udies based on permeability measurements of marine sediments have suggested that co rrelation between permeability and porosity could provide insight to a large range of sediments in nature (Bryant, 2002). Investigations based on numerical mode ling have shown that permeability is a crucial parameter in accretionary co mplex hydrology (e.g., Bekins et al., 1995; Bruckmann et al., 1997; Saffer and Bekins, 1998). According to Saffer and Bekins (1998) sediment permeability is the most im portant factor that controls modeled pore pressures because it can vary by several orde rs of magnitude. Thus, using a systematic relationship between porosity and permeability is a powerful way to approximate the permeability structure in an accretionary wedge model (Saffer and Bekins, 1998). Results from modeling studies have shown that pore pressures are highly sensitive to the

PAGE 30

17 permeability-porosity relationship (e.g., Gamage and Screaton, 2006). Prior to the availability of core samples of marine sediments, many studies extrapolated permeabilities from fine-grained terrigenous sediments found on-shore and in many cases these values produced ambiguous results (Bryant et al., 1981). With the availability of more samples, the quantity of permeability data has significantly increased. However, difficulties in laboratory measurements and finding undisturbed cores have limited the amount of permeability measurements repres entative of different lithologies and structural domains at subduction zones. The focus of this study is to synthesize available permeability data from four different subduction zones with the aim of pr edicting permeability-porosity relationships for a number of sediment types found in m odern accretionary complexes and to examine what parameters affect the relationship be tween permeability and porosity. The samples used in this study represent sediments from Northern Barbados, Costa Rica, Nankai, and Peru. Samples representative of the Nort hern Barbados and Nankai subduction zones mainly consist of fine-grained clays and silts that are commonly grouped as hemipelagics. Samples from Costa Rica cons ist of both hemipelagics and calcareous oozes while samples from Peru consisted of calcareous oozes and siliceous oozes. Using existing permeability data, permeability-porosity relationships were developed based on depositional environment, gr ain size distribution and structural domain. These relationships were compared and examined to evaluate the relative importance of each variable.

PAGE 31

18 Background Barbados The Barbados accretionary complex is located in the Caribbean where the north American Plate (Figure 2-1) is being subducte d beneath the Caribbean Plate at a rate of about 2 cm/yr in an east-west direction (D eMets et al., 1990). Active accretion of sediments at the Barbados accretionary comp lex takes place at the eastward end of the complex. The complex is partially exposed a bove sea level at Barbados Island (Figure 21). At the location of DSDP and OD P drilling, the incoming sediments are predominantly clay and claystones. Figure 2-1. A) Map of large-scale regional setting location of the drill sites at the Barbados accretionary complex. B) Cross section from the seismic depth N W

PAGE 32

19 section extending from west of Site 949 (ODP Leg 156) to Site 672 (ODP Leg 110), (Shipboard Scientific Party, 1998). This study used core permeability measurements from ODP Leg 156 Sites 948 and 949 and ODP Leg 110 Sites 671, 672 and 676. Site 948 is located 4.5 km west of the deformation front and coincides with the loca tion of Site 671 while Site 949 is located 2 km northeast of Site 948 (Shi pley et al., 1997). Site 676 wa s drilled 0.25 km arcward of the deformation front and Site 672 was drille d 6 km east of the deformation front to provide an undeformed referen ce site (Mascle et al., 1988). Costa Rica The Middle American Trench (MAT) is formed by the eastward subduction of the Cocos Plate beneath the Caribbean Plate (Figur e 2-2) at a rate of about 8.8 cm/yr (Silver et al., 2000). At Costa Rica the incoming sedimentary seque nce is about 380 m thick and consists of approximately160 m of siliceous hemipelagic sediments overlying about 220 m of pelagic carbonates (Silver et al., 2000). As indicated by drilling on DSDP Legs 67 and 84, this stratigraphy is regionally continuous between the Leg 170 transect and offshore Guatemala (Aubouin and von Huen e, 1985; Coulbourn, 1982). During ODP Leg 170, two locations penetrated the dco llement zone. Site 1043 is located 0.5 km landward of the trench and Site 1040 is lo cated 1.6 km seaward of the trench. The incoming sediments at Site 1039, which is loca ted at 1.5 km seaward of the deformation front, were also drilled during ODP Leg 170. In a more recent visit to the MAT, ODP Leg 205 drilled Sites 1253, 1254 and 1255 (Figure 2-2). Site 1253 is located 0.2 km seaward of the deformation front while Site s 1254 and 1255 are located coincident with Sites 1040 and 1043, respectively.

PAGE 33

20 Figure 2-2. A) Map showing location of OD P drilling along the Costa Rica subduction zone. B) Cross-section indicating ODP Leg 170 and 205 drilling sites used for this study (Silver et al., 2000). Nankai The Nankai accretionary complex is formed by the subduction of the Shikoku Basin on the Philippine Sea Plate beneath the southwest Japan arc on the Eurasian plate (Figure 2-3) at a rate of about 4 cm/yr (Seno et al., 1993).

PAGE 34

21 Figure 2-3. A) Location map of ODP Leg 190 (solid circles) a nd previous ODP/DSDP drill sites (solid squares) in the Nankai Trough. B) Seismic reflection profile through the Muroto Transect reference (Site 1173) and prism toe sites (Site 1174 and 808). C) Seismic reflection profile through Ashizuri Transect showing the reference Site 1177 (Shi pboard Scientific Party, 2001). During ODP Leg 190, Sites 1173 and 1174 were drilled along the Muroto Transect while Site 1177 was drilled approximately 100 km west of Muroto along the Ashizuri Transect (Figure 2-3). Site 1173 was drilled 11 km seaward of the deformation front and

PAGE 35

22 provides an undeformed reference site of the incoming sedimentary sequence (Shipboard Scientific Party, 2001). Site 1174 is located about 1.8 km landward of the deformation front and penetrates the dcoll ement within the proto-thrust zone (Figure 2-3). Site 1177 was drilled approximately 18 km seaward of th e deformation front as the reference site for the Ashizuri Transect (Shipboard Scie ntific Party, 2001). At both Sites 1173 and 1174, the hemipelagic sediments of the upper and lower Shikoku Basin are overlain by the turbidite-rich trench-wedge facies, which was not tested for permeability. At Site 1177, the trench-wedge facies was not cored. Peru The Peru accretionary complex is fo rmed by the northeastward subduction at approximately 6.1 cm/yr of the Nazca pl ate (Hampel, 2002) below the Andean continental margin along the Peru Trench (F igure 2-4). During ODP Leg 201, seven sites were drilled into a wide range of subsur face environments in both open-ocean (Sites 1225, 1226 and 1231) and ocean-margin provinces (Sites 1227 and 1230) of the eastern tropical Pacific Ocean. These subsurface e nvironments include carbonates and siliceous oozes typical of the equatorial Pacific, cl ays and nannofossil-rich oozes of the Peru Basin, biogenic and terrigenous-rich sediments of the shallow Peru shelf, and clay-rich deepwater sequences of the Peru slope (Shipboard Scientific Party, 2003).

PAGE 36

23 Figure 2-4. A) Map showing general locations of drill sites occupied during ODP Legs 138 (rectangle B) and 112 (rectangle C) at Peru subduction zone. B) Location map of equatorial Pacific primary sites. ODP site designations are in parentheses. C) Location map of Peru margin primary sites. Previous DSDP/ODP site designations are in pa rentheses (Shipboard Scientific Party, 2003). Laboratory Permeability Data Laboratory measured permeability data were provided by several sources for each of the four subduction zones. Two widely used methods for permeability measurements are through direct flow test s (e.g., falling or constant head, constant flow) and consolidation tests. Bryant et al. (1981) cited that results of calculated permeability from consolidation tests are one orde r of magnitude less than from direct measurements using direct flow methods. Because Bryant et al. (1981) observation was based on an individual direct flow method, it cannot be c onfirmed that all direct flow methods are incompatible with consolidation tests. However, to be consistent, I limited the permeability data only to those obtained from direct flow methods and excluded data

PAGE 37

24 obtained from consolidation tests. Permeability data that were reported without porosity or void ratio information were also ex cluded from this study, which examines permeability as a function of porosity. Methods of permeability measurements used in this study are briefly discussed in the following section. Where only hydraulic conductivity was provided, permeability was calculated using (Fetter, 1994) k = K /g (1) where k is intrinsic permeability [L2], K is hydraulic conductivity [L T-1], is fluid density [M L-3], g is the gravitational constant [L T-2] and is the kinematic viscosity [M L-1 T-1]. Values of fluid density and viscos ity were determined based on temperature values reported for the experiment and the sa linity of the permeant used. In cases where temperature and/or permeant used were not reported, I assumed a temperature of 25 C and a salinity (for permeant) of 35 kg/m3. Barbados Vrolijk et al. (unpubl. data, cited in Zwar t et al., 1997) used a sample from ODP Leg 156 Site 948 to measure permeability us ing a constant-head permeameter at an effective stress of 241 kPa. A sample with 2.5 cm in diameter and about 5 cm high was contained in a triaxial cell. The sample was backpressured at 350 kPa to dissolve any trapped air in the system. The permeant was saline whereas the confining fluid was oil. Bruckmann et al. (1997) measured permeability using three whole round core samples of size 6.2 cm in diameter and about 2 cm high from Leg 156, Site 949. Permeability was measured at individual load st eps using low gradient flow te sts as described in Olsen et al. (1985). Fresh water was us ed as both the permeant and the confining fluid. Maltman and others (presented in Zw art et al., 1997) measured pe rmeability of a cylindrical

PAGE 38

25 subsample from Leg 156, Site 949 using low gradie nt (~ 25 kPa), constant -rate flow tests. The sample was 3.8 cm in diameter and 7.6 cm in height. Two sets of permeability measurements were obtained. During the fi rst set, the effective stress was varied by maintaining a constant confining pressure while varying the pore pressures within the sample. In the second set, effective stress was varied by maintaining a constant pore pressure and varying the confining pressure. Taylor and Leonard (1990) used the fa lling head method to measure permeabilities of samples from Leg 110. Samples were bac kpressured to ensure complete saturation according to Lowe et al. (1964). Permeabilities were obtained at least 24 hr after the application of each new load, yielding a dist ribution of permeabilities at incremental void ratios. Although both vertical and horizont al permeabilities were measured, I only used the vertical permeability values in this study for consistency with the other samples. Permeabilities were estimated based on hydrau lic conductivities reported in Taylor and Leonard (1990). Nankai and Peru Gamage and Screaton (2003), Gamage et al., (2005), and Hays (unpublished data) used the constant flow method, which indu ces a hydraulic gradient across the sample where the measurements of the pressure difference allow determination of permeability. The samples measured by this method we re from ODP Leg 190 Sites 1173, 1174 and 1177 at the Nankai margin and ODP Leg 201 Sites 1225, 1226, 1227, 1230 and 1231 at the Peru margin. Samples had a minimum diam eter of 5.84 cm and a height that ranged from ~ 5.84 to10 cm. Samples were backpre ssured prior to flow tests. Several consolidation steps were run w ith the confining fluid pressure used for the consolidation. Permeability was measured at the end of each consolidation step. An idealized solution

PAGE 39

26 of seawater was used as the permeant while deionized water was us ed as the confining fluid in the cell. Corresponding porosities for estimated permeability were calculated using the change in volume of fluid contained in the cell after each consolidation step. Masters and Christian (1990) used constant head tests on two Peru samples from ODP Leg 112 Sites 679 and 681. Whole-round samples of 10 cm in height were backpressured to ensure complete saturation. Constant head tests were performed at different hydraulic gradients at varying stress levels. All flow tests were performed in the upward vertical direction. De-aired, filt ered seawater with a salinity of 35 kg/m3 was used as the permeant. Permeability measurements of Nankai sediments conducted by Taylor and Fisher (1993), Byrne et al. (1993), Bourlange et al. (2004) and Adatia and Maltman (2004) were not used in this study due to insufficient da ta presented and/or inconsistencies found in the methodology. For example, Byrne et al. ( 1993), Bourlange et al. (2004) and Adatia and Maltman (2004) failed to report poros ity/void ratios. Although Bourlange et al. (2004) include amounts of void ratio decrease at certain confining pressures the data did not include an initial porosity to estimate poros ities at each effective stress. Taylor and Fisher (1993) used air in permeability testi ng and also failed to backpressure their samples prior to flow tests. According to Saffer and Bekins (1998) samples that are not reconsolidated could overestimate permeabilities due to fabric expansion, and the use of air in permeability testing would further overestimate the permeability measurements. Costa Rica Saffer et al., (2000) performed constant flow tests on samples measuring 6.25 cm in diameter and 1.5 to 1.6 cm in length from ODP Leg 170, Sites1039 and 1040. Permeabilities were measured at several st ages during the sample consolidation to

PAGE 40

27 acquire permeability values at varying void ratio s. Multiple flow tests were conducted at each void ratio. McKiernan and Saffer (2005) performed flow through permeability tests on samples that were 2cm tall and 5cm in diameter of ODP Leg Sites 1253, 1254 and 1255. During flow tests, fresh wa ter was pumped into the top of the sample at a constant rate while pressure was maintained at th e cell base. The pressure difference was determined by monitoring the pressures at th e top of the cell duri ng each flow rate. Varying flow rates were used to produce va rying pressure difference across the sample. Distilled, de-aired water was used both as the permeant and confining fluid. Screaton et al. (2005) used c onstant flow permeability te sts and constant pressure difference tests on samples from ODP Le g 170 Sites 140 and ODP Leg Sites 1253 and 1255. Testing conditions were the same as de scribed in Gamage and Screaton (2003) and Gamage et al. (2005). The only exception to this method was using a constant pressure difference to induce flow through the sample rath er than applying a constant flow rate for several of the samples. Permeability-Porosity Relationship Bryant et al. (1975) and Neuzil (1994) observed that permeability of argillaceous sediments follows a log-linear relationship w ith porosity. The log linear relationship is given by log (k) = log (k0)+bn (2) where k0 is the projected permeability at zero porosity, b is a parameter describing the rate of change of the logarithm of permeability with porosity, and n is the porosity. Description of Statistical Methods The coefficient of correlation (R2) of the regression equation describes the variability of the estimates around the mean. However, it inherits the problem of small

PAGE 41

28 sample size. Thus, in such situations the de rived statistics are not necessarily the best indicator of “goodness of fit”. Examination of residuals helps to reaffirm the “goodness of fit” of the regression equa tion in conjunction with the R2 value. This examination involved plotting the residuals vs. the depende nt variable. If the residuals exhibit a random distribution and have a more or less even split above and below the zero line then it is possible to say that the equation de scribes the relationship well (Kirkup, 2002) The other test requires formulating a nu ll hypothesis and an alternate hypothesis, which are then tested using ANOVA and t-st atistic. This test helps to access the suitability of the best-fit equation that desc ribes the relationship between the variables (Kirkup, 2002). The hypothesis test utilize in this analysis is a one-tailed ANOVA. In this case the null and alternate hypotheses are, H0: the equation has a zero slope; Ha: the equation has non-zero slope. Since the linear regression equation relates porosity to permeability using the slope and the intercept of the equati on, the null hypothesis is that of zero slope, which yields a constant function. The hypothesis tests were performed at the 95% confidence interval. Results Permeability values were plotted on an outline of Neuzil’s (1994) compiled range of permeabilities as a function of porosity fo r argillaceous sediments (Figure 2-5). The majority of permeability values were envel oped within Neuzil’s (1994) plot. However, several samples from Site 1231 of Peru and Sites 1039 and 1040 of Costa Rica plotted outside Neuzil’s (1994) plot ted area. These include nonargillaceous sediments of calcareous oozes and few samples containing siliceous oozes.

PAGE 42

29 Figure 2-5. Plot of laboratory derived permeab ility measurements from Barbados, Costa Rica, Nankai, and Peru subduction zones superimposed on outline of Neuzil’s (1994) plot for argillaceous sediments. As compared to Neuzil's (1994) paper, the axes have been transposed. Effects of Depositional Environment According to Boggs (2001) there is little agreement regarding th e classification of deep-sea sediments and thus, suggested classi fications range from those that are largely genetic to those that are largely descrip tive. Unfortunately there is no single classification that take into account both genesis and descriptiv e properties of all kinds of deep-sea sediments (Boggs, 2001). Thus, here I used generalized descriptions used by Boggs (2001) to categorize hemipelagic and pe lagic sediments. Hemipelagic muds are defined as mixtures of fine-grained terri genous mud with biogenic remains that are deposited under very low current velociti es. According to Stow and Piper (1984), hemipelagic muds contain more than 5% bi ogenic remains and a terrigenous component of more than 40%. The terrigenous compone nt of the hemipelagic muds are commonly composed of fine terrigenous quartz, feldsp ar, micas, and clay minerals while biogenic

PAGE 43

30 remains include siliceous organisms such as diatoms and calcareous organisms such as foraminifers and nannofossils (Boggs, 2001). Following Boggs (2001), I subcategorized the pelagic sediments into two basic groups based on the abundant type of biogenic remains present in the sediment. Pelagic sediments may be composed mainly of clay -size particles of te rrigenous or volcanogenic origin or it may contain significant amount s of silt to sand-size planktonic biogenic remains (Boggs, 2001). Pelagic sediments that contain significant amounts of biogenic remains are called oozes. However, little agre ement exists with regards to the amount of biogenic remains required to qualify a sedi ment as an ooze (Boggs, 2001). Boggs (2001) suggested that oozes have more than tw o-thirds of biogenic components. Thus, depending on which biogenic component is domi nant, a sediment can be classified either as a calcareous or a siliceous ooze. The pelagic sediments that are predominantly of calcium carbonate tests were grouped as cal careous oozes while sediments that are predominantly of diatom tests were grouped as siliceous oozes. In situations where similar lithologic al descriptions ba sed on depositional environments are used, sedime nts from different locations were grouped together. For example, Northern Barbados, Nankai and Cost a Rica all provide samples of hemipelagic sediments. Because even slight differences in permeability-porosity relationships could affect results of fluid flow models (e.g., Gamage and Screaton, 2006), it is worthwhile comparing the permeability-porosity relationships based on depositional environment between individual locations. Based on the depositional classification desc ribed above all samples from Northern Barbados and Nankai were grouped as hemipela gic sediments. Samples from Costa Rica

PAGE 44

31 represented both hemipelagic sediments as well as calcareous oozes. Samples from Peru consisted of both calcareous and siliceous sedi ments. Unfortunately the only quantitative data available for the biogenic component of these sediments are from smear slide analyses. As a result it was difficult to de termine a representative percentage of the biogenic component. The nannofossil rich sa mples were grouped as calcareous oozes. The most problematic was to classify th e diatom-rich sediments. Lithological descriptions and smear slide analysis favored them in the “siliceous ooze” category, and quantitative data on the components were lackin g. However, it should be noted that these samples could contain considerable amounts of biogenic and terrig enous components and may fall in between hemipelagic and pelagic sediments. Peru samples from Site 1227 were rich in organic carbon and did not fit either of the pelagic sediment categories identified in this study. Thus they were excluded from this study. It should be noted that the sa mples from Sites 1225 and 1226 represents the sediments of the equatorial Pacific where th e direction of plate movement carries them away from the trench. Thus, these two site s do not represent the se diments of the Peru subduction zone. However, they consist dom inantly of calcareous oozes, which can be compared to the calcareous oozes of Costa Rica, and for that purpose were included in this study. The two main lithological groups of hemipelagics and pelagics were plotted separately in Figures 2-6. Based on the calculated R2 value for the Barbados hemipelagics, the equation explains only 20% of the correlation be tween permeability and porosity. Although the analysis of residuals showed randomness, the hypothesis testi ng did not support the existence of a statistically significant correlation between permeability and porosity for

PAGE 45

32 the Barbados hemipelagics. When six of the data points, which were obtained from Taylor and Leonard (1990), were removed from the regression, the R2 value of the remaining 19 samples increased from 0.20 to 0.50, predicting a permeability-porosity relationship of log (k)=-24.24 + 11.31n. Although statistically this makes the six data points outliers, a closer look at the lithological descriptions, grain size data and CaCO3 percentages suggest that the hemipelagics samples used by Taylor and Leonard (1990) are likely to be diffe rent from the rest of the hemipe lagics sediments representative of Barbados because samples used by Taylor and Leonard (1990) repr esented calcareous muds whereas other samples represented claystones. Even though many of the samples from Ba rbados, Nankai, and Costa Rica fall in the depositional classification of “hemipelag ics”, they are not well represented by the same permeability-porosity re lationship (log (k ) = -19.91+4.9n, R2 = 0.5). The Barbados permeability-porosity relationship predicts similar values of permeability to Nankai and Costa Rica relationships at por osities between 0.55-0.70 (Table 2-1). However, at lower porosities the Barbados permeability relationshi p predicts lower values than those of Costa Rica and Nankai. Because Barbados permeabilities are constrained by a small range of porosities, one should be cautious out side the porosity range of the laboratory results. The log-linear rela tionships for Nankai and Costa Rica plot roughly parallel to each other with slightly lowe r permeabilities at Costa Rica for a given porosity than at Nankai (Figure 2-6). The R2 value obtained for Nankai was 0.79 while for Costa Rica it was 0.70 suggesting reasonable correlation be tween permeability and porosity at these two locations. The R2 value obtained for the log linear relationships for the calcareous oozes were less than 0.5 while for the siliceous oozes it is greater th an 0.5 (Figure 2-6).

PAGE 46

33 The calcareous oozes of Costa Rica plot approxi mately in the same permeability range as those from Peru, near the upper bounda ry of Neuzil’s (1994) plot. Table 2-1. Log linear permeability-porosity relationships predicted for varying depositional environments at Barb ados, Costa Rica, Nankai and Peru. Location and Depositional environment Permeability-porosity relationship Barbados hemipelagics log (k) = -22.02 + 8.25n (R2 = 0.20) Costa Rica hemipelagics log (k) = -20.84 + 6.27n (R2 = 0.70) Nankai hemipelagics log (k) = -19.80 + 5.37n (R2 = 0.79) Costa Rica calcareous oozes log (k) = -18.09 + 4.83n (R2 = 0.33) Peru calcareous oozes log (k) = -20.87 + 7.79n (R2 = 0.40) Peru siliceous oozes log (k) = -18.64 + 3.55n (R2 = 0.67) Figure 2-6. Permeabilities classified based on depositional environment and location. A) Predicted log-linear permeability-porosity relationships for hemipelagic samples. B) Permeabilities and predicted log-linear permeability-porosity relationships for pelagic samples. Peru calcareous oozes lie at a higher por osity range (0.55-0.80) while Costa Rica calcareous oozes represent porosities between 0.45 and 0.70. The Peru siliceous oozes plot between the lower permeability values of calcareous oozes and the higher permeability values of hemipelagic sedi ments for porosities ranging from 0.45-0.8. Although the predicted R2 values were fairly low for th e two calcareous oozes, a linear relationship between the logarithm of perm eability and porosity for those two groups was

PAGE 47

34 not rejected based on the re sidual distribution and hypothe sis testing. Relative to relationships predicted for hemipelagic sedime nts, the permeability-porosity relationship predicted for the siliceous oozes show simila r permeabilities at porosities >0.7 and higher permeability values at porosities 0.7. Effects of Grain Size Bryant et al. (1981) st ated that grain size is the most important characteristic of a sediment that determines permeability as it a ffects the porosity of the sediment. This contention has been supported by studies su ch as by Koltermann and Gorelick (1995), in which grain size distribution was used to predict permeability and porosity in various sediment mixtures. Ninety percent of thei r predicted hydraulic conductivities matched the hydraulic conductivities estimated from fi eld tests within one order of magnitude (Koltermann and Gorelick, 1995). Accordingl y, I categorized the sediment samples in this study based on a grain size classification de scribed by Bryant (2002) In contrast to Bryant's (2002) study, which exclusively used sediments from the Gulf of Mexico, I combined samples from all four subduction zo nes due to limited va riation in grain size data found at any single location. Bryant (2002) noted that fine-grained marine sediments with low amounts of carbonate did not affect the permeability-porosity relationship, but did not specify what percent of carbonate was considered. Based on our calcareous pelagic samples I used 45 CaCO3 wt% as the limit between the high and lo w carbonate content. The 45 wt% was used because majority of the samples cont ained either distinctively high (>> 45 wt% CaCO3) or low (<<45 wt% CaCO3) amounts of CaCO3. For samples that did not have CaCO3 wt% data, I used information from othe r methods such as X-ray diffraction and

PAGE 48

35 inorganic carbon percentages to infer whether samples were like ly to contain greater than 45 wt% of carbonates. Although grain size data were available for samples from ODP Leg 112 Sites 679 and 681 (Peru), the scale used fo r the particle size cr iteria (Masters and Christian, 1990) was different from the rest of the grain size data used in this study. Thus, I excluded grain size data from Sites 679 and 681 in this part of the study. All samples used in the grain size classification (marked with an asterisk) and values of available weight percentages of CaCO3 are tabulated in Appendix A. All samples were categorized into the following four groups based on their grain size distribution: Group 1, sediment containing more than 80% clay size material. Group 2, sediment containing 60 -80% clay-size material. Group 3, sediment-containing silty-clays with less than 60% clay and less than 5% sand. Group 4, sediment-containing sandy-silts with less than 60% clay and more than 5% sand. The following particle size criteria were used for classification: sand (>63 m), silt (63-4 m) and clay (<4 m). The log linear relationships for each of the groups obtained by least squares regression fit are given in Table 2-2. Group 4 was excluded from the grain size classification as it only contained a tota l of four samples. Table 2-2. Permeability-porosity relationships based on grain size analyses. Group Description Permeability-porosity relationship Permeability-porosity relationship for Gulf of Mexico (Bryant 2002)* 1 > 80% clay log (k) = -24.28 + 11.32n (R2 = 0.53) log (k) = -20.9 + 6.54n 2 60-80% clay log (k) = -19.73 + 4.49n (R2 = 0.56) log (k) = -20.53 + 6.16n 3 Silty-clays with <60% clay and <5% sand log (k) = -19.91 + 5.45n (R2 = 0.43) log (k) = -20.59 + 6.77n

PAGE 49

36 Bryant (2002) used hydraulic conductivity in stead of intrinsic permeability. Thus, the hydraulic conductivities were converted to permeability (Table 2-2) using a viscosity of 0.000966 Pas and density of 1023 kg/m3 at a temperature of 25C and a salinity of 35 kg/m3. Based on the grain size classification, Gr oup 1 consists only of hemipelagic samples that are representative of Barbados Group 2 consists of hemipelagic samples from Barbados, Nankai and Costa Rica while Group 3 consists of Nankai and Costa Rica hemipelagics. In general the permeability-por osity relationship obtai ned from grain size classification suggests an increase in permeabilit y with a decrease in cl ay size particles. The R2 values predicted from the permeability-porosity relationships show R2 > 0.5 for Groups 1 and 2 and 0.48 for Group 3 (Figure 27). It should be noted that the 95% confidence intervals for the grain size groupi ngs show an overlap, suggesting that the permeability-porosity relationships are not statistically significantly different from each other. This overlap is probably caused by th e difficulties in permeability measurements, and the likelihood that sediments with slight ly varying grain size percentage (e.g., 79% clay versus 81% clay) may not have distingu ishable values of permeability, despite being classified in different categories. Although the R2 values predicted from grain size classification were higher than the ones predicted by depositional environment only, it predicted lower R2 values than those predicted fr om classification based on location except for Barbados. This may suggest that varying deformational processes at different locations may affect sediment pr operties at varying degrees.

PAGE 50

37 Figure 2-7. Permeabilities classified based on grain size distribution. Solid lines represent permeability-porosity relationship predicted for samples used in this study. Dashed line represents permeabil ity-porosity relationships of Bryant (2002). In order to understand the general trend of permeabil ity-porosity relationships based on grain size distribution, I compared my data with that of Br yant’s (2002) Gulf of Mexico data, which were obtained through cons olidation tests. The dashed lines in Figure 2-7 represent the relations hips of Bryant (2002) and so lid lines represent predicted relationships from this study. For Group 1, the predicted permeability from this study matched the permeabilities of Bryant’s (2002) at porosities around 0.6 to 0.7. As

PAGE 51

38 porosities decrease our relationship for Group 1 predicted lower values of permeability than Bryant’s (2002) permeability-porosity rela tionship. The large discrepancy between Bryant’s (2002) and this st udy’s relationship in Group 1 mi ght be caused by the narrow range of porosities (0.50-0.70) represented by the Group 1 data. In contrast Groups 2 and 3 represent a larger range of porosities between approximately 0.25-0.70. Group 3 shows fairly similar permeability-porosity relationshi ps to those of Bryant ’s (2002) while our Group 2 relationship crosses Bryant’s (2002) permeability-porosity relationship at a porosity of 0.5. Effects of Structural Domain Samples that were used for the grain size analyses were further categorized based on the structural domains of each sample to test the effect of deformation on the permeability-porosity relationship. Samples th at represented the underthrust sediments and the incoming sediments at reference sites were grouped together as they represented the undeformed or minimally deformed se diments of the subduction complex. Prism sediments and sediments that represented the dcollement zone were grouped together as these samples generally are highly deformed dur ing the accretionary pr ocess. Only grain size Group 2 of our samples had a considerable amount of samples that represent both the underthrust/incoming sediments and the prism/ dcollement sediments. Using the data from Group 2, I fitted linear regression lines for both the underthrust/incoming sediment and the prism/dcollement sediment groups (Figure 2-8).

PAGE 52

39 Figure 2-8. Permeabilities classified based on structural domain. Permeability-porosity relationships only shown for underthrust/incoming samples (dashed line) and prism/dcollement samples (solid line) of Group 2. The prism/dcollement sediments group onl y contained eight samples and thus, gave a relatively high R2 value of 0.91. The underthrust/incoming sediment group contained twenty-two samples and showed moderate correlation between permeability and porosity with a R2 value of 0.64. Even with the limited data available, the two relationships exhibited similar permeabilities at lower porosity values (0.2-0.3). With increasing porosities the two relationships pr edicted permeabilities that diverge from each other. For example, the permeability-porosity relationship of prism/dcollement group predicted a permeability value of 1 x 10-16 m2 at a porosity value of 0.50 while at the same porosity level the underthrust/incomi ng group predicted a lower permeability value of 1 x 10-18 m2. The differences exhibited in these two permeability-porosity relationships, while inconclusive suggest that it might be useful to study more samples representing each group, particularly the pr ism/dcollement group, to test the predicted relationships.

PAGE 53

40 Discussion The hemipelagic samples used in this study were well confined w ithin the limits of Neuzil’s (1994) plot of argi llaceous sediments. Although hemipelagic samples plotted in the same region, the predicted permeability-por osity relationships va ried with location. In the case of Barbados, the permeability-p orosity relationship showed scatter even within an individual location due to varia tions found in the hemipelagic sediments. These varying relationships suggest that sa mples categorized as “hemipelagics” could have different permeability-porosity relationshi ps at different locations. Thus, combined classification of both depositional environment and location may provide better correlation between permeability and porosity. In general the observed differences in the log permeability-porosity relationship appears correlated to the amount of clay a nd silt size particles present in the sample. Except for Group 1, our predicted permeability-por osity relationships based on grain size distribution are in good agreement with the rela tionships of Bryant (2002). Groups 2 and 3 exhibited similar trends as those predicted by Bryant’s (2002). This similarity suggests that even though samples were taken from diff erent subduction zones, samples classified based on their grain size distri bution exhibit similar trends between permeability and porosity compared to those samples that were taken from a single location (e.g., Bryant’s (2002) samples from Gulf of Mexico). Thus it could be concluded that the permeabilityporosity relationships obtained based on grain size distribution are generally applicable to samples from marine settings. However, ad ditional data, particularly in Group 1, would further test this conclusion. At high porosities, the relationship pred icted for the siliceous oozes suggests similar permeabilities to those predicted for he mipelagic sediments. However, there were

PAGE 54

41 no samples that represent lower porosities (l ess than 0.45) of the siliceous oozes and therefore the predicted relati onship should be used with caution at lower porosities. The lack of correlation exhibited between pe rmeability and porosity in calcareous oozes suggest that one should consider other variables such as depth, consolidation rates and relative age of the sediment to obtain mean ingful relationships for pelagic sediments. According to Bryant et al. ( 1981), unlike grain size, the influence of calcium carbonate is more pronounced with increasing depth becau se as burial increases less reduction of porosity is observed for calcareous sediment s compared to non-calcareous sediments. Mechanical compression tests conducted by Terzaghi (1940) a nd Robertson (1967) demonstrated that calcareous muds compact less than non-calcareous muds. A similar study by Bryant et al. (1981) demonstrated th at under a similar load, carbonate sediments do not consolidate to as low a void ratio as non-carbonates. Based on this finding they speculated that the resistance towards consolid ation in carbonates coul d be a result of the relative age of the sediment or the differences in particle shap e or the structural strength of the individual particles. Even though comparison of underthrust/incoming and prism/dcollement structural domains from Group 2 suggested the possibi lity of different varying permeabilityporosity relationships, it is r ecommended to further investigate these relationships using more samples; the results presented here are based on a limited number of samples especially those represented by the prism/dcollement group. It would be worthwhile to further test the effects of structural domain on permeability-porosity relationship, as this information will allow future studies based on permeability-porosity relationships to more realistically represent fluid flow.

PAGE 55

42 This study used a large number of data fr om many different sources to investigate the relationship between permeability and por osity. Assembling data for this study highlighted the importance of documenting de tailed methodology including critical data such as the temperature during permeability tes ting and the type of permeant used in the flow test. It is also im portant to have porosity and grain size data available for documented permeability values, as these parameters are valuable estimating relationships between permeability and porosit y. This study also identified gaps in available data. For example, only a few se diment samples were available from the structural domain representing the prism/d collement and few samples from Barbados were tested at low porosities. Thus, future data collections sh ould focus on collecting samples representing the prism and the dcoll ement as well as deeper parts of the underthrust sediments, where much remains to be learned. Conclusions I examined permeability-porosity relationships for sediments from four different subduction zones based on their depositional environment, grain size distribution, and structural domain. Greater correlation was observed between permeability and porosity for hemipelagic sediments and for siliceous oo zes while relatively low correlations were predicted for calcareous oozes. Based on the he mipelagic samples used in this study, it is clear that permeability-porosity relationships vary among hemipelagics at different locations and thus, classification based on de positional environment should be used with caution when applied at different locations Grain size predicts more meaningful correlation between permeability and porosity than depositional environment only, and these relationships are generally consistent with results from other marine settings. However, grain size classification shows le ss correlation between permeability and

PAGE 56

43 porosity relative to the relations hips predicted by location. Due to lack of data, the effect of structural domain on permeability-porosity relationship could not be evaluated. To predict meaningful relationships for permeab ility of carbonaceous sediments, one should consider other factors such as depth, consolidation rates, and re lative age of the sediment.

PAGE 57

44 CHAPTER 3 CHARACTERIZATION OF EXCESS PORE PRESSURES AT THE TOE OF THE NANKAI ACCRETIONARY COMPLEX, OC EAN DRILLING PROGRAM SITES 1173, 1174, AND 808: RESULTS OF ONE-DIMENSIONAL MODELING Introduction Examining the fluid flow of the deep hydrosphere is extremely important because fluid flow alters the physical and chemical properties of the Earth’s crust, affecting the ocean and the atmospheric chemistry that is vital for human existence (COMPLEX, 1999). At active plate margins, fluid pressu res can influence movement along faults and thus, the nature of the earthquake cycle (Moo re and Vrolijk, 1992). Pore fluid pressures build up in areas where sediment permeabilities are low enough to prevent pore fluid escape at a rate comparable to the rate of loading due to tect onic and gravitational stresses. These elevated fluid pore pressures play an important role in the development of accretionary complexes. It is speculated that high pore pressures control the formation of the dcollement, which separates highl y deformed overlying wedge sediments from slightly deformed underthrust sediments (Westbrook and Smith, 1983). Furthermore, pore pressures affect deformation within a nd the taper angle of the accretionary wedge (Davis et al., 1983). At the Nankai accretionary complex, a thic k terrigenous sequence rapidly deposits over a low-permeability hemipelagic sequence. Previous studies have used indirect estimates to document the development of pore pressures within the underthrust sediments at the toe of the Nankai accretiona ry complex. For example, Screaton et al. (2002) estimated depth-averaged excess pore pressures using porosity -depth profiles.

PAGE 58

45 The estimated overpressures within the underthrust sediments suggest insufficient permeability for fluid escape at a rate compar able to sediment loading. Saffer (2003) used shipboard observations of porosity and la boratory measurements of consolidation to estimate pore pressures and eval uate their variations down s ection. His results indicate undrained conditions in the underthrust sedime nts that may have been caused by rapid sedimentation and loading due to underth rusting. A modeling study conducted by Le Pichon and Henry (1992) suggested that trench sedimentation and ra pid accretion at the toe of the prism could generate excess pore pr essures in the underthr ust sediments at and even seaward of the deformation front. However, that study did not have permeability data, and used estimates based on lithologies. In this study, I used laboratory permeabil ity data to investig ate porosity reduction and the generation of excess pore pressures at the toe of the Nankai accretionary complex. Based on measured permeabilities I developed a permeability-porosity relationship for hemipelagic sediments at the Na nkai accretionary prism. I then used this permeability-porosity relationship in a one-dimensional numerical model to simulate the excess pore pressures and porosit ies that would result from sedimentation and loading by the prism at the toe of the accretionary comple x. Finally I tested the sensitivity of excess pore pressures and porosities to bulk permeability lateral stresses within the prism, and a hypothetical low-permeability barrier at the dcollement. Background Geologic Setting The Nankai accretionary complex is formed by the subduction of the Shikoku Basin on the Philippine Sea Plate beneath the southwest Japan arc on the Eurasian plate (Figure 3-1) at a rate of 4 cm/yr (Seno et al., 1993). This study focused on the Muroto

PAGE 59

46 Transect (Figure 3-1) where the thickness of th e complex varies from 750 m at the toe to ~4500 m at 50 km arcward. Along the Muroto Transect, the prism toe has a taper angle of 4-5 (Shipboard Scientific Party, 2001). Based on this low taper angle, it has been inferred that the Muroto Transect has high dcollement pore pressures or low intrinsic dcollement strength (Saffer and Bekins, 2002). During ODP Leg 131, Site 808 was drilled on the Muroto Transect approximately 3 km landward of the deformation front. A 560 m thick sequence of turbidites was found above the hemipelagic muds of the uppe r and lower Shikoku Basin facies. The dcollement zone was identified by intense brittle deformation at 945 to 964 meters below seafloor (mbsf), and develops from a homogeneous interval of hemipelagic mudstones within the lower Shikoku Basin f acies (Shipboard Scie ntific Party, 2001). During ODP Leg 190, Sites 1173 and 1174 were drilled seaward of Site 808 along the Muroto Transect (Figure 3-2). Figure 3-1. Location map of the study area in the Nankai accretionary complex and sites used for this study (modified from Moore et al., 2001).

PAGE 60

47 Site 1174 penetrated the dco llement within the proto-th rust zone (Figure 3-2). The dcollement was observed between 808 and 840 mbsf and was marked by fractures and brecciation in the lower Shikoku Basin f acies (Shipboard Scie ntific Party, 2001). The turbidites extended 431 m above the hemipelagic muds of the upper and lower Shikoku Basin facies (Shipboard Scientific Party, 2001). Figure 3-2. Schematic interpretation of the Muroto Transect showing tectonic domains and location of Leg 190 drill sites used in this study (modified from Moore et al., 2001). Site 1173 was drilled 11 km seaward of the deformation front providing an undeformed reference site of the incoming sedimentary sequence (Shipboard Scientific Party, 2001). The turbidite laye r at Site 1173 is much thinne r (~102 m) than at Site 1174, because it is farther away from the trench. At Site 1173, the age equivalent of the Site 1174 dcollement zone occurs between 390 and 420 mbsf, within the lower Shikoku Basin facies (Shipboard Sc ientific Party, 2001). At the Nankai accretionary complex, heat flow values ranging from 180 mW/m2 at Sites 1173 and 1174 to 130 mW/m2 at Site 808 have been estimated (Shipboard Scientific Party, 2001). These measured high heat flow values are related to the fossil spreading ridge represented by the Kinan Seamount on the Philippine Sea Plate which ceased spreading ~ 15 Ma ago (Shipboard Scientif ic Party, 2001). Previous studies have

PAGE 61

48 suggested that high temperatures at the Muroto Transect could affect in situ dehydration reactions such as the smectite to illite tran sition (Kastner et al., 1993) as well as observed cementation by authigenic clays (Ujiie et al., 2003). Previous Hydrologic Studies Previous workers have examined several aspects of fluid flow and pore pressure development at the Nankai accretionary comp lex. Le Pichon and Henry (1992) used a one-dimensional model of sedimentation repres enting the stratigraphic sequence at Site 808. Their model consisted of a coarser te rrigenous sediment layer with permeability values of 10-16 to 10-17 m2, rapidly depositing over a less permeable hemipelagic sequence (10-19 to 10-17 m2). Because permeability data for Nankai sediments was not available, they used permeabilities that are representative of uncompacted terrigenous and hemipelagic sediments. Their modeling results showed that this stratigraphic succession could potentially form a low mechanical re sistance horizon within the upper portion of the low permeability sediments, but testing of their model requires permeability data. Taylor and Fisher (1993) performed permeability measurements on subsamples of Site 808 cores using two methods. One me thod directly measured flow under known head values and the other method indirectly m easured flow from the rate of consolidation due to a known applied stress. Their measured permeabilities ranged from 10-14 to 10-19 m2 in horizontal and vertical directions. Saffer and Bekins (1998) used a log-lin ear permeability-porosity relationship and an assumed porosity distribution in a twodimensional numerical model to match pore pressures estimated from the critical taper theo ry of Davis et al. (1983) The critical taper is the shape for which the wedge is on the verge of failure unde r horizontal compression and to maintain a small critical taper it re quires high basal pore pressures (Davis et al.

PAGE 62

49 1983). For varying assumptions about permeabil ity and porosity dist ribution within the underthrust sediments, Saffer and Bekins (1998) simulated values between 0.2-0.4 in the region of Sites 1174 and 808. It should be noted that these simulations assumed a constant dcollement permeability of 10-16 m2. The estimated permeability-porosity relationships of Saffer and Bekins (1998) fall within the range for argillace ous rocks defined by Neuzil ( 1994), but was approximately two orders of magnitude lower than the direct permeability measurements given by Taylor and Fisher (1993). Saffer and Bekins (1998) suggested that direct permeability measurements by Taylor and Fi sher (1993) were overestimates due to fabric expansion, because these measurements were performed on samples that were not reconsolidated under pressure. Moreover, Saffer and Bekins (1998) argued that the use of air in the Taylor and Fisher (1993) permeability te sts would tend to further overestimate permeabilities. Screaton et al. (2002) used porosity measurements of core samples from Sites 808 (Shipboard Scientific Party, 1991), 1174, and 1173 (Shipboard Scientif ic Party, 2001) to estimate pore pressures in the underthrust sedime nts. In this study, Screaton et al. (2002) assumed that the solid volume is constant thus relating the change in volume to change in porosity. Site 1173 was assumed to provide a reasonable proxy for conditions of the Sites 1174 and 808 sediments when they were seaward of the trench. The average porosities of the underthrust (or equivalent) sediments of the lower Shikoku Basin facies decrease landward from 0.42 at Site 1173 to 0.34 and 0.33 at Sites 1174 and 808, respectively (Screaton et. al ., 2002). The porosity profile from Site 1173 shows steadily decreasing porosity with incr easing depth (Figure 3-3).

PAGE 63

50 Figure 3-3. Porosity profiles of Sites 808, 1174, and 1173. Light shading shows lower Shikoku Basin facies. Darker shading represents the dcollement zone at Sites 808 and 1174 and at Site 1173 the age-equivalent level of Site 1174 dcollement zone. At Sites 1174 and 808 the poros ities below the dcollement increase relative to the porosities above. Comparison between the porosity profiles of the underthrust sediments from Sites 1174 and 808 with the reference site (Site 1173) suggest excess pore pressure ratio of 0.42 at Site 1174 and 0.47 at Site 808 (Screaton et al., 2002), where represents the magnitude of excess fluid pressure relative to sediment overburden pressure and is defined as (P Ph)/(Pl Ph) (1) where P is pore fluid pressure [M L-1T-2], Ph is hydrostatic pressure [M L-1T-2], and Pl is lithostatic pressure [M L-1T-2]. The excess pore pressure ratio removes the effect of the overlying water column, and describes excess fluid pressures with in the sedimentary sequence relative to the overburden load minus hydrostatic pressure. For lithostatic pore pressures, 1, whereas at hydrostatic pressures, 0.

PAGE 64

51 Saffer (2003) integrated measurements of logging-while-dril ling (LWD), physical properties data, and laboratory consolidation tests to comp are pore pressure development and progressive dewatering in the undert hrust sediments. Calculations of based on Saffer’s (2003) results indicat e values of 0.30 and 0.44 at 15 m below the dcollement at Sites 1174 and 808, respectively. Th ese values are slig htly lower than the average values over the entire underthrust profile inferred fr om porosity data by Sc reaton et al. (2002). The estimated values calculated from results of Sa ffer (2003) generally decrease with depth below the dcollement at both Sites 1174 and 808. Evidence concerning possible la teral fluid flow is ambiguous. Broad low chloride anomalies were observed within the lower Shikoku Basin facies (Kastner et al., 1993; Gieskes et al., 1993; Spivack et al., 2002). C ontroversial ideas have been suggested as possible explanations for the observed low chlo ride anomalies. Kastner et al. (1993) and Underwood et al. (1993) suggested that sm ectite dehydration is the most likely mechanism for the observed low-chloride anomaly. On the basis of geochemical analyses, Kastner et al (1993) suggested late ral fluid flow along th e dcollement or along a deeper conduit within the underthrust sedime nts while Spivack et al. (2002) suggested continuous lateral fluid flow between Sites 808 and 1174 within the underthrust sediments. Brown et al. (2001) suggested that in situ dehydra tion of ~10-15 wt% of smectite could generate most of the freshe ning observed at Site 808 and therefore, provides a possible explanation for pore fluid freshening. However, if the initial amount of smectite is less than 10-15 wt% then the low-chloride anomaly could reflect the combined effects of both in situ and lateral fl ow at depth (Brown et al., 2001). This in situ dehydration hypothesis has been further supported by a recent study by Henry et al.

PAGE 65

52 (2003) using cation exchange capacity. Henr y et al. (2003) show ed that there are sufficient amounts of smectite to explain the ch lorinity anomalies by in situ reactions, and thus, lateral fluid flow is not required. Although fluid fr om dehydration greatly affects pore water chemistry, Saffer and Bekins ( 1998) suggest that fluid sources from dehydration of smectite are 10-1000 times smalle r than the compaction fluid sources. Thus, dehydration fluid sources are unlikely to be significant in generating pore fluid pressures at the toe of the prism. Laboratory Permeability Measurements Although previous studies have provided ev idence for excess pore pressures at the Nankai accretionary complex, the causes of it were not sufficiently characterized due to lack of well-constrained data in sediment properties. Because sediment permeability is the most important factor in controlling modeled pore pressures, it is important to establish a systematic relationship between permeability and porosity to approximate permeability in the accretionary complex (Sa ffer and Bekins, 1998). Thus, as the first step of the study I used laboratory measured permeability data from nine core samples at varying depths to establish a permeability-p orosity relationship. Gamage and Screaton (2003) provide detailed information about test methods. Vertical permeability tests were performed using the constant flow method on nine core samples taken from the Shikoku Basin facies at Sites 1173 and 1174. The cons tant flow permeability method induces a hydraulic gradient across the sample, and meas urement of the pressure difference allows determinations of permeability. To determ ine porosities for the permeabilities presented by Gamage and Screaton (2003), I calculated change in porosity during consolidation based on the change in volume of fluid [L3] contained in the cell at the end of each

PAGE 66

53 consolidation step, and used shipboard poros ity measurements for the initial porosities (Table 3-1). Table 3-1. A summary of laboratory measured permeabilities for samples from ODP Leg 190 Sites 1173 and 1174. Permeability valu es were recalculated (as compared to Gamage and Screaton, 2003) based on a viscosity value of 0.0008 kg/s.m and a density of 1020 kg/m3, based on laboratory temp eratures during testing (Lide, 2000). Values for Sample 1174B-84R-3, which are included in Gamage and Screaton (2003) are not presen ted here due to insufficient data to calculate porosities at measured permeability values. Sample Effective Stress (MPa) Porosity Permeability (m2) 1173A-22H-2, 199.9 mbsf, Silty clay 0.24 0.57 5.14 x 10-17 0.42 0.56 4.02 x 10-17 0.54 0.55 3.90 x 10-17 1173A-31X-1, 284.59 mbsf, Silty claystone to 0.27 0.62 1.98 x 10-17 0.42 0.60 1.28 x 10-17 1173A-39X-5, 367.07 mbsf, Silty claystone, 0.26 0.41 2.46 x 10-18 0.39 0.36 2.18 x 10-18 0.83 0.33 1.51 x 10-18 1173A-41X-cc, 388.75 mbsf, Silty claystone, 0.29 0.45 2.48 x 10-18 0.41 0.43 1.66 x 10-18 0.55 0.43 1.18 x 10-18 1173A-46X-1, 428.59 mbsf, Silty claystone 0.25 0.45 1.90 x 10-18 0.40 0.43 1.52 x 10-18 0.51 0.41 1.24 x 10-18 1174B-42R-3, 538.23 mbsf, Silty claystone, altered 0.19 0.37 7.84 x 10-18 0.48 0.34 2.00 x 10-18 0.62 0.32 1.52 x 10-18 1174B-59R-5, 704.95 mbsf, Silty claystone 0.55 0.32 5.18 x 10-19 0.69 0.30 3.53 x 10-19 0.83 0.28 2.94 x 10-19 1174B-69R-2, 795.17 mbsf, Silty claystone, 0.55 0.29 8.48 x 10-19 0.69 0.28 5.78 x 10-19 0.83 0.26 2.83 x 10-19 1174B-74R-1, 842.75 mbsf, Silty claystone-0.55 0.30 6.40 x 10-19 0.69 0.28 6.16 x 10-19 0.83 0.27 4.70 x 10-19 For clay-rich sediments, Bryant et al. (1975) and Neuzil (1994) established that logarithm (base 10) of permeability (k) decreases systematically with decreasing porosity (n) and thus, k can be represented as a function of porosity,

PAGE 67

54 log (k) = log (k0)+n (2) where k0 is the projected permeability at zero porosity and is a parameter describing the rate of permeability ch ange with porosity. With respect to Neuzil’s (1994) compila tion of permeability data for argillaceous sediments, most of the Nankai core permeability values fall in the central portion of the data range (Figure 3-4), while a few permeab ility values with porosities between 0.2 and 0.3 fall closer to the higher-permeability ma rgin of the data range. The measured permeability values are best fit by the following relationship: log (k) = -19.82+n (3) Figure 3-4. Permeability data measured for sa mples from ODP Leg 190 (solid diamonds) superimposed on a large-scale permeability versus porosity outline for argillaceous sediments compiled by Neuzil (1994). Solid line represents the best-fit line for the measured permeabili ties. The dashed line with + mark represents the permeability-porosity relationship for the lower range of measured permeability values. The shaded area represents the range of permeability-porosity relationships established by Saffer and Bekins (1998), with the dashed line (log (k)=-20+ 5.25n) indicating a mid-range value. The permeability-porosity relationship obtaine d from this study has a similar slope to the relationships used by Saffer and Bekins (1998) (log (k) = 20+5.25 n) but predicts slightly higher permeability values (Figure 34). The lower range of permeability values

PAGE 68

55 are near the bottom of the range used by Sa ffer and Bekins (1998) relationship, and can be represented by a line with the relationship, log (k) = -20.15 + 5.39n (Figure 3-4). Modeling Methods Theoretical Background As sediments are loaded, the sediment column beneath will compact if the pore fluid can escape at a rate comparable to the loading rate. However, if the permeability of the sediment column is low, dewatering will be inhibited, excess pore pressures will build, and compaction will be prevented. Darcy’s Law expresses the relationship between pore pressures, sediment properties, and fluid velocities. For constant density saturated flow, Darcy’s Law can be written in terms of the hydraulic head (Voss, 1984). h n K v (4a) where z fh g P h (4b) where v is pore fluid velocity [L T-1], K is hydraulic conductivity [L T-1], n is porosity, h is hydraulic head [L], f is fluid density [M L-3], g is gravitational acceleration [L T-2], and hz is elevation head [L]. Hydr aulic conductivity is defined by K=kf g/, where k is intrinsic permeability [L2] and is dynamic viscosity [M L-1 T-1]. Intrinsic permeability is a function of the porous medium while fluid density and viscosity depend on the properties of the fluid and may change with te mperature, salinity, a nd to a lesser extent with pressure. Combining the mass conservation of fluid with Darcy’s law (Eq. 4a), the following equation can be written for diffusion of head in porous media for saturated fluid flow. Q t h S h Ks ) ( (5)

PAGE 69

56 where Ss is specific storage [L-1], and t is time [T]. Specific storage is a measure of the volume of formation fluid per unit volume of media per unit head of a saturated formation that is stored or e xpelled from storage due to comp ressibility of the matrix and change in fluid pressure. Specific storage is defined as Ss=f g( +n ) where is matrix compressibility [M-1 L T2], and is the compressibility of the formation fluid [M-1 L T2]. Matrix compressibility ( ) is defined by = [1/(V)](dV /d e) where V is the bulk volume [L3] and e is the effective stress [ML-1T-2]. The second term on the right hand side is a source term representing changes to fluid volume or pressure, such as due to loading, per time per unit volume of media [T-1]. During loading, the stress added from the sediment load is partitioned between the pore fluid and the matrix. The loading efficiency, denotes the fraction of the stress added to the pore fluid pressu re and, assuming the sediment grains are incompressible relative to the matrix, is defined as follows (after Neuzil, 1986) ( n) (6) For highly compressible sediments, the loading efficiency is near 1. The increase in pore pressure caused by lo ading drives fluid flow. When initial conditions, boundary conditions and the controlling hydrogeologic parameters k and are known, Equation 5 can be solved for hydrau lic head at any point at any time, and thus, pore fluid velocity relative to the se diment framework can be calculated using Equation 4a. The pore pressure is related to effective stress ( e) by e = t – P (7)

PAGE 70

57 where t is the total stress. As fluid escapes, pore pressure decreases and effective stress increases, causing porosity to decrease. For sediments that are normally compacting (sediments at hydrostatic pore pressure), porosity has been observed to decrease exponentially with depth, as suggested by Athy (1930) n = noexp[bz ] (8) where no is initial porosity, b is a constant [L-1], and z is burial depth [L]. For hydrostatic conditions, the change in effective stress, e, with depth is given by (de / d z ) = (s f) (1n ) g (9) where s is grain density. Combining Equations 8 and 9 results in (dn / de ) = (bn) / [(s f ) (1 n) g] (10) Equation 10 is only applicable for sedi ments undergoing compaction. For this investigation, it is assume d that sediments cannot decompact or expand when the effective stress is reduced. Thus, porosities will remain constant unless the effective stress exceeds the previous maximum effective stress value. The change in volume (d V ) can be related to (d n ) by d V / V = (d n ) / (1 – n ) (11) In one dimension, the volume change re presents only the change in vertical thickness and thus the horizontal dimension of the sediment column stays constant. As porosity decreases with depth, matrix compressibility ( is also reduced. Based on Equations 10 and 11, the matrix compressibility is calculated as = (bn)/[(s f) (1-n)2 g] (12)

PAGE 71

58 Model Implementation This modeling study focused on Sites 1173, 1174, and 808 to examine the development of pore pressures along the Mu roto Transect duri ng early subduction. Modeling is one-dimensional, and thus, lateral fluid flow cannot be included. However, as discussed above, evidence for lateral fluid fl ow at this Transect is inconclusive. A method previously described by Screaton and Ge (2000) and Ke merer and Screaton (2001) was modified for this investigation to add the permeability-porosity relationship (Eq. 2). The modeling method combines a lo ading program to simulate pore pressure increases due to sedimentation and initial subduction with an existing fluid flow and transport code, SUTRA (Voss, 1984). Based on the rate of sedimentation or thic kening of the overriding prism for each of the segments, the loading program calcula tes the additional thickness of each added sediment layer. As new sediment layers ar e added to the top of the model, the layers beneath were moved down one row, and their pore pressures were incremented with the additional pore fluid pressure due to the load of the new layers multiplied by (Eq. 6). The additional load is calculated from the thickness and bulk density of the new layer, with the hydrostatic pressure subtracted. The updated pore pressures were used as i nputs to SUTRA as initial conditions to perform transient fluid flow simulations. On ce the pore pressures (P) at the end of each loading step were calculated by SUTRA, they were transferred back into the loading program, in which effective stress was calcula ted using Equation 7. If effective stress increased relative to the previous loading step, the porosity decrease was determined using an iterative method to solve Equation 10 for change in porosity. Vertical spacing

PAGE 72

59 was reduced to maintain constant solid volum e and to ensure mass balance (Eq. 11). The new porosity values were then used to calcula te the matrix compressibility for the next sedimentation step (Eq. 12). SUTRA was modi fied to calculate speci fic storage for each node based on the compressibility and poros ity calculated by th e loading program. Model Dimensions, Boundary Conditions, and Initial Conditions The geometry and hydrogeology of the Nankai accretionary complex were simplified into three hydrogeologic units in cluding the turbidites, upper Shikoku and lower Shikoku Basin hemipelagic sequences. The model domain was discretized into 200 rows. Because the model is one-dimensional, lateral flow along the dcollement could not be simulated and thus, special phy sical properties such as higher permeabilities were not assigned to the dcollement. The upper and lower Shikoku Basin layers were assigned si milar porosity and permeability parameters as the two layers were mainly composed of hemipelagic muds. The physical properties for hemipelagic units were assigned based on geological observations and laboratory measurements. The lithology of the turbidite unit at Nankai is composed of a variety of thick to th in bedded sand and silt turbidites with some hemipelagic muds (Shipboard Scientif ic Party, 2001). Values of no (0.77) and b (1.1 x 10-3 m-1) for hemipelagic sediments were obtaine d from Screaton et al. (2002) while for turbidites (no=0.65, b = 7 x 10-4 m-1), the porosity-depth relationship of Bekins and Dreiss (1992) was used. To test the sensitivity of the b coefficient for hemipelagic sediments, I used the porosity data from Site 1173 to estima te the standard deviation for the b values as 3 x 10-4 m-1. Using the standard deviation, I check ed the sensitivity of b values at the minimum possible value of 1.4 x 10-3 m-1 and a maximum possible value of 8 x 10-4 m-1. For the turbidites, I used b values of 1.0 x 10-3 m-1 and 4 x 10-4 m-1 as the maximum and

PAGE 73

60 minimum based on the b values used by Be kins and Dreiss (1992). The estimated values were only increased by 0. 04 at the maximum b value while values decreased by the same amount at the minimum b value, s uggesting the sensitivity to the b value is small. The permeability-porosity relationship for the hemipelagic sediments was obtained from the laboratory permeability measurements (Equation 3). Generally, the sandy portions of the turbidites would be expect ed to have higher permeabilities than the hemipelagic sediments. However, because the lowest permeability layer generally controls vertical fluid flow (e.g., Bear a nd Verruijt, 1987), I used the same permeabilityporosity relationship for both tu rbidites and hemipelagics. I examined the effects of the permeability-porosity relationship by varying the log (k0) value for both hemipelagics and turbidites during the sensitivity analyses. The top boundary of the model was set as hydr ostatic to simulate the effect of the water column above the complex. I assume that the hydraulic connection between the sediments and the oceanic crust is low due to an ash layer that would be expected to alter to low-permeability clays (Saffer and Bekins, 1998). Thus, permeability of 10-23 m2 was assigned to all elements below the lower Shi koku basin facies. The ocean crust consisted of sufficient rows at the beginning of the simulation so that only “ocean crust” rows would be dropped from the base of the mode l as the sediment layers are added, to maintain a constant number of elements throughout the simulation. Matrix compressibility of the ocean crust was set at 1.0 x 10-11 Pa-1. To approximate the effects of in situ temperature on viscosity, I applied temperature and heat flow boundary condi tions, and used SUTRA to calculate

PAGE 74

61 temperature distribution. Both the seafloor and oceanic cr ust boundaries were treated as specified temperature boundaries. The seafloor boundary was set at 2o C while the heat flow at the base of the model was assigned to produce temperatures consistent with those obtained from shipboard measurements. Therma l conductivities and specific heat values for both fluid and solid matrix are given in Table 3-2. SUTRA then calculated fluid viscosity as a function of temperature using the followi ng relationship (Voss, 1984) 133.15) (T 248.37)10 10 (239.4 (T)7 (13) where is pore fluid viscosity [kg m-1s-1] and T is temperature in C. Table 3-2. Fluid and solid matrix prop erties used for numerical simulations. Parameter Value Fluid compressibility [Pa-1] 4.40E-10 Fluid density [kg m-3] 1035 Fluid specific heat [J kg-1 C-1] 4180 Fluid thermal conductivity [J s-1 m-1 C-1] 0.7 Solid grain density [kg m-3] 2650 Solid grain specific heat [J kg-1 C-1] 1000 Solid grain thermal conductivity [J s-1 m-1 C-1] 3.0 Sedimentation and Prism Thickening Rates Sedimentation seaward of the deformation front and loading due to the over-riding prism were applied separately in two differe nt phases (Table 3-3) During phase one, sedimentation rates calculated from the biostr atigraphy were used to build the sediment columns at Sites 808, 1174, and 1173. At S ite 808 and 1174, the sediment columns were built using 137 time steps of 100,000 years and at Site 1173, 140 time steps were used. The average initial sedimentation rate for each unit was based on the initial thickness of that unit and its correspondi ng deposition time as provided by the Shipboard Scientific Party (2001). The initial thickness of each uni t (i.e., the thickness of the sediment layer prior to consolidation) was calculated base d on an assumed initial porosity of 0.77 and

PAGE 75

62 0.65 for hemipelagics and turbidites respec tively, and a final porosity based on core measurements of shipboard porosity (Shipboa rd Scientific Party, 2001). During phase two, prism thickening rates for Sites 1174 and 808 were calculated using change in prism thickness and a convergence rate of 4 cm/yr. The convergence rate was used to calculate the time needed for the incoming sediment s to underthrust to Sites 1174 and 808. In phase two, each stress step was 50,000 years, and Site 1174 was loaded for 6 stress steps while Site 808 was loaded for 7 stress steps. Table 3-3. Summary of sedimentation rates cal culated from biostratig raphy at Sites 1173, 1174 and 808 and prism thickening rates calculated from prism geometry and convergence rate. Site Unit Thickness (m) Time (Ma) Vertical loading Initial sedimentation rate (m/yr) Prism thickening rate (m/yr) 1173 Turbidites 120 Present-0.25 5.07 x 10-4 upper Shikoku 220 0.25-2.85 7.42 x 10-5 lower Shikoku 50 2.85-3.75 1.21 x 10-4 Proto-dcollement 25 3.75-4.95 4.55 x 10-5 lower Shikoku 290 4.95-14.00 8.08 x 10-5 1174 Turbidites 420 Present-0.30 1.9 x 10-3 Trench-wedge facies 50 0.3-0.90 1.91 x 10-4 upper Shikoku 190 0.90-2.5 2.61 x 10-4 lower Shikoku 149 2.5-6.25 1.12 x 10-4 Dcollement 25 6.25-7.15 7.85 x 10-5 lower Shikoku 256 7.15-14.30 1.20 x 10-4 808 Turbidites 550 Present-0.35 3.6 x 10-3 Trench-wedge facies 75 0.35-0.95 4.28 x 10-4 upper Shikoku 188 0.95-2.85 2.20 x 10-4 lower Shikoku 125 2.85-5.35 1.46 x 10-4 Dcollement 25 5.35-6.35 7.30 x 10-5 lower Shikoku 265 6.35-14.35 1.03 x 10-4 Results Model Results Results of the base run using the one-dime nsional loading and fluid flow model are summarized in Table 3-4. For the base run, values of k0 and were assigned according

PAGE 76

63 to the permeability-porosity relationship determined from laboratory measurements (Equation 3). The simulated porosity profiles at all three sites d ecrease smoothly with depth in the lower Shikoku Basin facies. At Site 1173, the simulated porosity profile indicates a small overestimate of porosities compared to the observed porosity profile (Figure 3-5). This reflects that Si te 1173 is slightly overpressured ( *=0.04) in the simulation (Figure 3-5), wher eas the porosity-dep th relationship assumed hydrostatic conditions at Site 1173 (S creaton et al., 2002). Th is would imply that estimated based on the assigned b value would be an underestimate. It should be noted that results of Scre aton et al. (2002) and Saffer (2003) make a similar assumption of hydrostatic conditions at Site 1173. As a result, their pore pressure results may also underestimate actual pore pre ssure ratios. At bot h Sites 1174 and 808, the simulated porosities reasonably match th e observed porosities a bove the dcollement while underestimating the porosities below the dcollement (Figure 3-6). The estimated values below the dcollement at Site 1173 sl ightly increase with depth while at Sites 1174 and 808, values slightly decrease with depth below the dcollement. Simulated excess pore pressure ratios *, observed at 15 meters below the dcollement increase landward from 0.04 at Site 1173, to 0.14 a nd 0.22 at Sites 1174 and 808 respectively (Figure 3-5 and Figure 3-6), but are less than that inferred by Screaton et al. (2002) and Saffer (2003). Thus, using the permeabilityporosity relationship from Equation 3, the simulations suggest that sedi mentation and prism loading alone is not sufficient to generate excess pore pressures as large as predicted from previous studies.

PAGE 77

64 Table 3-4. Summary of simulation runs at Site 1174 and 808. Values of and porosity for each simulation are given at 15 m below the dcollement (855 mbsf at Site 1174 and 979 mbsf at Site 808). At th ese depths, shipboard measurements indicate porosities of 0.37 at Site 1174 and 0.36 at Site 808 (Shipboard Scientific Party, 1991; 2001). Italics i ndicate parameters that were changed from the base run. Run Unit log (k0) Lateral Stress Ratio 10-m Low k Barrier (m2) Site 1174 Site 1174 n Site 808 Site 808 n Base run Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0 NA 0.14 0.31 0.20 0.28 Saffer and Bekins (1998) Turbidites Hemipelagics -20.00 -20.00 5.25 5.25 0 NA 0.20 0.33 0.31 0.33 Raise turbidite k Turbidites Hemipelagics -16.82 -19.82 5.39 5.39 0 NA 0.10 0.30 0.15 0.28 Raise bulk k Turbidites Hemipelagics -18.82 -18.82 5.39 5.39 0 NA 0.06 0.29 0.10 0.25 Lower bulk k Turbidites Hemipelagics -20.82 -20.82 5.39 5.39 0 NA 0.47 0.45 0.47 0.42 Lower bulk k Turbidites Hemipelagics -20.15 -20.15 5.39 5.39 0 NA 0.23 0.34 0.29 0.32 Vary lat.stress Turbidites Hemipelagics -19.82 19.82 5.39 5.39 0.1 NA 0.13 0.31 0.20 0.29 Vary lat.stress Turbidites Hemipelagics -19.82 19.82 5.39 5.39 0.3 NA 0.15 0.31 0.21 0.29 Vary lat.stress Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0.5 NA 0.18 0.32 0.23 0.30 Vary lat.stress Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0.7 NA 0.21 0.32 0.26 0.30 Vary lat.stress Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0.9 NA 0.24 0.32 0.29 0.31 Lower bulk k + lat stress Turbidites Hemipelagics -20.15 -20.15 5.39 5.39 0.3 NA 0.26 0.35 0.32 0.34 Low k barrier Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0 10-19 0.15 0.31 0.21 0.29 Low k barrier Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0 10-20 0.28 0.35 0.34 0.32 Low k barrier Turbidites Hemipelagics -19.82 -19.82 5.39 5.39 0 10-21 0.45 0.40 0.52 0.39

PAGE 78

65 Figure 3-5. Simulated porosity and profiles for base run at Site 1173. A) Comparison of observed porosity to simulated por osity at Site 1173. B) Simulated profile at Site 1173. Shading represents the age-equivalent level of Site 1174 dcollement zone. Sensitivity to Bulk Permeability Because sediment permeability is the most important factor that controls modeled pore pressures (Saffer and Bekins, 1998), here I conducted a sensitivity analysis to test its effects on porosities and excess pore pressures. Bulk permeabilities of hemipelagics and turbidites were changed separately. Results indicate that simulated values were not very sensitive to an increase in turbidite permeability. For example, when turbidite permeability was increased 3 orders of magnitude by changing log (k0) from -19.82 to – 16.82 the simulated decreased by only 0.05 at S ite 808 and by 0.04 at Sites 1174 (Figure 3-6).

PAGE 79

66 Figure 3-6. Simulated porosity and profiles with varying bulk permeability at Sites 1174 and 808. A & C) Comparison of obs erved porosity to simulated porosity at Sites 1174 and 808. B & D) Simulated profiles at Sites 1174 and 808. Shading represents the dcollement zone. The values estimated by Saffer (2003) were based on LWD (Site 1174) and Shipboard data (Site 808). Vertical lines represents es timate by Screaton et al. (2002). When the bulk permeability of both turbidit es and hemipelagics was lowered by an order of magnitude (from log (k0) =-19.82 to –20.82 the pore pressures below the dcollement significantly increased compared to the base run (Table 3-4, Figure 3-6). The simulated values reached above 0.45 at both Sites 808 and 1174 and the simulated porosities above and below the dcollement were overestimated compared to the observed porosity values (Figures 3-6). As bulk permeability was lowered, the steepness of the slope of the profile gradually incr eased (Figure 3-6). The profile for log (k0) = -20.15 represents the rela tionship shown in Figure 3-4 for the lower range of permeability values. The profile of this simulation shows higher values for compared to the results obtained for th e base run. The estimated porosities

PAGE 80

67 below the dcollement at both sites were less than observed values. The simulated porosities above the dcollement were higher at Site 1174 than observed, while at Site 808 they reasonably matched the observed porosities. I also examined the sensitivity of porosity and to a permeability relationship (k=20+5.25n) in the range used by Saffer and Beki ns (1998) (Table 3-4, Figure 3-6). The simulated values using this relationship predicts greater values than those predicted from the base run suggesting that simulated values are very sensitive to even slight changes in the log-linear permeability-porosit y relationship (Eq. 2). The predicted using this relationship were greater by 0. 06 (Site 1174) and 0.11 (Site 808) compared to the values predicted from the base run (Table 3-4). In comparison to the profile of Saffer (2003), the values of predicted for log (k0) = -20.82 is in the same range as the values predicted by Saffer (2003). The general trend of the profile matches values predicted by Saffer (2003) best in the upper (Site 808) and mid (S ite 1174) portions of the undert hrust and with depth the profile gradually deviates to lower values than predicted by Saffer (2003). Sensitivity to Lateral Stress In reality, the prism sediments are struct urally deformed by lateral stresses caused by tectonic compression. To examine the e ffects of tectonic comp ression on excess pore pressures and porosities, I included lateral stress within the prism sediments as an additional fraction of the vert ical loading (Jaeger and C ook, 1969). Following Domenico and Schwartz (1998) the ratio of horizontal to vertical stress becomes greater than one in areas of tectonic compression. I varied the la teral stress factor from 0.1 to 0.9 (i.e., the

PAGE 81

68 total stress at any point in the prism was assume d to be 1.1 to 1.9 times the vertical stress, Table 3-4). Figure 3-7. Simulated porosity and profiles with added lateral stress at Sites 1174 and 808. A & C) Comparison of observed por osity to simulated porosity at Sites 1174 and 808. B & D) Simulated profiles at Sites 1174 and 808. Shading represents the dcollement zone. The values estimated by Saffer (2003) were based on LWD (Site 1174) and Shipboard data (Site 808). Vertical lines represents estimate by Screaton et al. (2002). When lateral stress was added to the pr ism sediments, the porosities smoothly decreased with depth above th e dcollement, and abruptly increased to a maximum at the dcollement (Figure 3-7). Below the dcoll ement, the porosities gradually decreased with depth. The sharp change in porosity at the dcollement is due to the added lateral stress increasing effective stress and lowering porosities within the prism sediments. The simulated values with added lateral stress show a gradual increase of pore pressures above the dcollement while reaching the maximum within the dcollement (Figure 37). Below the dcollement, values gradually decreased with depth. Even though lateral stress is added only in the prism, the values beneath the prism increased because the vertical fluid fl ow from the underthrust is restricted by the

PAGE 82

69 elevated pressures within the prism. Thus, when the lateral stress was increased, both and porosity values were slightly increased below the prism (Table 3-4). However, the effect of prism lateral stress on pore pressu res in the underthrust is relatively small compared to the sensitivity to bulk perm eability. Although it may be possible to have greater values of the la teral stress factor than 0.9, si mulations suggest that they would have little additional affect on pore pr essures in the underthrust sediments. Increasing the lateral stress f actor results in a greater offset in porosity across the dcollement zone. For example, the simulated porosity profile with a lateral stress factor of 0.5 underestimated observed porosity valu es both above and below the dcollement (Figure 3-7). Raising the lateral stress factor increases porosities in the underthrust sediments (Table 3-4), yielding a better match to observed, but decreases porosities in the prism, yielding a poorer match. Thus, by itself, changing lateral stress (with the porositypermeability relationship obtained in this study) cannot generate porosities that matched observed porosities. When lateral stress was combined with lo wer bulk permeability, the pore pressures were significantly increased with respect to la teral stress alone. For example, when the bulk permeability of the hemipelagic sediments were lowered to a log (k0) of -20.15 (the lower boundary to permeability data in Figure 34) with a lateral stre ss factor of 0.3, the at 15 m below the dcollement at Site 808 reached 0.32. The simulated porosity profile for this combination well matche d the observed porosit y profile above the dcollement while the porosities below the dcollement were very slightly underestimated at both Site s 808 and 1174 (Figure 3-7).

PAGE 83

70 Sensitivity to a Low-Permeability Barrier Previous accretionary complex studies have suggested the idea of a lowpermeability barrier at the dcollement as a possible cause of excess pore pressures. For example, at the Barbados accretionary prism, Bekins et al. (1995) simulated a 15-m thick low-permeability cap above the top of the dcollement based on an anomalously low permeability value (6.5 x 10–21 m2) measured by Taylor and Le onard (1990). In a more recent study based on inferred porosity vari ations obtained from inverted seismic reflection data at Nankai, Bangs and Gulick (2005) argued that c onsolidation of the uppermost lower Shikoku Basin strata forms a barrier blocking the fluid flow from below. Because the barrier lie s just above the projected leve l of the dcollement, they suggest that higher-porosity, underconsolidated and overpressured sediment below forms a surface of potential dco llement propagation. To test the effects of a low-permeability barrier at the dcollement at Nankai, I ran simulations with a 10-m thick low-permeability barrier above the top of the dcollement zone. This barrier was implem ented after the initial sediment ation steps and prior to the loading by the prism (Table 3-4). As expected, when the low-permeability was added, the values below the barrier abruptly increase d to a maximum because the fluids in the underthrust could not migrate thr ough the barrier fast enough to keep pace with loading. The simulated values at both Sites 1174 and 808 gra dually decrease with depth below the dcollement (Figure 3-8).

PAGE 84

71 Figure 3-8. Simulated porosity and profiles with added low-permeability barrier at Sites 1174 and 808. A & C) Comparison of observed porosity to simulated porosity at Sites 1174 and 808. B & D) Simulated profiles at Sites 1174 and 808. Shading represents the dcollement zone. The values estimated by Saffer (2003) were based on LWD (S ite 1174) and Shipboard data (Site 808). Vertical lines represents es timate by Screaton et al. (2002). The simulated porosities smoothly decrease both above and below the dcollement while reaching a maximum at the dcolleme nt (Figure 3-8). The porosity profiles generated by the low-permeability barrier with a permeability of 1 x 10–21 m2 were slightly higher below the dcollement at both Sites 808 and 1174 than the observed porosities. Above the dcollem ent, the porosities closely matc hed the observed values at Site 1174 while slightly underest imating the observed values at Site 808. The simulated values at both sites were similar to the valu es presented by Screat on et al. (2002). In comparison to the profile of Saffer (2003), the values of predicted for the lowpermeability barrier is roughly in the same range as those predicted by Saffer (2003). However, the general trend of the profile of the low-permeability barrier is different from that of Saffer (2003). The low-permeab ility barrier predicts a sharp decrease in

PAGE 85

72 with increasing depth while the profile of Sa ffer (2003) predicts a gradual decrease of with increasing depth. Although modeling resu lts indicate high values at and below the dcollement due to the added low-permeability barrier, core samples immediately above the dcollement are not available for laboratory -measurements to assess the presence of a low-permeability barrier. Although I do not have any independent evidence suggesting that the permeability is low in the direction perpendicular to the plane of dcollement, previous studies have suggested the forma tion of a low-permeabil ity barrier during the accretionary process due to sh earing of sediments below the wedge as it moves above the underthrust sediments. For example, Maltman et al., (1992) suggested that at the Nankai accretionary complex, high stresses at the de pth of the dcollement could seal fractures lowering its permeability despite the presence of highly brecciated and fractured material. This idea is further strengthened by lack of evidence for fluid flow such as veins, mineralized surfaces or clastic dykes at the d collement (Maltman et al., 1992). If a lowpermeability barrier does exist, then it is more likely to form at the base of the wedge where the high shearing is present increasing the pore pressures below the barrier. The presence of a low-permeability barrier is lik ely to generate excess pore pressures below the barrier, as shown in Figure 3-8. The affects of lateral fluid flow on pore pres sures are difficult to predict. Lateral fluid flow could allow pore pressures to be lo wered, if fluids can escape to the seafloor near the toe of the prism. However, if the d collement transmits fluids to near the toe of the prism, and these fluids cannot escape, por e pressures will be increased at the updip end of the lateral flow path. Lateral fluid flow within the dcollement could decrease the

PAGE 86

73 significance of the low-permeability barrier, if fluid could escape along the dcollement zone. Similarly, lateral fluid escape along th e dcollement may decrease the impact of lateral stresses within the prism on values within the underthrust sediments. Because the dcollement has been speculated as a possi ble, yet controversial, pathway for lateral flow it will be valuable to address in future investigations. Implications Both simulations using the added lateral stress and low-permeability barrier show a sharp increase in near the dcollement zone at both Sites 1174 and 808 (Figure 3-7 and Figure 3-8). However, each scenario generated a distinct profile. The profile generated by the added latera l stress demonstrate a gradual increase in pore pressures both above and below the dcollement, and the value peak at the dcollement. The maximum corresponds to the maximum simulate d porosity found near the top of the dcollement zone (Figure 3-7). In contrast, the profile generated by the lowpermeability barrier shows an abrupt increase of pore pressures at the lower part of the dcollement. Above the dcollement, the profile shows a very slight decrease with depth while below the dcollement decreases more gradually. As observed in both scenarios, the maximum is located near the dcollement, consistent with previous inferences (e.g., Hubbert and Rubey, 1959) that pore fluid pressures play a major role in the mechanics of thrust faulting. The excess pore pressures observed near the dcollement can also be related to the sliding of the dcollement as proposed by the cr itical wedge theory (Davis et al., 1983). Critical wedge theory predicts that pore pr essures significantly gr eater than hydrostatic are needed to maintain small taper angles su ch as at the Muroto Transect of Nankai.

PAGE 87

74 Generally a small critical taper is an indication of very little basal friction, due either to low intrinsic strength or elevated pore pressure s, or that the wedge consists of a strong material, which need not deform for stable sl iding to occur (Davis et al., 1983). Thus, it can be speculated the maximum values observed in the profiles at the dcollement could represent the level of least mechanical resi stant that promotes stable sliding. It is possible that a permeability barrier would also affect fault zo ne initiation, if elevated pore pressures are transmitted seaward of the deformation front. A two-dimensional model would be needed to fully examine this possibility. Many previous studies based on the critic al taper theory have assumed the pore pressure ratios (pore fluid pressure/lithosta tic pressure) at the ba se of the wedge and within the wedge to be equal. The results from simulations with the low-permeability barrier demonstrate a mechanism for basal pore pressure ratios to be significantly greater than those in the wedge which could increase taper angle for a given basal pore pressure. Lateral stresses in the prism may also result in basal pore pressure ra tios to be higher than in the wedge, but the difference is not as gr eat as with the low-pe rmeability barrier. Conclusions I examined the effects of different parame ters on excess pore pressures at the toe of the Nankai accretionary complex. Using a permeability-porosity re lationship based on a best fit to laboratory data, simulations sugge st that sedimentati on and prism thickening generate excess pore pressures, but not as high as predicted from previous studies. Results from this study demonstrated si gnificant increase in pore pressures at the dcollement with lower bulk permeability, such as obtained by using the lower boundary of permeability-porosity data. Because th e lowest permeabilities generally control

PAGE 88

75 vertical flow, this relationship may be more appropriate for the simulations than the best fit equation. However, if the high excess pore pressures suggested by Screaton et al. (2002) or Saffer (2003) are correct, permeabilitie s must be even lower, requiring either a bulk permeability represented by log (k0) = -20.82, a low permeability (<10-20 m2) barrier above the dcollement, or a combination of lower bulk permeability and permeability barrier. Alternatively, other factors must c ontribute, such as recent prism growth rates greater than the time-avera ged rates simulated here. The magnitude of pore pressures in the underthrust sediments demonstrated only slight sensitivity to added lateral stresses in the prism, although the profile of the pore pressure ratio is affected. Results furthe r illustrated that the simulations with lowpermeability barrier and lateral stress both pr oduced a sharp increase in porosities below the dcollement zone, as is observed in the measured values. Furthermore, in both scenarios, maximum excess pore pressure ra tios were found at the dcollement, which could contribute to stable sliding of the dcollement zone.

PAGE 89

76 CHAPTER 4 EVOLUTION OF HIGH PORE PRESSURES AND IMPLICATIONS FOR EPISODIC FLUID FLOW AT THE NORTHERN BARBADOS ACCRETIONARY COMPLEX Introduction Variations in fluid flow pressures an d its distribution w ithin subduction zones regulate the mode of deformation, affecting th e evolution of the accretionary complex. It has been speculated that excess pore pressure s play a major role in the mechanics of thrust faulting (Hubbert and Rubey, 1959) thus, allowing the weak semilithified sediments in accretionary complexes to g lide over the subducting plate along a low-angle detachment surface (Davis et al., 1983). Exce ss pore fluid pressures are also responsible for increasing sediment permeability associated with reduction in effective stress (Yeung et al., 1993; Fisher and Zwart, 1996). In a ddition, pore pressure has been claimed to influence seismogenic faulting, through its control on effective st ress and consolidation state (e.g., Moore and Saffer, 2001; Scholz, 1998). Thus, an understanding of the development of excess pore pressures will provide valuable insight to the evolution of accretionary complexes as well as to subduction z one processes such as fault mechanics. The abundant evidence for excess pore pres sures at Barbados accretionary complex provides an excellent opportunity for the study of evolution of pore pressure generation. The distribution of mud volcanoes (Gretene r, 1976) and the overall shape of the accretionary prism (Davis et al, 1983) suggest the existence of excess pore pressures at the Barbados accretionary complex. In additi on, elevated pore fluid pressures have been inferred at the Barbados accretionary comple x from fluid flow modeling (e.g., Shi and

PAGE 90

77 Wang, 1988; Screaton et al., 1997; Bekins et al., 1995; Henry, 2000) and direct borehole measurements along the dcollement (Fouche r et al., 1997; Becker et al., 1997). Studies have speculated that fault zone s play a major role in focusing fluid expulsion (e.g., Bekins et al., 1995; Moore et al., 1998; Henr y, 2000). Indirect evidence for transient fluid flow along the dcolleme nt is provided by geochemical and thermal anomalies (e.g., Bekins et al., 1995; Fish er and Hounslow, 1990). A widely held hypothesis relating elevated pore pressures and evidence for flui d flow is that flow must be episodic, although the mechanisms produci ng the episodic fluid flow events are not fully understood. Field evidence for episodic fl uid flow includes the presence of multiple episodes of fracturing and vein filling in accreted sediments (Labaume et al., 1997). However, the time scales of these events are difficult to estimate (K nipe et al., 1991). Several mechanisms have been put forward to explain the time-variable permeability in the fault zone. One mech anism is that the permeability in the dcollement is enhanced by the episodic ope ning of horizontal hydrof ractures when pore pressures reach values above lithostatic (e .g., Behrmann, 1991; Moore and Vrolijk, 1992; Brown et al., 1994). Similarly, numerical modeling by Bekins et al. (1995) predicted results with observed chloride anomalies within the dcollement of the Barbados accretionary complex by assuming an instan taneous permeability increase within the dcollement. Another mechanism for episod ic fluid flow is based on in situ bulk permeability measurements that were made at a variety of fluid pressure conditions. According to Fisher and Zwart (1996), this relationship between bulk permeability and effective stress may explain the dynamics of fluid-fault interactions and the transient nature of hydrologic processe s at convergent margins. A dditional hydrologi c tests that

PAGE 91

78 were conducted at a sealed borehole penetr ating the dcollement at the Barbados accretionary complex support the conclusion by Fisher and Zwart (1996) that significant permeability increase can occur within the dco llement at pore pressures below lithostatic pressures (Screaton et al., 1997). Studies ba sed on fluid budgets show s that fluid flux varies with both arcward di stance and through time (Saffer and Bekins, 1999). In addition, fluid budget studies sugge st that initiation of connected flow conduits is delayed with respect to the time of accr etion and may be related to bur ial below a critical depth, where channelized fluid escape is more effici ent than diffuse flow to the sea floor or where sediments may behave brittlely (Sa ffer and Bekins, 1999). Even though previous modeling studies have investig ated the production of overpre ssures, these models did not indicate when the overpressures are generated in the evolution of the complex and thus, the connectivity between excess pore pressu res and episodic fluid flow is not well understood. Models used by Henry a nd Wang (1991), Shi and Wang (1994), and Stauffer and Bekins (2001) focused on processe s that take place at the toe of the prism during initial offscraping. Bekins et al. ( 1995) focused on steady-stat e pore pressures and transient fluid flow assuming instantaneous dcollement permeability after the entire prism had grown. However, to fully unders tand the development of pore pressures and thus, hypothesized episodic fl uid flow, one should examine the development of pore pressures both at the toe and deeper parts of the accretionary complex through both space and time. Thus, in this study I modeled 50 km of the accretionary complex as a timedependent evolving prism. A combined prism growth and fluid flow model was used to examine the development of pore pressures. Mechanisms for episodic fluid flow were examined during the evolution of the accretionary complex by including hydrofracture or

PAGE 92

79 a dcollement with varying permeability based on a relationship of bulk permeability vertical effective stress. Background The Barbados accretionary complex is located in the Caribbean where the Atlantic Plate (Figure 4-1) is being subducted beneath the Cari bbean Plate at a rate of 2 cm/yr in an east-west direction (DeMets et al., 1990). At Barbados, active accretion of sediments takes place at the eas tward end of the complex as the more stabilized portion lies westward, where the complex is partiall y exposed above sea level at the Barbados Island (Figure 4-1). The complex varies in thickness from 200 km south of Tiburon Rise at 14 N, to approximately 10 km north of Tiburon Rise at 16 N (Bangs and Westbrook, 1991). The variation in thickness of the comp lex is related to the distance from the terrigenous sediment source from South America (Underwood and Deng, 1997), as well as the affect of local barriers such as th e Tiburon Rise (Figure 4-1), which slows the influx to the complex. The northern Barbados accretionary comple x is rich in hemipelagic sediments (Bekins et al., 1995), whereas the southern pa rt is dominated by turbidites (Langseth et al., 1990). The age of the sediments at the Barbados accretionary complex varies from Late Eocene to Late Cretaceous (Underw ood and Deng, 1997). Mud volcanoes and mud diapirs, which indicate excess pore pressures, are abundant in the southern part of the complex where the sediment sequence is thicker (Moore et al., 1982). Based on seismic reflection images the dcollement was estimated to be ~14 m thick at the northern Barbados accretionary complex (Shipley et al., 1994). It was inferred as a high-porosity zone with undercom pacted sediments and high-fluid pressures

PAGE 93

80 (Moore et al., 1995). During Deep Sea Dr illing Project (DSDP), Leg 78A high pore pressures were encountered at the toe of the northern Barbados accretionary complex while attempting to drill through the dcollem ent (Biju-Duval et al., 1984). The seismic profiles show prism sediments being subjected to high lateral strain while sediments in the underthrust remain undeformed (e.g., We stbrook et al., 1988), possibly due to the presence of excess pore pressures on the dcollement. Figure 4-1. Location map and cross-section of the Barbados accretionary complex. A) Map of the eastern Caribbean showin g the deformation front. B) Crosssection of the Barbados accretionary complex (Shipboard Scientific Party, 1998). W N

PAGE 94

81 Ocean Drilling Program (ODP) Leg 110 revis ited northern Barbados and three sites were drilled (Sites 671, 672 and 676). S ites 671 and 676 (Figure 4-1) were drilled arcward of the deformation front at a di stance of 4.5 km and 0.25 km respectively (Mascle et al., 1988). Site 672 was drilled 6 km east of the deformation front to provide an undeformed reference site (Mascle et al., 1988). The core samples from Leg 110 were analyzed for structural features, chemical signals, permeability, consolidation behavior, and bulk composition. Taylor a nd Leonard (1990) inferred n ear-lithostatic pore pressures directly above the proto-dcollement zone ba sed on consolidation tests on core samples. In contrast several other studies (e.g., Shi and Wang, 1985; Screaton and Ge, 2000) concluded that sedimentati on rates alone were not suffi cient to produce excess pore pressures. During Leg 156 Logging While Drilling (LWD) was performed at Site 948, which coincides with the location of Site 671 (Shipl ey et al., 1997). Site 949 was drilled 2 km northeast of Site 948 (Shipley et al., 1997) During Leg 156, a hydrologic borehole packer was successfully deployed and fluid fl ow experiments were conducted at Sites 948 (671) and Site 949 (Leg 171A, Site1046) (Fisher and Zwart, 1996). Results of packer experiments further supports the pr esence of excess pore pressures at the dcollement, although perturbations from drilling and testing were difficult to separate from natural pore pressures. Estimated values for fault zone permeability from in situ packer tests vary from 10-12 m2 when fluid pressure is at lithostatic to 10-18 m2 when fluid pressure is at hydrostatic (F isher and Zwart, 1996). Results of hydrogeologic tests that were performed at a sealed borehole (949C) penetrating the dcoll ement support the idea

PAGE 95

82 of significant permeability increase within th e dcollement with increasing pore pressure (Screaton et al., 1997). Fluid Flow and Pore Pressures Evidence for fluid flow at the Barbados accretionary complex comes from the presence of low-chloride anomalies observe d along the dcollement (Kastner et al., 1993). It has been suggested by Bekins et al. (1995) that smectite dehydration is the most likely mechanism for low-chloride anomalies. In contrast, Fitts and Brown (1999) suggested that low-chloride anomalies occur as a result of arti ficial squeezing of sediments released from smectite interlayers during pore water sampling. However, even after accounting for the effects of sample squ eezing process, the low-chloride anomaly is still 12% fresher than seawat er (Fitts and Brown, 1999) s upporting the clay dehydration as a possible explanation for low-chloride anomalies. The clay dehydration reaction takes place at temperatures between 60-160 C (Perry and Hower, 1970). Based on kinetic modeling of clay dehydration in the Ba rbados accretionary complex, Bekins et al. (1995) estimated the peak reaction window to be at 50 km arcward of the deformation front. Thus, if the fluid released from smectite dehydration is responsible for the observed low-chloride anomalies at the toe of the prism (Site 948), then pore fluids must migrate over 46 km to reach Site 948 (Bekins et al., 1994). Heat flow anomalies have also been observed at the Barbados accretionary complex from temperature measurements (F isher and Hounslow, 1990) and surface heat flow surveys (Foucher et al., 1990). Seafloor heat flow values near the toe of the complex range from 96 to 192 mW m-2 (Fisher and Hounslow, 1990) and are well above the 53-55 mW m-2 expected for 90 Ma oceanic lithosphere (Ferguson et al., 1993). One of the possible explanations for the observed heat flow anomalies is fluid flow along the

PAGE 96

83 dcollement (Langseth et al., 1990). In c ontrast, Fisher and H ounslow (1990) and Moore et al. (1998) suggest possibl e lateral fluid flow along turbidites in the underthrust sequence. Muller and Smith (1993) argued that the uniform ly high background heat flow in the ODP drilling area could be due to cr ustal thinning of the Tiburon Rise. Based on both steady state and transient state models, fluid fluxes need ed to explain the observed heat flow anomaly are approximately one order of magnitude higher than the fluxes needed to explain the low-ch lorinity anomaly (Henry, 2000) Similarly several other studies (Foucher et al., 1990; Saffer and Bekins 1999; Cutillo et al., 2003; Bekins and Screaton, 2006) have noted that the outflow from the underthrust necessary to create a thermal anomaly (greater than 60 mW m-2) seems excessive compare to the in flow thus, suggesting that the hypothesis on crustal thin ning under Tiburon Rise should be further explored. In addition to low-chloride and heat flow anomalies, the presence of mineralized veins supports the idea of transi ent fluid flow along the dcoll ement. Mineral veins were found in the upper part of the dcollement (L abaume et al., 1997). The orientations of the veins suggest that pore fluids around the fault zones is at near lithostatic pressure during vein formation (Labaume et al., 1997). Veins that were formed during several growth phases reflect the episodicity of flui d flow along the dcollement (Labaume et al., 1997). Bekins et al. (1995) used low chloride anomalies as constraints on a transient model. In their model they raised permeabilities from 10-12 -10-15 m2 along the entire dcollement zone to match observed low-chloride anomalies. To justify this, they hypothesized that pressures in the accretionary complex build until it reaches values that

PAGE 97

84 are close to lithostatic. These high pore pressures lower the ef fective stress along the dcollement creating fractures or dilating existing fractures, raising the permeability along the dcollement. To simulate this hypot hesized scenario base d on a steady –state model, they assigned varying valu es of dcollement permeability (kd) until sublithostatic pressures were simulated. Once the appropriate kd value was determined they assumed that the pressures estimated from this soluti on represent those immediately before a slip event that increases kd and used these pressures as initial conditions in the transient simulation. When dcollement permeability was suddenly raised the pore pressures along the dcollement reached va lues that are closer to li thostatic during the first few thousand years supporting the concept of pressu re build-up and releas e cycles or episodic fluid flow. In another study, Henry (2000) modeled the ep isodicity of fluid flow in a slightly different manner. In his model he started a pressure pulse at the arcward boundary of the complex and allowed it to propagate along the dcollement. The dcollement was assigned a permeability that increased continuously as pore pressure increased following the relationship of Fisher and Zwart (1996). Al so it is assumed that all fluid flow occurs either along the dcollement or along sandy laye rs at 200 m below the dcollement within the underthrust sequence. Thus, a bulk permeability of 3 x 10-13 m2 is assigned either to the dcollement or to the sand layer to simulate fluid flow. According to this simulation, significant fluid flow persists for several thousands to 10,000 years and it is in agreement with a diffusion-advection model of the chlorinity anomaly (Henry, 2000). Based on long term monitoring at Hole s 948D and 949C during ODP Leg 156, the estimated values (where = (P-Ph)/(Pl-Ph), P = pore pressure, Ph = hydrostatic

PAGE 98

85 pressure and Pl = lithostatic pressure) were 0.50 (F oucher, et al., 1997) and 0.36 (Becker et al, 1997), respectively. Thes e results were similar to thos e previously obtained using a steady state model with a perm eability-depth relation for clay -rich sediments (Bekins et al; 1995). Moreover, the increase of by 0.15 between Sites 949 and 948 over a lateral distance of 2.2 km was also estimated by Stauffer and Bekins (2001) based on inferred consolidation state. In addition, constraints for pore pressure di stribution also follow from an analysis of the mechanical force balan ce in accretionary wedges presen ted by Davis et al. (1983). According to Davis et al. (1983), in order fo r the sediments of the wedge to move over the underthrust sequence along a low-angl e dcollement, high pore pressures ( = 0.92 for the overall taper, where is the ratio of pore fluid pr essures to the vertical normal traction exerted by the lithostatic overburden must be present along the dcollement. The presence of lower pressures would result in a steeper taper angle than that observed at Barbados (Bekins et al., 1995). Hydrofractures Behrmann (1991) suggested that hydrofracturing enhances permeability in argillaceous rock sequences. According to Behrmann (1991) th e capability of rocks to hydrofracture depends on the mode of faulting and the effective mean stress. He also noted that the depth to which hydrofracture c ould occur is a function of both the faulting mode and the ratio of fluid and lithostatic pr essures. Thus, wrench and normal faults hydrofracture even at ratios of fluid and lithosta tic pressure is less than one. In contrast thrust faults always require ratio of fluid and lithostatic pressu re to be greater than one to hydrofracture. Vertical fluid flow has been indicated by 3 –D seismic images at the

PAGE 99

86 northern Barbados accretionary complex wh ich shows at least 50 m offset in the turbidites along the normal faults extending upward from the basement (Zhao and Moore, 1998) suggesting that vertical hydrofracture could occur in th e underthrust. According to Behrmann (1991), the criteria requires to be > 0.8 for vertical hydrofracturing in soft rocks. Vertical hydrofractures occur perpendicular to the least principle stress axis. According to Price (1975), when sediment s hydrofracture, the hydr aulic properties are expected to change dramatically. Modeling Methods Model Implementation A model developed by Screaton and Ge (2000) was modified to simulate the effects of subduction beneath a prism. This model bu ilds the accretionary complex in segments through time. The prism growth/flow model cons ists of two sub programs. The first sub program is a modified loading program (G amage and Screaton, 2006), which builds the initial sediment column that enters the accr etionary complex at the deformation front. The second sub program (prism loading) a dvances the accretionary complex over the subducting sediments in segments through time at a convergence rate of 2 cm/yr. It is assumed that the taper of the prism is constant. The prism-growth/flow model calculates the time necessary for prism to advance one column by dividing the horizontal dimension of the column by the convergence rate. The vertical dimension is calculated using the prism thickening rates. As the prism a dvances seaward the lo ading program adds sediments on each advanced column according to the assigned prism thickening rates. Based on the calculated increase in overbur den and sediment properties, the prismgrowth/flow model calculates the pore fluid pressures. These pore pressures are then input into SUTRA (Voss, 1984), a finite-ele ment code that simulates transient two-

PAGE 100

87 dimensional fluid and thermal transport. SUTRA simulates the fluid flow and the pore pressure changes with time. Once the pore pressures were calculated in SUTRA they were transferred back into the prism-growth/flow model, which calculates effective stress, porosity, and permeability before beginning the next loading step. During hydrofracture, the model checks for re gions that have reached the criteria for vertical hydrofracture afte r every seaward advancement of the prism. If pressures meet the vertical hydrofracture criteria, th en the model increase s vertical permeability from the point of lithostatic pressure up to the dcollement. If pressures are less than the assigned criteria for hydrofracture the perm eability values are assigned according to permeability-porosity relationship. SUTRA uses a backwards finite-differe nce scheme which enhances numerical stability. The large permeability contrast during the introduction of hydrofractures into the model was challenging for the iterative so lver, especially for the thermal transport simulations. However, by reducing the cont rast between the highest and the lowest permeabilities and increasing upstream weigh ting for the transport, convergence was obtained. Model Equations Combining the mass conservation of fluid w ith Darcy’s law, the following equation can be written for two-dimensional transient flow: Q t h S y h K x h Ks yy xx 2 2 2 2 (1) where Ss is specific storage [L-1], h is hydraulic head [L], x and y are spatial coordinates, Q is a source term reflecting pr ocesses such as loading that affect fluid volume or pressure [T-1]. The left hand side term accounts for fluid flow into and out of a

PAGE 101

88 given element in two dimensions. The ri ght hand side terms account for changes in storage of fluid mass due to change in hydrauli c head and addition or removal of fluid or pressure. When initial conditions, boundary conditions and the controlling hydrologic parameters are known, Equation (1) can be solved for head at any point in a twodimensional field at any given time. The e quations present here are changes made to those in Chapter 3. For a complete explanation of the model equations refer to the “Model Equations” in chapter 3. Model Dimensions, Boundary Conditions, and Initial Conditions The cross-section model consists of four zones: the upper hemipelagics that form the wedge, the dcollement, the turbidite s and the underthrust hemipelagic sequence (Figure 4-2). The properties for the four hydrogeologic units were assigned based on geological observations a nd laboratory measurements. The model domain was discretized into finite element grid cons isting 8100 nodes and 7920 quadrilateral elements (Figure 4-2). Horizontally the model exte nds to a maximum of 50 km arcward of the deformation front during the total simulation time. Vertically the model extends from the seafloor to a maximum depth of 3 km at full growth. At the deformation front the incoming sediment sequence was divided into 470 m of underthrust sediments, 15 m of dcollement zone and 173 m of accreted sediments. Values of no (0.7) and b (8 x 10-4 m1) for hemipelagic sediments were obtained from Screaton and Ge (1997) while for turbidites (no=0.6, b = 7 x 10-4 m-1), the porosity-depth relationship of Bekins and Dreiss (1992) was used. The n0 and b values used for turbidites were similar to those values used by Screaton and Ge (1997) for Barbados sediments south of Tiburon Rise, as these sediments are rich in turbidites.

PAGE 102

89 Permeability for the hemipelagic units were assigned based on the permeabilityporosity relationship used by Bekins et al. (1995) (log (k) = -22.0+8.44n). This relationship was based on laboratory permeability measurements of ODP Leg 110 obtained by Taylor and Leonard (1990). Because these permeability measurements only represented samples from less than 500 m of depth, to estim ate the permeability-porosity relationship, Bekins et al. (1995) extended the ODP data to depth by visually placing a line, which roughly bisects th e outline for argillaceous sediments compiled by Neuzil (1994). Because there were no permeability measurements at lower porosities, the constraints for permeability are limited. Thus, it should be noted that the permeabilityporosity relationship predicted by Bekins et al. (1995) might not necessarily represent permeability values at lower porosities. If th e true permeability at lower porosities is lower than what is given by the permeability-p orosity relationship, then the estimated pore pressure values will be underestimated. For turbidites, permeabilities were assigned based on the permeability-porosity relationshi p (log (k) = -20.0+5.25n) used by Screaton and Ge (1997) and Saffer and Bekins (1998) Permeabilities in the dcollement were assigned using the hemipelagic permeability-porosity relationship, unless otherwise noted. The top boundary of the model was specifi ed to be at hydrostatic pressure ( = 0) while the landward and bottom boundaries we re set at no-flow. The landward boundary was treated as no flow to prevent loss of fluids acro ss the boundary. The bottom boundary was treated also as a no-flow boundary, based on the assumption that the permeability of the oceanic crust is low compared to the sediments above. This assumes that most fractures are filled in the 82 m.y old crust (Bek ins et al., 1995). The seaward

PAGE 103

90 boundary was set at hydrostatic constant pr essure to accommodate fluid flow through dcollement during periods of hydrofracture. The model builds up from the ocean crust above. The ocean crust was assi gned a permeability value of 10-23 m2, a porosity of 0.1 and a matrix compressibility of 1.0 x 10-11 Pa-1. Figure 4-2. Grid and boundary conditions for the prism growth/flow model. To incorporate the effects of in situ temp erature on viscosity, I applied temperature and heat flow boundary conditions, and us ed SUTRA to calculate temperature distribution. The seafloor boundary was set at of 2o C (Bekins et al., 1994) while the heat flow at the base of the model was assigned as 55 mW m-2 (Ferguson et al., 1993) to produce temperatures consistent with t hose given in Davis and Hussong (1984) and Bekins et al. (1994). The ar cward and seaward b oundaries were set as no heat flow boundaries. Thermal conductivities and specifi c heat values for both fluid and solid matrix are given in Table 4-1.

PAGE 104

91 Table 4-1. Fluid and solid matrix proper ties used for numerical simulations. Parameter Value Fluid compressibility [Pa-1] 4.40E-10 Fluid density [kg m-3] 1035 Fluid specific heat [J kg-1 C-1] 4180 Fluid thermal conductivity [J s-1 m-1 C-1] 0.7 Solid grain density [kg m-3] 2650 Solid grain specific heat [J kg-1 C-1] 1000 Solid grain thermal conductivity [J s-1 m-1 C-1] 3.0 Sedimentation and Loading Rates Sedimentation seaward of the deformation front and loading due to the over-riding prism were applied separately in two different phases (Table 4-2). Sedimentation rates for phase one were calculated based on the biostratigraphy of Site 672. During phase one, the incoming sediment column was built using 25 time steps of 2,680,000 years. The average initial sedimentation rate for each unit was based on the initial thickness of that unit and its corresponding deposition time as provided in the biostratigraphy of Site 672. The initial thickness of each unit (i.e., the thickness of the sediment layer prior to consolidation) was calcula ted based on an initial porosity of 0.70 and 0.60 for hemipelagics and turbidites respectively. Final porosity was estimated based on bulk density data from Leg 171A, Site 1044 (Shipboa rd Scientific Party, 1998). Porosity was estimated using a solid grain density of 2.65 g/cm3 and a fluid density of 1.035 g/cm3 as used in Screaton and Ge (2000). According to Screaton and Ge ( 2000) the largest error associated in this conversion is due to the presence of smec tite interlayer water, which will overestimate porosity values. During phase two, prism-thickening rates were calculated using prism thickness at Site 671 an d a convergence rate of 2 cm/yr. Because the prism was built in segments, the number of loading steps used in phase two varied from 10 to 40 with a constant lo ading step size of 67,500 years.

PAGE 105

92 Table 4-2. Summary of sedimentation and prism thickening rates calculated from biostratigraphy at Site 672. Site Unit Thickness (m) Time (Ma) Vertical loading Initial sedimentation rate (m/yr) Prism thickening rate (m/yr) 672 Accreted sediments Present-5 2.60 x 10-3 3.90 x 10-5 Hemipelagics 173 5-18 1.50 x 10-6 Dcollement 15 18-25 1.70 x 10-6 Turbidites 285 25-43 1.45 x 10-5 Hemipelagics 187 43-50 1.70 x 10-5 According to Bekins et al. (1995), in deeper parts of th e accretionary complex, fluid released from clay dehydration becomes im portant and had been used to explain the observed low-chloride anomalies at the toe of the prism. Moreover, if the rate of fluid released during the dehydration reaction is high enough this mechanism could also contribute to the generation of excess pore pressures. However, studies have not shown that fluids released during smectite dehydration are large enough to affect pore pressures. A fluid flow and budget study by Saffer and Bekins (1998) concluded that dehydration fluid sources are 10-1000 times smaller than fluid released from co mpaction sources and thus, calculated pore pressures are largely independent of the clay content. Because the model used here only extends to 50 km from the deformation front while based on previous studies (e.g., Bekins et al., 1994) the smectite dehydration is important at distances >50 km from the deformation front, thus fluid sources fr om dehydration were not included in this study. Studies based on fluid budget modeling show that prism fluids do not contribute significantly to the dcollement flow (Bek ins and Screaton, 2006) thus prism fluid reaching the underthrust should be even be sma ller. However, elevated pore pressures within the prism due to sources could i nhibit upward migration of fluid from the

PAGE 106

93 dcollement and underthrust sediment. Because prism sources were not included in this model, low permeabilities were assigned to the prism to compensate. This approach was tested as part of the modeling. Results and Discussion In all of the model runs, the absence of prism sources was compensated for by assigning a very low permeability (1 x 10-26 m2) to the prism sediments. This assumption creates a barrier between the bottom of the prism and the t op of the dcollement zone, which inhibits fluid flow in and out of th e underthrust and thus predicting the maximum possible value within the dcollement and unde rthrust sediments. To examine the validity of this approach, I pe rformed a simulation with the same parameters as used by Bekins et al. (1995), except instead of incl uding sources in the prism as they did, a permeability of 1 x 10-26 m2 was assigned to the prism sediments (Table 4-3). The estimated values from this simulation at 2.7 million years were in good agreement with those simulated in the underthrust sedime nts by Bekins et al. (1995). Furthermore, the simulated values at Site 949C and 948D were c onsistent to those values obtained from long term monitoring, as were the re sults from Bekins et al. (1995). Modeled values along the dcollement were 0.31 at Site 949C and 0.48 at Site 948D while values obtained from long term monitoring were 0.36 at Site 949C and 0.50 at Site 948D. This suggests that assigning a low permeability to the prism is a reasonable compensation for omitting prism sources. The temperature di stribution of the accretionary complex in this simulation is also in good agreement w ith that used by Bekins et al. (1994). For the base run simulation, there were several significant differences from the parameters used in Bekins et al. (1995) The base run simulation contained both

PAGE 107

94 hemipelagic and turbidite units whereas in Beki ns et al. (1995), all sediment units were treated as hemipelagics. In the base run, for hemipelagics, the same permeabilityporosity relationship (log (k) = -22.0+8.45n) as in Bekins et al. (1995) was used. For turbidites a relationship with higher permeab ility was used (log (k) = -20.0+5.25n). In addition, the dcollement was assigned the same permeability-porosity relationship as for the hemipelagic sediments instead of using a hi gher permeability value as in Bekins et al. (1995) simulation. The results of the base run indicate that maximum extends from about 4 km arcward from the deformation front and laterally extends arcward with time (Figure 4-3). At all four time periods, the maximum value is contained in the area directly below the prism. During the first 0.6 million years, the maximum reached a value of 0.89 in the underthrust sediments and 0.9 in the dcollement. Both values increased by 0.03 between 0.6 and 1.3 million years. After reaching a maximum values of 0.93 in the underthrust and 0.94 in the dcollement at the end of 2 million years the maximum values decreased to 0.90 at 2.7 million years. This decrease in value with time may have result from the dissipation of pore pressures during the 2.7 million years of subduction. It may be possible th at some fluid will flow arcward within the turbidite layer (although it canno t flow out of the arcward boundary) thus, decreasing the peak pore pressures in the dcollement a nd underthrust with time. Calculations by Hanshaw and Bredehoeft (1968) shows that even if a high pore pressure region is surrounded by a low-permeability (10-21 m2) material the pore pressures cannot be maintained above 75 percent of the lithostati c stress for more than 10,000 years. Based on these calculations, the excess pressures bleed off with time because the specific storage, which is a function of the pore volum e and the compressibility of water and rock

PAGE 108

95 is so low that only a very small amount of wa ter needs to escape to drastically to lower the pore pressure (Byerlee, 1990). However, it is difficult to confirm flow in the arcward direction based on the resu lts of this study. The results of the base run has values lower than the results of Bekins et al. (1995) by 0.1 at the deformation front, while having values 0.2-0.3 higher 50 km arcward from the deformation front. Thes e differences are caused by the different parameters (permeability-porosity relationships in the dcollement and turbidite unit) used in the two simulations. The values of the base run provi des the upper limit values of within the dcollement and the underthrust sediments b ecause the sediments in the dcollement and underthrust would not lose flui d to the prism sediments above. Based on the estimated values from the base run, the values does not reach lithostatic during 2.7 million years of prism growth, and thus does not meet th e criteria for horizontal hydrofracture in the dcollement ( = 1). However, it should be noted th at it may be possible that at greater depths, addition of pressure due to smectite dehydration might be important, and may even increase pore pressu res above lithostatic. Based on the results of the base run it was not possible to reach lithostatic pore pressures resulting in horizont al hydrofracture in the dcollement. However, according to the criteria given by Behrmann (1991), the values needed for vertical hydrofracture is less than 1. Furthermore, vertical focu sing of flow has been observed by 3-D seismic images, which shows normal faults extending up ward within the turbidite unit (Zhao and Moore, 1998). If vertical hydrofracture coul d increase pore pressure s in the dcollement then it is possible for the pressures in the dcollement to reach lithostatic. To simulate

PAGE 109

96 this scenario, I used the same parameters as us ed in the base run, a nd introduced vertical hydrofractures to the model w ithin the underthrust unit. The model assigned a higher vertical permeability value (1 x 10-13 m2) relative to the background bulk permeability, to those elements that have met the vertical hydrofract ure criteria of < 0.8. Horizontal permeabilities were kept at the background bulk permeability. The hydrofractures were vertically extended to the base of the dcollement by assigning higher vertical permeability values to all elements above any el ement that meets the criteria. This allows fluids to flow towards the dcollement through the vertical hydrofractures. The values at the dcollement were examined to see whether it has reached values greater than lithostatic, allowing horizont al hydrofracturing. The estimated values in the dcollement and in the underthrust sediments were very similar to those predicted from the base r un. Results indicate that adding vertical hydrofractures in the underthru st does not increase pore pressu res in the dcollement to lithostatic, causing horizontal hyd rofracture. One possible expl anation would be that the amount of fluids entering the vertical hydr ofracture is not suffi cient to increase pore pressures to lithostatic. The criterion for vertical hydrof racture was met within the turbidite unit, which has been previously suggested as a possi ble pathway for lateral flow (e.g., Henry, 2000). It may also be possible th at pore pressures reach lithostatic pressures and dissipate at smaller time frames than th e loading step size used in this study (< 67,500 yrs).

PAGE 110

97 Figure 4-3. Simulated values for base run at 0.6, 1.3, 2.0, and 2.7 million years. Because vertical hydrofractures in the unde rthrust sediments did not increase pore pressures in the dcollement to lithostatic, an alternative to hydrofracture was tested in which the dcollement was assigned pe rmeabilities based on a bulk permeabilityeffective stress relationship. The bulk pe rmeability and vertical effective stress relationship was obtained from in-situ bulk perm eability measurements that were made at varying effective stress values in a borehole th at intersected the dco llement zone (Fisher and Zwart, 1996). This scenario tests th e possibility of episodic fluid flow in the dcollement while maintaining pore pressures th at are close to lithostatic pressures at the base of the dcollement. The relationship is given by: log kbulk = -12.3-1.6 v where v is the vertical effectiv e stress in MPa [ML-1T-2] (Fisher and Zwart, 1996). Based on this relationship, the upper extreme bulk permeability approaches 10-12 m2 when fluid pressure reaches lithostatic while the lower extreme of 10-18 m2 was predicted at

PAGE 111

98 hydrostatic pressures (Fisher and Zwart, 1996). All other sediment units were assigned permeabilities based on the permeability-poros ity relationships for hemipelagics and turbidites. The estimated maximum value of 0.87 was reached at 2 million years (Figure 4–4). During the first 0.6 million years the maximum value was 0.69 and gradually increased to 0.82 at 1.3 million years and to 0.87 at 2 million years. At 2.7 million years the maximum value decreased to 0.85 fr om 0.87. The dcollement permeability at 4 km arcward of the deformation front (Site 948) is about 1 x 10-15 m2. Due to the bulk-permeability-vertical effective stress relationship, the bulk permeability increases with an effective stress decrease. Thus, estimated at shallow depths closer to the deformation front were lower than those estimated at the arcward end. Because permeability in the dcolle ment is higher at shallow depths, the overpressured fluids will escape through the high permeability dcollement at the toe of the prism, reducing pore pressures. At Site 949C the estimated is 0.28 while the at Site 948D is 0.48. Compared to the values obtained from the long term monitoring, at Site 948D the values fairly matched while at Site 949C the value was underestimated by 0.08. The lower value at Site 949C can be explained by the higher permeabilities in the dcollement (at low eff ective stress) relative to the dcollement permeabilities at Site 948D which is located 4 km arcward of the deformation front. In comparison to the results of the base run, the values estimated here were only slightly lower. This suggests that assigning a bulk permeability -vertical effective stress relationship into the dcollement not only allows fluid to expe l at the toe of the complex where effective stress is relatively low and permeability is high, but also maintains

PAGE 112

99 relatively high pore pressures along the base of the prism thus, ma intaining its narrow taper. To investigate the combined effect of both vertical hydrofracture and the dcollement with the bulk permeability-effectiv e stress relationship the two parameters were combined. The maximum (0.87) estimated from the combined simulation is similar to the maximum estimated from bulk permeability-effective stress relationship in the dcollement and in th e underthrust (Table 4-3). Table 4-3. Summary of simulation runs with the estimated values of maximum ( max) in the underthrust and the dcoll ement. The location of the max is given seaward from the arcward end of the model during 2. 7 million years of prism growth. However, the combined simulation indicate a slight variation in the profile closer to the deformation front relative to those estimated using only the bulk permeability-effective stress relationship in the dcollement (Figure 4-4). The combined simulation predicts slightly higher values (by 0.02) of ~28 km seaward from the arcward end (Figure 4-4). This slightly high may represent the minor effect of vertical hydrofracture. The observe d sudden increase in the profiles closer to the deformation of both in the bulk permeability-effective stress simulation and the combined simulation may suggest a possible transient response. However, these transient events probably Underthrust Dcollement Run Unit log (k0) Vertical Hydrofracture (m2) kd (m2) max locati on max location Base Run Turbidites Hemipelagics 20 -22 5.25 8.44 NA NA 0.93 15.0 0.94 10.8 Bekins et al. (1995) Turbidites Hemipelagics -22 -22 8.44 8.44 NA 10-15 0.64 39.0 0.64 0.40 Vertical hydrofractur e Turbidites Hemipelagics -20 -22 5.25 8.44 10-13 NA 0.93 15.0 0.94 10.8 Varying kd Turbidites Hemipelagics -20 -22 5.25 8.44 10-13 kbulkv 0.87 10.0 0.87 12.0 Vertical hydrofractur e+ varying k d Turbidites Hemipelagics -20 -22 5.25 8.44 10-13 kbulkv 0.87 10.0 0.87 10.8

PAGE 113

100 occur at smaller time periods than the time step used in this study and thus, examining the pore pressures at smaller time steps may provide insight to possible episodic flow events in the dcollement.

PAGE 114

101 Figure 4-4. Estimated values in the dcollement at 0.6, 1.3, 2.0 and 2.7 million years of prism growth for all simulations.

PAGE 115

102 Conclusions The prism growth and flow model used in this study allows examination of pore pressure development as sediments subduct be neath the accretionary prism. Simulations with the dcollement given the same permeability as surrounding sediments predicted a maximum possible value of 0.94 in the dcolle ment. Although the predicted value was close to lithostatic pressures it was unable to reach lithostatic as speculated by previous studies, and thus would not produce horizontal hydro fracture in the dcollement zone. To test the effects of vertical h ydrofractures on episodic fluid flow, vertical hydrofractures were introduced in the underthr ust sediments based on a criteria presented by Behrmann (1991). The results indicate that vertical hydrofractures were unable to increase pore pressures in the dcollement, e ither due to lack of fluids entering the fracture increasing pore pressure s or because pore pressures reached lithostatic pressures and were dissipated within a smaller time frame compared to the loading step used in this study. Another mechanism for episodic fluid flow was tested by assigning a bulk permeability-vertical stress relationship to the dcollement. The results of this simulation indicate that excess pore pressures close to lit hostatic can be sustaine d at the base of the prism while fluid is expelled at the toe of the complex. The results suggest that the dcollement may not need to hydrofracture to ge nerate transient latera l fluid flow. With the bulk permeability-vertical effective stress relationship the dcollement could potentially propagate fluid pulses towards the toe of the complex releasing fluids at the toe. The effects of vertical hydrofractur e and assigning the bulk permeability-effective

PAGE 116

103 stress relationship in the dcollement should be further investigated at smaller time steps to examine if possible episodic ev ents occur in the dcollement.

PAGE 117

104 CHAPTER 5 SUMMARY AND CONCLUSIONS In fluid flow modeling studies sediment permeability is the most important factor that controls modeled pore pr essures. In such studies, permeability is approximated based on a systematic relationship of permeability and porosity. Marine sediments from Northern Barbados, Costa Rica, Nankai and Peru subduction zones were used in this study to examine permeability-porosity relati onships based on depositional environment, grain size distribution and st ructural domain. Results suggest moderate to high correlation between permeability and porosity for argillaceous sediments and little correlation for carbonate dominant sediment s. For argillaceous sediments, the classification based on locati on and grain size distribution provides greater correlation between permeability and porosity than the depositional environment of the sediment alone. The effects of struct ural domain on permeability-por osity relationship could not be evaluated due to limited data. Elevated fluid pore pressures play a critical role in the development of accretionary complexes, including the development of the dcollement zone. A one-dimensional loading and fluid flow model was used to simu late excess pore pressu res and porosities. Simulated excess pore pressure ratios (as a fr action of lithostatic pr essure – hydrostatic pressure) using the best-fit pe rmeability-porosity relationship were lower than predicted from previous studies. The model was also used to test the sensitivity of excess pore pressure ratios in the underthrust sediment s to bulk permeability, lateral stress in the prism, and a hypothetical low-permeability barrier at the dcollement. Results

PAGE 118

105 demonstrated significant increase in pore pressures below the dcollement with lower bulk permeability, such as obtained by using the lower boundary of permeability-porosity data, or when a low-permeability barrier is added at the dcollement. In contrast, pore pressures in the underthrust sediments demons trated less sensitivity to added lateral stresses in the prism, although the profile of the excess pore pr essure ratio is affected. Both simulations with lateral stress and a low-permeability barrier at the dcollement resulted in sharp increases in porosity at th e dcollement, similar to that observed in measured porosities. Furthermore, in both scenarios, maximum excess pore pressure ratios were found at the dcollement sugges ting that either of these factors would contribute to stable slid ing along the dcollement. Changes in fluid flow pressures and its distribution regulate the mode of deformation affecting the evolution of the accr etionary complex. To fully understand the development of pore pressures and thus, hypot hesized episodic fluid flow, one should examine the pore pressure development thr ough both space and time. A two-dimensional prism growth and flow model was used to examine the pore pressure development through the evolution of the acc retionary complex and mechanics of episodic fluid flow. Results indicate values (0.94) close to lithostatic pressures. However, estimated values were unable to reach lithostati c pressures thus not allowing horizontal hydrofracture in the dcollement zone. Vert ical hydrofractures we re introduced in the underthrust sediments to examine whether th ey would increase pore pressures in the dcollement. The results indicate that ver tical hydrofractures were unable to increase pore pressures in the dcollement zone, either due to lack of fluids entering the fracture increasing pore pressures or pore pressure s reached lithostatic pressures and were

PAGE 119

106 dissipated within a smaller time frame compare to the time step used in this study. To examine another possible mechanism for epis odic fluid flow, a bulk permeability-vertical stress relationship was assigned to the dco llement. Results indicate that excess pore pressures close to lithostatic can be sustaine d at the base of the prism while fluid is expelled at the toe of the complex. This fu rther suggest that dcollement may not need to hydrofracture initiating transient lateral flui d flow because the bulk permeability-vertical effective stress relationship could propagate fluid pulses towards the toe of the complex releasing fluids at the toe of the prism. Howe ver, the effects of ver tical hydrofracture and assigning the bulk permeability-e ffective stress relationship in the dcollement should be further investigated at smaller time steps to examine if episodic events occur in the dcollement.

PAGE 120

APPENDIX LISTING OF PERMEABILITY, PO ROSITY, GRAIN SIZE DATA WITH REFERENCES USED IN THIS STUDY

PAGE 121

108Table A-1. Listing of permeability, porosity, grain size data with references used in this study. † Grain size and carbonate d ata were assigned based on the grain size data available for simila r samples from Screaton et al. (2005) for Costa Rica Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference Barbados log k (m2) Sand (>63 m) Silt (63-4 m) Clay (<4 m) 156-948C-13X-3* Vrolijk et al. in Zwart et al. (1997) 530.4 -18.3098 4.9000E-19 0.55 Underthrust Gray claystone 0 8 92 Meyer and Fisher (1997) 156-949C-2X-1* Maltman et al. in Zwart et al. (1997) 254.08 -17.3279 4.7000E-18 0.62 Prism Gray claystone 0 9 91 Meyer and Fisher (1997) -17.3468 4.5000E-18 0.61 -17.7799 1.6600E-18 0.59 -17.7852 1.6400E-18 0.59 -17.8386 1.4500E-18 0.59 -17.8861 1.3000E-18 0.58 -17.8665 1.3600E-18 0.57 -18.0000 1.0000E-18 0.57 -18.5229 3.0000E-19 0.56 156-949C-15X-5* Bruckmann et al. (1997) 366.23 -15.0937 8.0600E-16 0.70 Prism Light olivegray claystone 0 9 91 Meyer and Fisher (1997) -16.3788 4.1800E-17 0.69 -17.4855 3.2700E-18 0.69 156-949C-19X-1* 399.2 -16.9508 1.1200E-17 0.62 Decollement Yellowish brown claystone 3 12 85 Meyer and Fisher (1997) -17.4855 3.2700E-18 0.62 -17.7878 1.6300E-18 0.61 156-949C-22X-1* 428.75 -17.1046 7.8600E-18 0.55 Decollement Light brownish gray claystone 1 7 92 Meyer and Fisher (1997) -17.4353 3.6700E-18 0.55

PAGE 122

109Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference -17.9508 1.1200E-18 0.54 110-671B-9H-2* Taylor and Leonard (1990) 76.8 -16.0836 8.2492E-17 0.57 Prism Marl 7 18 76 29.7 Taylor and Leonard (1990) 110-672A-2H-3 7.7 -14.5933 2.5508E-15 0.67 Reference site (incoming sediments) Calcareous Mud 20 35 44 35 Taylor and Leonard (1990) 110-672A-19X-3* 165.8 -15.6186 2.4064E-16 0.67 Reference site (incoming sediments) Calcareous Mud 0 33 67 0 Taylor and Leonard (1990) 110-672A-19X-3* 165.9 -16.2622 5.4674E-17 0.67 Reference site (incoming sediments) Calcareous Mud 0 33 67 0 Taylor and Leonard (1990) 110-676A-9H-3* 77.2 -15.7134 1.9348E-16 0.56 Prism Calcareous Mud 13 44 43 38.01 Taylor and Leonard (1990) 110-676A-12X-3* 105.6 -15.8234 1.5016E-16 0.49 Prism Calcareous Mud 2 23 75 22.23 Taylor and Leonard (1990) Nankai 190-1173A-22H-2* Gamage and Screaton (2003) 199.9 -16.2857 5.1793E-17 0.57 Reference site (incoming sediments) Silty clay 1 48 51 Steurer and Underwood (2003) -16.3924 4.0516E-17 0.56 -16.4064 3.9227E-17 0.55 190-1173A-31X-1* 284.59 -16.7012 1.9896E-17 0.62 Reference site (incoming sediments) Silty claystone to clayey siltstone 2 39 59 Steurer and Underwood (2003) -16.8898 1.2888E-17 0.60

PAGE 123

110Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 190-1173A-39X-5* 367.07 -17.6054 2.4809E-18 0.41 Reference site (incoming sediments) Silty claystone, moderate biotuebation 1 55 44 Steurer and Underwood (2003) -17.6594 2.1909E-18 0.36 -17.8175 1.5224E-18 0.33 190-1173A-41XCC* 388.75 -17.6026 2.4970E-18 0.45 Reference site (incoming sediments) Silty claystone, mottled due to moderate bioturbation 1 51 48 Steurer and Underwood (2003) -17.7780 1.6674E-18 0.43 -17.9266 1.1841E-18 0.43 190-1173A-46X-1* 428.59 -17.7192 1.9090E-18 0.45 Reference site (incoming sediments) Silty claystone 0 31 69 Steurer and Underwood (2003) -17.8152 1.5304E-18 0.43 -17.9036 1.2485E-18 0.41 190-1174B-42R-3* 538.23 -17.1027 7.8938E-18 0.37 Prism Silty claystone, altered ash 1 44 56 Steurer and Underwood (2003) -17.6960 2.0137E-18 0.34 -17.8152 1.5304E-18 0.32 190-1174B-59R-5* 704.95 -18.2824 5.2196E -19 0.32 Prism Silty claystone 1 32 68 Steurer and Underwood (2003) -18.4495 3.5522E-19 0.30

PAGE 124

111Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference -18.5281 2.9642E-19 0.28 190-1174B-69R-2* 795.17 -18.0686 8.5382E-19 0.29 Prism Silty claystone, bioturbated throughout, scattered worm tubes 1 37 62 Steurer and Underwood (2003) -18.2354 5.8156E-19 0.28 -18.5449 2.8514E-19 0.26 190-1174B-74R-1* 842.75 -18.1909 6.4439E-19 0.30 Underthrust Silty claystonecarbonate cemented 0 26 74 Steurer and Underwood (2003) -18.2074 6.2023E-19 0.28 -18.3246 4.7363E-19 0.27 190-1177A-25R-2* Hays (unpubl. data) 533.2 -17.7878 1.6300E-18 0.45 Reference site (incoming sediments) Silty claystone, moderate bioturbation 0 30 70 Steurer and Underwood (2003) -17.9700 1.0715E-18 0.42 -18.0850 8.2224E-19 0.40 190-1177A-46R-2* 732.54 -17.8133 1.5371E-18 0.37 Reference site (incoming sediments) Silty claystone, localcarbonate cementation 38 27 32 Steurer and Underwood (2003) -18.2880 5.1523E-19 0.36 Costa Rica

PAGE 125

112Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 170-1039B-10H-2 Saffer et al. (2000) 80.85 -16.4000 3.9811E-17 0.62 Reference site (incoming sediments) Upper hemipelagic sec: diatom ooze with ash -16.4000 3.9811E-17 0.68 -16.0000 1.0000E-16 0.75 -15.9000 1.2589E-16 0.75 170-1039B-16X-8 141.54 -17.6000 2.5119E-18 0.67 Reference site (incoming sediments) Upper hemipelagic section: calcareous clay -17.1000 7.9433E-18 0.68 -16.4000 3.9811E-17 0.76 -15.5000 3.1623E-16 0.79 -14.8000 1.5849E-15 0.80 170-1039B-26X-6 237.25 -15.5000 3.1623E-16 0.53 Reference site (incoming sediments) Pelagic section: siliceous nannofossil ooze -15.2000 6.3096E-16 0.56 -15.3000 5.0119E-16 0.58 -15.0000 1.0000E-15 0.59 -14.8000 1.5849E-15 0.61 -14.2000 6.3096E-15 0.62

PAGE 126

113Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 170-1040C-30R-4 -15.1000 7.9433E-16 0.79 Underthrust Upper hemipelagic section: silty clay stone with ash layers -15.8000 1.5849E-16 0.78 -16.8000 1.5849E-17 0.69 -16.9000 1.2589E-17 0.72 205-1253-2R-4* McKiernan and Saffer (2005) 380.07 -15.9889 1.0260E-16 0.46 Reference site (incoming sediments) Calcareous mudstones †1 49 50 57.41 Screaton et al. (2005) 205-1253-3R-2* 386.83 -16.2952 5.0680E-17 0.50 Reference site (incoming sediments) Calcareous mudstones †2 43 55 47.41 Screaton et al. (2005) 205-1253-4R-1* 394.91 -16.9024 1.2520E-17 0.47 Reference site (incoming sediments) Calcareous mudstones †8 50 42 Screaton et al. (2005) 205-1254-16R-4 366.74 -18.5810 2.6240E-19 0.39 Decollement Hemipelagic mudstones †6 39 55 1.43 Screaton et al. (2005) 205-1255-2R-CC* 134.89 -19.0337 9.2540E-20 0.32 Underthrust Hemipelagic mudstones †3 46 51 4.67 Screaton et al. (2005) -15.3177 4.8120E-16 0.50 -17.1045 7.8620E-18 0.47 -17.8170 1.5240E-18 0.44 -18.3405 4.5660E-19 0.39 -18.8283 1.4850E-19 0.32

PAGE 127

114Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 205-1255-3R-CC* 146.48 -19.0201 9.5480E-20 0.41 Underthrust Hemipelagic mudstones †1 34 65 1.59 205-1255-4R-CC* 152.38 -19.1157 7.6610E-20 0.26 Underthrust Hemipelagic mudstones †1 34 65 1.77 Screaton et al. (2005) -17.0106 9.7580E-18 0.65 -17.0708 8.4950E-18 0.65 -17.7881 1.6290E-18 0.57 170-1040C-38R-2 Screaton et al. (2005) 518 -14.8827 1.3100E-15 0.57 Underthrust Pelagic section: siliceous nannofossil chalk 1 54 46 89.66 Screaton et al. (2005) -14.9208 1.2000E-15 0.55 -14.9706 1.0700E-15 0.54 170-1040C-46R-4* 598 -14.9957 1.0100E-15 0.60 Underthrust Pelagic section: calcareous diatomite and breccia 1 58 41 69.75 Screaton et al. (2005) -15.0400 9.1200E-16 0.59 -15.0526 8.8600E-16 0.58 205-1253-02R-3* 379 -15.1656 6.8300E-16 0.66 Reference site (incoming sediments) Pelagic section: nannofossil chalk with diatom 1 49 50 57.41 Screaton et al. (2005) -15.2255 5.9500E-16 0.64 -15.2660 5.4200E-16 0.62

PAGE 128

115Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 205-1253A-03R-1* 386 -15.0177 9.6000E-16 0.71 Reference site (incoming sediments) Pelagic section: nannofossil chalk with diatom 2 43 55 47.41 Screaton et al. (2005) -15.2652 5.4300E-16 0.69 -15.3872 4.1000E-16 0.67 205-1255A-02RCC* 135 -17.5129 3.0700E-18 0.57 Underthrust Lower hemipelagic section 3 46 51 4.67 Screaton et al. (2005) -17.6216 2.3900E-18 0.56 -17.7721 1.6900E-18 0.53 205-1255A-03RCC* 147 -17.7570 1.7500E-18 0.49 Underthrust Lower hemipelagic section 1 34 65 1.59 Screaton et al. (2005) -17.7595 1.7400E-18 0.47 -17.7852 1.6400E-18 0.46 205-1255A-04RCC* 152 -17.4425 3.6100E-18 0.66 Underthrust Lower hemipelagic section 1 34 65 1.77 Screaton et al. (2005) -17.5229 3.0000E-18 0.64 -17.6421 2.2800E-18 0.59 Peru 201-1231B-3H* Gamage et al. (2005) 17.1 -16.5031 3.1400E-17 0.89 Peru Basin Diatom rich clay and Diatom ooze 22 67 11 Aiello (unpubl. data) -16.9914 1.0200E-17 0.87 -17.1694 6.7700E-18 0.86

PAGE 129

116Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference -17.4634 3.4400E-18 0.85 201-1231B-6H* 44.1 -15.8268 1.4900E-16 0.85 Peru Basin Silt with some volcanic glass 11 46 43 Aiello (unpubl. data) -16.2708 5.3600E-17 0.83 -16.6038 2.4900E-17 0.82 -16.8210 1.5100E-17 0.80 201-1231B-9H 75.7 -15.1107 7.7500E-16 0.59 Peru Basin Nannofossil ooze 54 46 0 Aiello (unpubl. data) -15.2472 5.6600E-16 0.58 -15.3420 4.5500E-16 0.57 -15.3706 4.2600E-16 0.56 201-1231B-13H 112.1 -15.3344 4.6300E-16 0.66 Peru Basin Nannofossil ooze 38 62 0 Aiello (unpubl. data) -15.3883 4.0900E-16 0.65 -15.4225 3.7800E-16 0.64 -15.4365 3.6600E-16 0.63 201-1230A-4H 31 -16.7328 1.8500E-17 0.68 Lower slope of Peru trench Diatom ooze with some clay and silt -16.7696 1.7000E-17 0.67 -16.8861 1.3000E-17 0.66 -16.9393 1.1500E-17 0.64 201-1230A-9H 70.7 -16.2403 5.7500E-17 0.68 Lower slope of Peru trench Diatom ooze with some clay and silt

PAGE 130

117Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference -16.3726 4.2400E-17 0.68 -16.4698 3.3900E-17 0.67 -16.5100 3.0900E-17 0.67 201-1230A-31X* 230.8 -16.4789 3.3200E-17 0.53 Lower slope of Peru trench Diatom ooze with some clay and silt 13.74 59.41 17.86 Aiello (unpubl. data) -16.7122 1.9400E-17 0.49 -16.7878 1.6300E-17 0.48 -16.8327 1.4700E-17 0.45 201-1230A-35X* 252.1 -16.6946 2.0200E-17 0.58 Lower slope of Peru trench Diatom ooze with some clay and silt 0.01 80.97 19.02 Aiello (unpubl. data) -16.7471 1.7900E-17 0.55 -16.8508 1.4100E-17 0.53 -16.9031 1.2500E-17 0.51 201-1227A-3H 19.3 -15.6216 2.3900E-16 0.71 Shallow Peru trench Silt -15.7471 1.7900E-16 0.69 -15.8125 1.5400E-16 0.67 -15.9469 1.1300E-16 0.64 201-1227A-12H 101.9 -15.6091 2.4600E-16 0.70 Shallow Peru trench Diatom ooze/ nannofossil/silt -15.7447 1.8000E-16 0.68 -15.7959 1.6000E-16 0.67 -15.9788 1.0500E-16 0.66

PAGE 131

118Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 201-1226B-4H 24.8 -15.1694 6.7700E-16 0.76 Equatorial pacific Nannofossil oozes and diatom oozes 18 64 18 Aiello and Kellett (unpubl. data) -15.2708 5.3600E-16 0.74 -15.3288 4.6900E-16 0.73 -15.3726 4.2400E-16 0.73 201-1226B-26H 239.6 -15.5171 3.0400E-16 0.67 Equatorial pacific Nannofossil oozes and diatom oozes 33 57 10 Aiello and Kellett (unpubl. data) -15.5143 3.0600E-16 0.67 -15.5346 2.9200E-16 0.66 -15.5436 2.8600E-16 0.65 201-1226B-43X 381.2 -16.8041 1.5700E-17 0.62 Equatorial pacific Nannofossil oozes and diatom oozes 14 67 19 Aiello and Kellett (unpubl. data) -16.8539 1.4000E-17 0.58 -16.9066 1.2400E-17 0.57 -16.9914 1.0200E-17 0.56 201-1226B-46X 409.4 -17.7328 1.8500E-18 0.58 Equatorial pacific Nannofossil oozes and diatom oozes 13 71 16 Aiello and Kellett (unpubl. data) -17.8508 1.4100E-18 0.56 -17.8761 1.3300E-18 0.53 -17.9031 1.2500E-18 0.52

PAGE 132

119Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference 201-1225A-4H 24.7 -15.4425 3.6100E-16 0.78 Equatorial pacific Nannofossil ooze and some diatom, radiolarian and foram oozes 26 62 12 Aiello and Kellett (unpubl. data) -15.5498 2.8200E-16 0.75 -15.7878 1.6300E-16 0.73 -15.7011 1.9900E-16 0.71 201-1225A-10H 83.2 -15.6180 2.4100E-16 0.68 Equatorial pacific Nannofossil and some diatom oozes 31 58 11 Aiello and Kellett (unpubl. data) -15.6757 2.1100E-16 0.66 -15.7721 1.6900E-16 0.65 -15.8447 1.4300E-16 0.64 201-1225A-26H 242.7 -15.1688 6.7800E-16 0.68 Equatorial pacific Nannofossil and diatom oozes 21 79 0 Aiello and Kellett (unpubl. data) -15.2211 6.0100E-16 0.66 -15.2807 5.2400E-16 0.65 -15.3478 4.4900E-16 0.64 201-1225A-34H 309.7 -15.6253 2.3700E-16 0.64 Equatorial pacific Nannofossil and diatom oozes 17 73 10 Aiello and Kellett (unpubl. data) -15.6596 2.1900E-16 0.63 -15.7212 1.9000E-16 0.62 -15.7423 1.8100E-16 0.61

PAGE 133

120Table A-1. Continued Location Permeability Reference Depth (mbsf) Permeability Porosity Structural domain Lithology Grain Size (wt %) Carbonate (wt%) Grain Size and Carbonate (wt%) Reference Sand (>75 m) Silt (75-5 m) Clay (<5 m) 112-679C-8H-2 Masters and Christian (1990) 68.9 -16.5394 2.8880E-17 0.69 Peru shelf Diatomaceous silty soils 9 65 26 Masters and Christian, 1990 -16.2384 5.7756E-17 0.69 -16.3176 4.8128E-17 0.69 -15.9791 1.0493E-16 0.74 -16.1135 7.7002E-17 0.74 -16.1189 7.6050E-17 0.74 -15.5394 2.8880E-16 0.79 -15.5852 2.5990E-16 0.79 -15.2384 5.7756E-16 0.81 -15.4725 3.3690E-16 0.81 -15.5394 2.8880E-16 0.81 -15.6943 2.0216E-16 0.81 -15.7378 1.8289E-16 0.81 -15.9026 1.2514E-16 0.81 112-681C-2H-3 10.3 -16.5694 2.6953E-17 0.57 Peru upper slope Diatomaceous silty soils 35 61 4 Masters and Christian, 1990 -16.4145 3.8503E-17 0.57 -16.3538 4.4279E-17 0.57 -16.6364 2.3099E-17 0.59 -16.3831 4.1390E-17 0.59 -16.3176 4.8128E-17 0.59 -16.2457 5.6794E-17 0.64 -16.1777 6.6420E-17 0.64 -16.1028 7.8922E-17 0.64

PAGE 134

121 LIST OF REFERENCES Adatia, R. H., and A. J. Maltman (2004), Data report: Initial permeability determinations on sediments from the Nankai Trough accretionary prism, ODP Sites 1173 and 1174. In Mikada, H., Moore, G.F., Taira, A., Becker, K., Moore, J.C., and Klaus, A. (Eds.), Proc. ODP, Sci. Results 190/196, 1-12 [Online]. Available from World Wide Web. [Accessed 2004-12-10] Anderson, M. P. and W. W. Woessner (1992), Applied groundwater modeling, Simulation of flow and advective trans port, Academic Press, San Diego, CA. Athy, L. F. (1930), Density, porosity, and compaction of sedimentary rocks, Am. Assoc. Pet. Geol. Bull. 14, 1-23. Aubouin, J., and R. von Huene (1985), Su mmary; Leg 84, Middle America Trench transect off Guatemala and Costa Rica, Proc. Deep Sea Drill. Project Init. Rep. 84, 939-957. Bangs, N. L.B., and S. P.S. Gulick (2005) Physical properties along the developing decollement in the Nankai Trough: infere nces from 3-D seismic reflection data inversion and Leg 190 and 196 drilling data. In Mikada, H., Moore, G.F., Taira, A., Becker, K., Moore, J.C., and Klaus, A. (Eds.), Proc. ODP, Sci. Results 190/196, 118 [Online]. Available from World Wide Web. [Accessed 2005-06-25] Bangs, N. L.B., and G. K. Westbrook (1991), Seismic modeling of the decollement zone at the base of the Barbados Ridge accretionary complex, J. Geophys. Res, 96, 38533866. Bear, J. and A. Verruijt (1987), Modeling Groun dwater Flow and Pollution, D. Reidel Publishing Company, Dordrecht, Holland. Becker, K., A. T. Fisher, and E. E. Davis (1997), The CORK experiment in Hole 949C: Long-term observations of pressure and te mperature in the Barbados accretionary prism, Proc. Ocean Drill. Program Sci. Results 156, 247-252. Behrmann, J. H. (1991), Conditions for hydr ofracture and the fluid permeability of accretionary wedges, Earth Planet. Sci. Lett ., 107, 550-558.

PAGE 135

122 Behrmann, J. H. (1992), Conditions for hydr ofracture and the fluid permeability of accretionary prisms, Earth Planet. Sci. Lett ., 107, 550-558. Bekins, B. A., and S. J. Dreiss (1992), A si mplified analysis of pa rameters controlling dewatering in accretionary prisms, Earth Planet. Sci. Lett., 109, 275-287. Bekins, B. A., A. M. McCaffrey and S. J. Dr eiss (1994), The influence of kinetics on the smectite to illite transition in the Barbados accretionary prism, J. Geophys. Res., 99, 18,147-18,158. Bekins, B. A., A. M. McCaffrey and S. J. Dreiss (1995), Episodi c and constant flow models for the origin of low-chloride wa ters in a modern accr etionary complex, W. Resour. Res., 31, 3205-3215. Bekins, B. A. and E. Screaton (in press), Pore pressures and fluid flow in the Northern Barbados Accretionary Complex, In Tim Dixon (Ed.), Proceedings from the Seismogenic Zone Theoretical Institute, Snowbird, Utah, march 2003, Columbia University Press. Bernstein-Taylor, B. L., K. S. Kirchoff-Stein, E. A. Silver and D. A. Reed (1992), Largescale duplexes within the New Britain accr etionary wedge: a po ssible example of accreted ophiolitic slivers, Tectonics 11, 732-752. Biju-Duval, B., and team (1984), Initial Reports Deep Sea Drilling Project vol. 78A, U.S. Government Printing Office, Washington. Bird, R. B., W. E. Stewart, and E. N. Lightfoot (1960), Transport Phenomena, John Wiley and Sons, Inc. Madison, WI. Boggs, S., (2001), Principles of sedime ntology and stratigraphy, pp. 378-416, New Jersey, Prentice-Hall, Inc. Bourlange, S., L. Jouniaux, P. Henry (2004) Data report: Permeability, compressibility, and friction coefficient measurements under confining pressure and strain, Leg 190, Nankai Trough. In Mikada, H., Moore, G.F ., Taira, A., Becker, K., Moore, J.C., and Klaus, A. (Eds.), Proc. ODP, Sci. Results, 190/196, 1-16 [Online]. Available from World Wide Web. [Accessed 2004-12-10] Brown, K. M., B. A. Bekins, B. Clenne ll, D. Dewhurst, and G. Westbrook (1994), Heterogeneous hydrofracture development and accretionary fault dynamics, Geology, 22, 259-262. Brown, K.M., D. Saffer, a nd B.A. Bekins (2001), Smectite diagenesis, pore-water freshening, and fluid flow at the toe of the Nankai wedge, Earth Planet. Sci. Lett., 194, 97-109.

PAGE 136

123 Brown, K. and G. K. Westbrook (1988), Mud diapirism and subcretion in the Barbados ridge accretionary complex: The role of fluid in accreti onary processes, Tectonics 7, 613-640. Bruckmann, W., K. Moran, and A. Macki llop (1997), Permeability and consolidation characteristics from Hole 949B, northern Barbados Ridge. Proc. Ocean Drill. Program Sci. Results 156, 109-114. Bryant, W. R. (2002), Permeability of cl ays, silty-clays and clayey-silts, Gulf Coast Associ. Geol. Soci. Trans., 52, 1069-1077. Bryant, W. R., R. Bennett, and C. Kather man (1981), Shear streng th, consolidation, porosity, and permeability of oceanic sediments. In Emiliani, C. (Ed.), The Sea (Vol. 7): New York (Wiley), 1555-1616. Bryant, W. R., W. Hottman, and P. Trabant (1975), Permeability of unconsolidated and consolidated marine sedi ments, Gulf of Mexico, Marine Geotechnol ., 1, 1-14. Byerlee, J. (1990), Friction, overpre ssure, and fault normal compression, Geophys. Res. Lett. 17, 2109-2112. Byrne, T., A. Maltman, E. Stephenson, W. Soh and R. Knipe (1993), Deformation structures and fluid flow in the toe region of the Nankai accretionary prism, Proc. Ocean Drill. Program Sci. Results 131, 83-95. Charlton, T. R. (1988), Tectonic erosion and accretioan in steady-state trenches, Tectonophysics 149, 233-23. COMPLEX (1999), Conference on Multiple Pl atform Exploration of the Ocean, Vancouver, British Columbia, edited by N. G. Pisias and M. L. Delaney, May. Coulbourn, W. T. (1982), Stratigraphy and structures of the Middle America Trench, Deep Sea Drilling Project Leg 67 transect off Guatemala, Proc. Deep Sea Drill Project Init. Rep., 67, 691-706. Cutillo, P.A., E. J. Screaton, and S. Ge ( 2003), Three-dimensional numerical simulation of fluid flow and heat transport with in the Barbados accretionary complex, J. Geophys. Res 108, B12, 2555 Dahlen, F. A. (1984), Noncohesive critic al Coulomb wedges: an exact solution J. Geophys. Res ., 89, 10,125-10, 133. Davis, D. M., and D. M. Hussong (1984), Geothermal observations during Deep Sea Drilling Project Leg 78A, Initial rep. Deep Sea Drill. Program 78A, 593-598. Davis, E. E., and R. D. Hyndman (1989), A ccretion and recent deformation of sediment along the northern Cascadia subduction zone, Geol. Soc. Am. Bull ., 101, 14651480.

PAGE 137

124 Davis, D. M., J. Suppe, and F. A. Dahlen ( 1983), Mechanics of fold and thrust belts and accretionary wedges, J. Geophys. Res ., 88, 949-963. DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1990), Current plate motions, Geophys. J. Int. 101, 425-478. Domenico, P.A. and F. W. Schwartz (1998) Physical and Chem ical Hydrogeology, 2nd ed., pp. 178-179, John Wiley and Sons, Inc., New York. Elliott, W. C., J. L. Aronson, G. Matisoff, and J. L. Gautier (1991), Kinetics of the smectite to illite transformation in the Denver Basin: Clay mineral, K-Ar data, and mathematical model results, Am. Assoc. Pet. Geol. Bull., 75 436-462. Ferguson, I.J., Westbrook, G.K., Langseth, M.G., Thomas, G.P. (1993), Heat flow and thermal models of the Barbados ridge accretionary complex, J. Geophys. Res, 98, B3, 4121-4142 Fetter, C. W. (1994), Applied Hydrog eology, Prentice Hall, New Jersey. Fisher, D., and T. Byrne (1987), Structural e volution of underthrusted sediments, Kodiak Islands, Alaska, Tectonics 6, 775-793. Fisher, D., and T. Byrne (1990), The character and distribution of mine ralized fractures in the Kodiak Formation, Alaska: implica tions for fluid flow in an underthrust sequence, J. Geophys. Res., 95, 9069-9080. Fisher, A. T., and M. W. Hounslow (1990), Tr ansient fluid flow through the toe of the Barbados accretionary complex: Constr aints from Ocean Drilling Program Leg 110 heat flow studies and simple models, J. Geophys. Res., 95, 8845-8858. Fisher, A. T., and G. Zwart (1996), Relati on between permeability and effective stress along a plate boundary fault, Barbados accretionary complex, Geology 24, 307310. Fitts, T. G., and K. M. Brown (1999), Stress -induced smectite dehydration: ramifications for patterns of freshening and fluid expul sion in the N. Barbados accretionary wedge, Earth Planet. Sci. Lett., 172, 179-197. Foucher, J. P., X. Le Pichon, S. Lallemant, M. A. Hobart, P. Henry, M. Benedetti, G. K. Westbrook, and M. G. Langseth (1990), Heat flow, tectonics a nd fluid circulation at the toe of the Barbados Ridge accretionary prism, J. Geophys. Res., 95 88598867. Foucher, J. P., P. Henry, and F. Harmegni es (1997), Long–term observations of pressure and temperature in ODP Hole 948D, Barbados accretionary prism, Proc. Ocean Drilling Program Sci. Results 156, 229-238.

PAGE 138

125 Fryer P., J. Pearce, L. Stokking, J. Ali, R. Ar culus, D. Ballotti, M. Burke, G. Ciampo, J. Haggerty, R. Haston, D. Heli ng, M. Hobert, T. Ishii, L. Johnson, Y. Lagabrielle, H. Maekawa, M. Marlow, F. McCoy, G. Milner M. Mottl, B. Murton, S. Phipps, C. Rigsby, K. Saboda, B. Stabell, S. Van Der Laan, and Y. Xu (1990), Proceedings of the Ocean Drilling Program, Initial Reports 125. ODP Leg 125 Shipboard Scientific Party, College Station, TX. Gamage, K., B. Bekins, and E. Screaton ( 2005), Data report: Permeabilities of eastern equatorial Pacific and Peru margin sediments. In Jrgensen, B.B., D'Hondt, S.L., Miller, D.J., et al. (Eds .), Proc. ODP, Sci. Results 201 [Online]. Available from World Wide Web: . [Accessed 2005-07-11] Gamage, K., and E. Screaton (2003), Data re port: Permeabilities of Nankai accretionary prism sediments. In Mikada, H., Moore, G. F., Taira, A., Becker, K., Moore, J.C., and Klaus, A. (Eds.), Proc. ODP, Sci. Results 190/196, 1-22 [Online]. Available from World Wide Web. [Accessed 2004-06-10] Gamage, K., and E. Screaton (in press), Char acterization of excess pore pressures at the toe of the Nankai accretionary complex, ocean drilling program sites 1173, 1174, and 808: results of one-dimensional modeling J. Geophys. Res Gieskes, J. M., T. Gamo, and M. Kastner ( 1993), Major and minor element geochemistry of interstitial waters of Site 808, Nankai Trough: An overview, Proc. Ocean Drill Program Sci. Results 131, 387-395. Gretener, P. E. (1976), Pore pressure: F undamentals of general ramifications and implications for structural geology in AAPG Continuing Educa tion Course Note Series #4, American Association of Pe troleum Geologists, Tulsa, Okla., 87. Hampel. A. (2002), The migration history of the Nazca Ridge along the Peruvian active margin: a re-evaluation, Earth Planet. Sci. Lett 203, 665-679. Hanshaw, B. B., and J. D. Bredehoeft ( 1968), On the maintenance of anomalous fluid pressures: II Source layer at depth, Geol. Soc. Am. Bull ., 79, 1107-1122. Hatcher, R. D. (1995), Structural Geology pp. 224, Prentice Hall, New Jersey. Henry, P. (2000), Fluid flow at the toe of the Barbados accretionary wedge constrained by thermal, chemical, and hydrologic observations and models, J. Geophys. Res., 105 25,855-25,872. Henry, P., L. Jouniaux, E. Screaton, S. Hunze and D. Saffer (2003), Anisotropy of electrical conductivity record of initial strain at th e toe of the Nankai accretionary wedge, J. Geophys. Res ., 108, doi:10.1029/2002JB002287.

PAGE 139

126 Henry, P., and C.-Y. Wang (1991), Modeling of fl uid flow and pore pressure at the toe of Oregon and Barbados accretionary wedges, J. Geophys. Res., 96 20,109-20,130. Hovland, M. (1990), Do carbonate reefs form due to fluid seepage, Terra Nova, 2, 8-18. Hubbert, M. K., and W. W. Rubey, (1959), Role of fluid pressure in the mechanics of overthrust faulting, I, Mechanics of fluid-fi lled porous solids a nd its application to overthrust faulting, Geol. Soc. Am. Bull ., 70, 115-166. Jaeger, J. C., and N. G. W. Cook, (1969), Fundamentals of Rock Mechanics, pp. 87-91, Methuen, London. Kastner, M., and H. Elderfield, W. J. Je nkins, J. M. Gieskes and T. Gamo (1993), Geochemical and isotopic evidence for fl uid flow in the western Nankai subduction zone, Japan, Proc. Ocean Drill. Program Sci Results, 131, 397-413. Kastner, M., H. Elderfield, and J. B. Martin (1991), Fluids in c onvergeny margins: What do we know about their composition, origin, role in diagenesis and importance for oceanic chemical fluxes, Philos. Trans. R. Soc. London, Ser. A 335, 275-288. Kemerer, A., Screaton, E. (2001), Permeabilitie s of sediments from Woodlark Basin: implications for pore pressures. In Huc hon, P., Taylor, B., Klaus, A. (Eds.), Proc. ODP, Sci. Results 180, 1-14 [Online]. Available from World Wide Web. Kirkup, L (2002), Data Analysis with Excel: An Introduction for Physical Scientists, pp. 213-360, Cambridge University Press. Knipe, R. J., S. M. Agar, and, D. J. Prio r (1991), The microstructural evolution of fluid flow paths in semi0lithified sediments from subduction complexes Philos. Trans. R. Soc. London, Ser. A 335, 261-273. Koltermann, C., and M. Gorelick (1995) Fractional packing model for hydraulic conductivity derived from sediment mixtures, Water Resour. Res ., 31(12), 32833297. Labaume, P., M. Kastner, A. Trave, and P. Henry (1997), Carbonate veins from the decollement zone at the toe o the N. Barb ados accretionary prism: Microstructure, mineralogy, geochemistry, and relations w ith prism structures and fluid regime, Proc. Ocean Drill. Program Sci. Results, 156 79-96. Langseth, M. G., and E. A. Silver (1996), The Nicoya convergent margin-A region of exceptionally low heat flow, Geophy. Res. Lett ., 23, 891-894. Langseth, M. G., G. K. Westbrook, and M. Hobart (1990), Cont rasting geothermal regimes of the Barbados Ridge accretionary complex, J. Geophys. Res., 95, 88298844.

PAGE 140

127 Le Pichon, X., T. Iiyama, J. Boulgue, J. Char vet, M. Faure, K. Kano, S. Lallemant, H. Okada, C. Rangin, A. Taira, T. Urabe, and S. Uyeda (1987), Nankai Trough and Zenisu Ridge: a deep-sea submersible survey, Earth Planet. Sci. Letters 83, 285299. Le Pichon, X., and P. He nry (1992), Erosion and accre tion along subduction zones: a model of evolution, Koninklijke Nederlandse Ak ademie van Wetenschappen Proceedings 95, 297-310. Lide, D. R. (Ed.) (2000), Handbook of Chem istry and Physics, Chemical Rubber Publishing Company, pp. 12-13, Boca Raton, Florida. Lowe, J., P. Zaccheo, H. Feldman (1964), C onsolidation testing with backpressure, J. Geotech. Eng. Div., Am. Soc. Civ. Eng ., 90, 69-86. Maltman, A., T. Byrne, D. Karig and S. La llemant (1992), Structural geological evidence from ODP Leg 131 regarding fluid flow in the Nankai prism, Japan, Earth Planet. Sci. Lett., 109, 463-468. Mascle, A., Moore, J. C., and Sh ipboard Scientific Party (1988), Proc. ODP, Init. Repts (Pt. A), 110: College Station, TX (Ocean Drilling Program). Masters, J. C., and H. Christian, (19 90), Hydraulic conductivity of diatomaceous sediment from the Peru continental ma rgin obtained during ODP Leg 112. In Suess, E., von Huene, R., et al ., Proc. Ocean Drill. Program Sci. Results 112, 633638. McKiernan, A.W., and D. M. Saffer (in press), Permeability and consolidation properties of subducting sediments off Costa Rica, ODP Leg 205. In Morris, J.M., Villinger, H.W., and Klaus, A. (Eds.), Proc. ODP, Sci. Results, 205 1-## [Online]. Available from World Wide Web: . [Accessed 2005-05-30]. Meyer, A. and A. Fisher (1997), Proc. Ocean Drill. Program Sci. Results 156, 337-341. Moore, J. C. (1989), Tectonics and hydrogeology of accretionary prisms: Role of the decollement horizon, J. Struct. Ge ol., 11, 95-106. Moore, J. C., B. Biju-Duval and team (1982) Offscraping and undert hrusting of sediment at the deformation front of the Barba dos Ridge: Deep Sea Drilling Project Leg 78A, Geol. Soc. Am. Bull., 93 1065-1077. Moore, J. C., A. Klaus, and Sh ipboard Scientific Party (1998), Proceedings of the Ocean Drilling Program, Initial Reports vol. 171A, Ocean Drill. Program, College Station, Tex.

PAGE 141

128 Moore, J.C., G. F. Moore, G. R. Cochra ne, and H. J. Tobin (1995), Negative-polarity seismic reflections along faults of the Oregon accretionary prism: Indicators of overpressuring, J. Geophys. Res., 100 12,895-12,906. Moore, J.C. and Saffer, D. (2001), Updip limit of the seismogenic zone beneath the accretionary prism of southwest Japan: An effect of diagenetic to low-grade metamorphic processes and increasing effective stress, Geology 29.183-186. Moore, J. C. and E. Silver (2004), Fluid flow in accreting a nd eroding convergent margins, In K. Becker et al. (Eds.), Achievements and Opportun ities of Scientific Ocean Drilling, Joides Journal 28, 91-96. Moore, G.F., A. Taira, N. L. Bangs, S. Kuramoto, T. H. Shipley, C. M. Alex, S. S. Gulick, D. J. Hills, T. Ike, S. Ito, S. C. Leslie, A. J. McCutcheon, K. Mochizuki, S. Morita, Y. Nakamura, J. O. Park, B. L. Taylor, G. Toyama, H. Yagi, and Z. Zhao (2001), Data report: Structural setting of the Leg 190 Muroto transect. In Moore, G.F., Taira, A., Klaus, A., et al., Proc. ODP, Init. Repts ., 190 [Online]. Available from World Wide Web: . [Accessed 2005-06-15] Moore, J. C., and P. Vrolijk (1992), Fluids in accretionary prisms, Rev. Geophys ., 30, 113-135. Muller, R. D. and W. H. F. Smith (1993), De formation of the ocean ic crust between the north American and South American plates J. Geophys. Res ., 98, 8275-8291. Neuzil, C. E. (1986), Groundwater flow in low-permeability environments, Water Resour. Res ., 22, 1163-1195. Neuzil, C. E. (1994), How permeable are clays and shale?, Water Resour. Res., 30, 145150. Olsen, H. W., R. W. Nichol s and, T. L. Rice (1985), Low gradient permeability measurements in a triaxial system, Geotechnique 35, 145-157. Peacock, S. M. (1990), Fluid processes in subduction zones, Science 248, 329-337. Perry. E., and J. Hower (1970), Burial diagen esis in Gulf Coast pelitic sediments, Clays Clay Miner ., 18, 165-177. Price, R. (1988), The mechanical paradox of large overthrusts, Geol. Soc. Am. Bull ., 100, 1898-1908. Robertson, E. C. (1967), Labor atory consolidation of carbonate sediments. In Richards, A.F. (Ed.), Marine Geotechnique : Urbana (Univ. Illinois Press), 118-127.

PAGE 142

129 Saffer, D. M. (2003), Pore pressure de velopment and progressive dewatering in underthrust sediments at the Costa Ri can subduction margin: comparison with northern Barbados and Nanaki, J. Geophys. Res ., 108, doi:10.1029/2002JB001787. Saffer, D., and B. A. Bekins (1998), Episodi c fluid flow in the Nankai accretionary complex: Timescale, geochemistry, flow rates and fluid budget, J. Geophys. Res ., 103, 30,351-30,370. Saffer, D., and B. A. Bekins (1999), Fl uid budgets at convergent plate margins: Implications for the extent and duration of fault-zone dilation, Geology 27, 10951098. Saffer, D., and B. A. Bekins (2002), Hydrologic controls on the morphology and mechanics of accretionary wedges, Geology 30(3), 271-274. Saffer, D., and A. McKiernan (2005), Permeab ility of underthrust sediments at the Costa Rican subduction zone: Scale dependen ce and implications for dewatering, Geophys. Res. Lett ., 32, doi:10.1029/2004GL021388. Saffer, D., E. Silver, A. Fisher, H. Tobin, and K. Moran, (2000), Inferred pore pressures at the Costa Rica subduction zone: imp lications for dewatering processes. Earth Planet. Sci. Lett. 177, 193-207. Scholz, C. H. (1998), Eart quakes and friction laws, Nature 391, 37-42. Screaton, E. J., B. Carson, and G. P. Lennon (1995), Hydrogeological properties of a thrust fault with in the Oregon accretionary prism: J. Geophys. Res ., 100, 20,02520,035. Screaton, E. J., A. T. Ficher, B. Cars on, and K. Becker ( 1997), Barbados Ridge hydrogeologic tests: Implica tions for fluid migration along an active decollement, Geology 25, 239-242. Screaton, E. J., and S. Ge (1997), An assessm ent of along-strike flui d and heat transport within the Barbados ridge accretionary co mplex: Results of preliminary modeling, Geophys. Res. Lett ., 24, 3085-3088. Screaton, E. J., and S. Ge (2000), Anomalous ly high porosity in the proto-decollement zone of the Barbados accretionary co mplex: Do they indicate overpressures?, Geophys. Res. Lett ., 27, 1993-1996. Screaton, E. J., T. Hays, K. Gamage and J. Martin (in press), Data report: Permeabilities of Costa Rica subduction zone sediments. In Morris, J.M., Villinger, H.W., and Klaus, A. (Eds.), Proc. ODP, Sci. Results 205, 1-## [Online]. Available from World Wide Web: . [Accessed 2005-05-30].

PAGE 143

130 Screaton, E. J., D. Saffer, P. Henry, S. Hunze and Leg 190 Shipboard Scientific Party (2002), Porosity loss within the underthrus t sediments of the Nankai accretionary complex: Implications for overpressures, Geology 30, 19-22. Seno, T., S. Stein, and A. E. Gripp (1993), A model for the motion of the Philippine Sea Plate consistent with NUVEL-1 and geological data J. Geophys. Res ., 98, 17,94117, 948. Shi, Y., and C.-Y. Wang (1985), Pore pressure generation in front of the Barbados Ridge Complex, Geophys. Res. Lett ., 12, 773-776. Shi, Y., and C.-Y. Wang (1988), Generation of high pore pressure s in accretionary prisms: Inferences from the Barbados subduction complex, J. Geophys. Res. 93, 8893-8910. Shi, Y., and C.-Y. Wang (1994), Models study of advective heat flow associated with compaction and dehydration in accretionary prisms, J. Geophys. Res ., 99, 93199325. Shipboard Scientific Party (1991), Site 808, In Taira, A., I. Hill, J. V. Firth Initial Reports, Ocean Drilling Program Leg 131, College Station, Texas, Ocean Drilling Program, pp. 71-269. Shipboard Scientific Party ( 1998), Introduction to logging-while-d rilling invest igations of faulting, fluid flow, and seismic images of the northern Barbados subduction zone. In Moore, J.C., Klaus, A., et al., Proc. ODP, Init. Repts., 171A: College Station, TX (Ocean Drilling Program), 5-10. Shipboard Scientific Party (2001), Leg 190 Pr eliminary Report: Deformation and fluid flow processes in the Nankai Trough accretionary prism, [Online] . Shipboard Scientific Party, (2003), Leg 201 su mmary. In D'Hondt, S.L., Jrgensen, B.B., Miller, D.J., et al., Proc. ODP Init. Repts., 201 [Online]. Available from World Wide Web: http://www-odp.tamu.edu/ publications/201_IR/chap_01/chap_01.htm. [Accessed 2004-12-15] Shipley, T. H., G. F. Moore, N. L. Bangs J. C. Moore, and P. L. Stoffa (1994), Seismically inferred dilatancy distributi on, northern Barbados Ridge decollement: Implications for fluid migr ation and fault strength, Geology, 22 411-414. Shipley, T.H., Y. Ogawa, P. Blum, and J. M. Bahr (Eds.), (1997 ), Proc. ODP, Sci. Results 156 [Online]. Available from World Wide Web: http://wwwodp.tamu.edu/publications/156_SR/156TOC.HTM. [Accessed 2004-12-15]

PAGE 144

131 Sibuet, M., A. Fiala, J. P. Foucher, and S. Otha (1990), Spatial distribution of clams colonies at the toe of the Nanakai accret ionary prism near 138 E, International Conference of Fluids in Subduction Zones, Abstracts Volume, Center National de la Recherche Scientifique, Paris. Silver, E.A., G. Kimura, P. Blum, and T. H. Shipley (Eds.), (2000), Proc. ODP, Sci. Results, 170 [Online]. Available from World Wide Web: http://wwwodp.tamu.edu/publications/170_SR/170sr.htm. [Accessed 2005-02-15] Spivack, A. J., M. Kastner and B. Ransom (2002), Elemental and isotopic chloride geochemistry and fluid flow in the Nankai Trough, Geophys. Res. Lett ., 29, 10.1029/2001GL014122. Stauffer, P. and B. A. Bekins (2001), Mode ling consolidation and dewatering near the toe of the northern Barbados accretionary complex, J. Geophys. Res., 106 6369-6383. Steurer, J. F., and M. Underwood (2003), Data report: The relation between physical properties and grain-size variations in hemipelagic sediments from Nankai Trough. In Mikada, H., Moore, G.F., Taira, A., Becker, K., Moore, J.C., and Klaus, A. (Eds.), Proc. ODP, Sci. Results 190/196, 1-25 [Online]. Available from World Wide Web. [Accessed 2004-12-10] Stow, D. A. V., and D. J. W. Piper (1984), Deep-water fine -grained sediments: Facies models, in Fine-grained sediments: Deep water processes and facies, edited by D. A. V. Stow and D. J. W. Piper, pp. 6 11-646, The geological Society, Blackwell, Oxford. Taylor, E., and A. Fisher (1993), Sediment pe rmeability at the Nankai accretionary prism, Site 808, Proc. Ocean Drill Program Sci. Results, 131, 235-245. Taylor, E., and J. Leonard (1990), Sediment consolidation and permeability at the Barbados forearc, In Proc. Ocean Drill Program Sci. Results, 110, 289-308. Terzaghi, K and R. B. Peck (1940), Soil Mech anics in Engineering Practice, 2nd ed. John Wiley and Sons, New York, 729. Ujiie, K., T. Hisamitsu and A. Taira (2003), Deformation and fluid pressure variation during initiation and evolution of the plate boundary decollement zone in the Nankai accretionary prism, J. Geophys. Res ., 108, doi:10.1029/2002JB002314. Underwood, M. B., and X. Deng (1997), Clay mineralogy and clay geochemistry in the vicinity of the decollement z one, northern Barbados ridge, Proc. Ocean Drill. Program, Sci. Results 156, 3-30.

PAGE 145

132 Underwood, M. B., K. Pickering, J. M. Gieske s, M. Kastner, and R. Orr (1993), Sediment geochemistry, clay mineralogy and diagenes is: A synthesis of data from Leg 131, Nankai Trough, Proc. Ocean Drill. Program Sci. Results 131, 343-363. Von Huene, R., and D. W. Scholl (1991), Obse rvations at convergent margins concerning sediment subduction, subduction erosion, a nd the growth of continental crust, Rev. Geophys ., 29, 279-316. Voss, C. I. (1984), A finite element simula tion model for saturated-unsaturated, fluid density-dependent groundwater flow with energy transport or chemically-reactive single species solute transport, U.S. Geol. Surv. Water Resour. Invest ., 84-4369. Vrolijk, P. J. (1987), Tectonically-driven fluid flow in the Kodiak accretionary complex, Alaska, Geology 15, 466-469. Vrolijk, P. J., G. Myers, and J. C. Moore ((1988), Warm fluid migration along tectonic mlanges in the Kodiak accr etionary complex, Alaska, J. Geophys. Res., 93, 10,313-10,324. Westbrook, G. K., J. W. Ladd, P. Buhl, N. Bangs, and G. J. Tiley (1988), Cross section of an accretionary wedge: Barbados Ridge Complex, Geology 16, 631-635. Westbrook, G. K. and M. J. Smith (1983), Long decollement and mud volcanoes; evidence from the Barbados Ridge complex for the role of high pore-fluid pressure in the development of an accretionary complex, Geology 11, 279-283. Yeung, K., C. Chakrabarty, and X. Zhang (1993), An approximate analytical study of aquifers with pressure-sensi tive formation permeability: Water Resour. Res ., 29, 3495-3501. Zwart, G., W. Bruckmann, K. Moran, A. MacK illop, A. Maltman, A. Bolton, P. Vrolijk, T. Miller, M. Gooch and A. Fisher (1 997), Evaluation of hydrogeologic properties of the Barbados accretionary prism: a synthesis of Leg 156 results. In Shipley, T. H., Ogawa, Y., Blum, P., and Bahr, J.M. (Eds.), Proc. Ocean Drill. Program Sci. Results 156, 303-310.

PAGE 146

133 BIOGRAPHICAL SKETCH Kusali Gamage was born in April 27, 1974, in Colombo, Sri Lanka. She earned her Bachelor of Science (1997) and Master of Science (1999) degrees from Bowling Green State University of Ohio. She continued her education at the University of Florida, where she received her PhD in December of 2005.


Permanent Link: http://ufdc.ufl.edu/UFE0013080/00001

Material Information

Title: Permeabilities of Subduction Zone Sediments and Their Effect on Pore Pressure Generation
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013080:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013080/00001

Material Information

Title: Permeabilities of Subduction Zone Sediments and Their Effect on Pore Pressure Generation
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013080:00001


This item has the following downloads:


Full Text












PERMEABILITIES OF SUBDUCTION ZONE SEDIMENTS AND THEIR EFFECT
ON PORE PRESSURE GENERATION















By

KUSALI R. GAMAGE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Kusali R. Gamage

































To my parents and all my teachers.















ACKNOWLEDGMENTS

I would like to express my sincere gratitude to my advisor, Dr. Elizabeth Screaton,

for her continuous support and encouragement throughout my graduate studies. She

guided me through the dissertation process, never accepting less than my best efforts.

Her technical and editorial advice was essential to the completion of this dissertation and

has taught me innumerable lessons and insights on the workings of academic research in

general. I cannot thank her enough for bearing my complaints and procrastination with

unbelievable patience. My thanks also go to the members of my committee, Drs. David

Foster, Jonathan Martin, Louis Motz, and Douglas Smith, who have patiently sat through

committee meetings and taken time to understand my work. I also thank them for

reading previous drafts of this dissertation and providing many valuable comments that

improved the presentation and contents of this dissertation.

Many thanks go to Dr. Barbara Bekins of the USGS, Menlo Park, California, for

providing core samples and suggesting working on the Peru permeability measurements

and Dr. Ivano Aiello of the Moss Landing Marine Laboratories, California, for providing

me with grain size data relating to this research. I am also very grateful to Kevin Hartl

for his invaluable technical support, Dr. John Jaeger and William Vienne for helping me

with grain size analyses and Dr. Jason Curtis for carbon analyses. My special thanks go

to my colleagues Troy Hays, George Kamenov, Jennifer Martin, Victoria Mejia and

many friends out in the real world for all their support and encouragement throughout my

graduate studies.









Last, but not least, I would like to thank my husband, Sanjaya, for his

understanding and love during the past few years. His encouragement was in the end

what made this dissertation possible. My parents, Punyasena and Kalyani Gamage, and

my brother, Rachita Gamage, receive my deepest gratitude and love for their dedication

and the many years of support during my undergraduate studies that provided the

foundation for this work.
















TABLE OF CONTENTS

pae

A C K N O W L E D G M E N T S ................................................................................................. iv

L IST O F T A B L E S ............. ........ .. ........................................... .. ...... ....... ix

LIST OF FIGURES ................................. ...... ... ................. .x

ABSTRACT ........ .............. ............. .. ...... .......... .......... xii

CHAPTER

1 GENERAL INTRODUCTION ..................................................... .....................

Deformation Processes of Subduction Zones......................................................2
Role of the Decollem ent Zone............................................................ ...........3
C critical T aper T heory ................................................... ............................. 4
Role of Fluid Flow at Convergent Margins..................................................6
Evidence for Fluid Flow and Pressures............... ................................................6
Basics of Fluid Flow Related to Subduction Zone................. ..............................10
P erm eab ility ...................................... ............................ ................ 1 1
Hydrogeologic Modeling........................................................... 12
Statement of Problem ............. ....... .......... .................. 13

2 A COMPARATIVE STUDY OF PERMEABILITY MEASUREMENTS FROM
THE SUBDUCTION ZONES OF NORTHERN BARBADOS, COSTA RICA,
N A N K A I, A N D P E R U .................................................................... ..................... 16

Introduction..................................... .................................. ........... 16
B background .............................................................................................................18
B arbados ............................................................................................. ........18
C o sta R ic a ...................................................................................................... 1 9
N a n k a i .......................................................................................2 0
P e r u ................................................................................................................ 2 2
Laboratory Perm ability Data..................................................................23
B arbados ................................................................................................... ....... 24
N ankai and P eru .............................................................2 5
C o sta R ic a ...................................................................................................... 2 6
Permeability-Porosity Relationship .......... ..... ........................ ........... 27
D description of Statistical M ethods................................. .................. 27









R results ................ ......... ......... ..... ............ ............ 28
Effects of Depositional Environment .................. ...............................29
Effects of G rain Size .................................... ................... ..... .... 34
E effects of Structural D om ain ................................................................... ... ..38
D iscu ssio n ...................................... ................................................. 4 0
C o n clu sio n s..................................................... ................ 4 2

3 CHARACTERIZATION OF EXCESS PORE PRESSURES AT THE TOE OF
THE NANKAI ACCRETIONARY COMPLEX, OCEAN DRILLING
PROGRAM SITES 1173, 1174, AND 808: RESULTS OF ONE-
D IM EN SION A L M O D ELIN G .................................................................................44

In tro d u ctio n .......................................................................................4 4
B a c k g ro u n d .....................................................................................................4 5
G eologic Setting .............................................................................. 45
Previous H ydrologic Studies ........................................ .......................... 48
Laboratory Permeability Measurements .............. .............................................52
M modeling M methods .................. ................................ .. .... ...... .............. 55
T heoretical B background ........................................................... .....................55
M odel Im plem entation ........................ ....................... ............... ... 58
Model Dimensions, Boundary Conditions, and Initial Conditions ...................59
Sedimentation and Prism Thickening Rates ....................................................61
R e su lts ...................................... .................................................... 6 2
M o d el R e su lts .................................................................................. 6 2
Sensitivity to B ulk Perm eability................................... .................................... 65
Sensitivity to Lateral Stress ....................................................... .... ........... 67
Sensitivity to a Low-Permeability Barrier.................. ... ...............70
Im p location s ........................................................................... 7 3
C o n clu sio n s..................................................... ................ 7 4

4 EVOLUTION OF HIGH PORE PRESSURES AND IMPLICATIONS FOR
EPISODIC FLUID FLOW AT THE NORTHERN BARBADOS
A CCRETION AR Y CO M PLEX ................................................................................76

In tro d u ctio n .......................................................................................7 6
B background ...............7.. ...................9.............................
Fluid Flow and Pore Pressures ........................................ ........................ 82
Hydrofractures ................................. .......................... .... ...... 85
M modeling M methods .................. ....................................... .. .......... 86
M odel Im plem entation ............................................... ............................. 86
M odel E qu action s ............... ............... .. .................... .......... ................. .. .. 87
Model Dimensions, Boundary Conditions, and Initial Conditions.............................88
Sedimentation and Loading Rates .............................................. ...............91
R results and D iscu ssion .............................. ........................ .. ...... .... ...... ...... 93
C o n clu sio n s.................................................... ................ 10 2

5 SUMMARY AND CONCLUSIONS ........... ................................. ...............104









APPENDIX: LISTING OF PERMEABILITY, POROSITY, GRAIN SIZE DATA
WITH REFERENCES USED IN THIS STUDY .................................................107

L IST O F R E F E R E N C E S ...................................................................... ..................... 12 1

BIOGRAPHICAL SKETCH ............................................................. ...............133
















LIST OF TABLES


Table page

2-1 Log linear permeability-porosity relationships predicted for varying lithologies
at Barbados, Costa Rica, Nankai and Peru .................................... ............... 33

2-1 Permeability-porosity relationships based on grain size analyses. ........................35

3-1 A summary of laboratory measured permeabilities for samples from ODP Leg
190 Sites 1173 and 1174 ............................ .............. .................. .... ...........53

3-2 Fluid and solid matrix properties used for numerical simulations.........................61

3-3 Summary of sedimentation rates calculated from biostratigraphy at Sites 1173,
1174 and 808 and prism thickening rates calculated from prism geometry and
convergence rate ........................................... ...... ................ ...... 62

3-4 Summary of simulation runs at Site 1174 and 808. .............................................64

4-1 Fluid and solid matrix properties used for numerical simulations.........................91

4-2 Summary of sedimentation rates calculated from biostratigraphy at Sites 1173,
1174 and 808. ..........................................................................92

4-3 Summary of simulation runs with the estimated values of maximum *.................99

A-i Listing of permeability, porosity, grain size data with references used in this
stu dy .............................................................................108
















LIST OF FIGURES


Figure pa

1-1 Diagram of accretionary prism and the processes of frontal accretion and
u n d e rp latin g ..............................................................................................................3

1-2 Schematic of an ideal coulomb wedge. ............................... ............................... 5

2-1 Map of large-scale regional setting location of the drill sites at the Barbados
accretionary com plex .................................................................... ...................18

2-2 Map showing location of ODP drilling along the Costa Rica subduction zone. ....20

2-3 Location map of ODP Leg 190 and previous ODP/DSDP drill sites in the
N ankai T rough .................................................... ................. 2 1

2-4 Map showing general locations of drill sites occupied during ODP Legs 138
and 112 at Peru subduction zone. ................................ ................................. 23

2-5 Plot of laboratory derived permeability measurements from Barbados, Costa
Rica, Nankai, and Peru subduction zones. .................................... .................29

2-6 Permeabilities classified based on depositional environment and location. ...........33

2-7 Permeabilities classified based on grain size distribution. ...................................37

2-8 Permeabilities classified based on structural domain. ........................................39

3-1 Location map of the study area in the Nankai accretionary complex and sites
used for this study. ........................................... ... .... ........ ......... 46

3-2 Schematic interpretation of the Muroto Transect showing tectonic domains and
location of Leg 190 drill sites used in this study...................................................47

3-3 Porosity profiles of Sites 808, 1174, and 1173. ...............................................50

3-4 Permeability data measured for samples from ODP Leg 190................................54

3-5 Simulated porosity and k* profiles for base run at Site 1173 ................................65

3-6 Simulated porosity and k* profiles with varying bulk permeability at Sites 1174
and 808. ..................................................................................66









3-7 Simulated porosity and k* profiles with added lateral stress at Sites 1174 and
80 8. .......................................................................................6 8

3-8 Simulated porosity and k* profiles with added low-permeability barrier at Sites
1174 and 808. .............................................................................7 1

4-1 Location map and cross-section of the Barbados accretionary complex. ...............80

4-2 Grid and boundary conditions for the prism growth/flow model..........................90

4-3 Simulated k* values for base run at 0.6, 1.3, 2.0, and 2.7 million years. ................97

4-4 Estimated k* values in the decollement at 0.6, 1.3, 2.0 and 2.7 million years of
prism growth for all sim ulations. ........................................ ....................... 101
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

PERMEABILITIES OF SUBDUCTION ZONE SEDIMENTS AND THEIR EFFECT
ON PORE PRESSURE GENERATION

By

Kusali R. Gamage

December 2005

Chair: Elizabeth Screaton
Major Department: Geological Sciences

Permeability is a fundamental sediment property influencing fluid flow, hence fluid

pressures in the subsurface. Because elevated fluid pore pressures play a critical role in

the development of accretionary complexes, including the development of the

decollement zone, it is important to simulate pore pressures based on a systematic

relationship of permeability and porosity. Based on laboratory permeability

measurements of Northern Barbados, Costa Rica, Nankai and Peru subduction zones

sediments, a high correlation between permeability and porosity was found for

argillaceous sediments while little correlation was found for carbonate dominant

sediments. Classification based on location and grain size distribution provided greater

correlation between permeability and porosity than the depositional environment of the

sediment alone. In the second part of the research, a one-dimensional loading and fluid

flow model near the toe of the Nankai subduction zone was used to examine the effects of

lower bulk permeability (sensitivity to a permeability-porosity relationship), lateral stress









in the prism, and addition of a low-permeability barrier to the decollement. The results

predicted significant increase in pore pressures below the decollement zone with lower

bulk permeability, or when a low-permeability barrier is added at the decollement. Both

simulations with lateral stress and a low-permeability barrier at the decollement resulted

in sharp increases in porosity at the decollement, similar to that observed in measured

porosities. In addition, these two scenarios predict maximum excess pore pressure ratios

at the decollement suggesting that either of these factors would contribute to stable

sliding along the decollement. In the third part of the research, results from a two-

dimensional prism growth and flow model indicate pore pressures close to lithostatic

pressures at the decollement when decollement was given the same permeability as the

surrounding sediments. However, these pore pressures were unable to reach lithostatic

pressures thus not allowing horizontal hydrofracture in the decollement zone. Addition

of vertical hydrofractures in the underthrust sediments did not increase pore pressures to

lithostatic pressures in the decollement. When a bulk permeability-vertical stress

relationship was assigned to the decollement, pore pressures reach values close to

lithostatic pressures, suggesting that high pore pressures can be sustained at the base of

the prism while fluid is expelled at the toe of the complex.














CHAPTER 1
GENERAL INTRODUCTION

Examining the fluid flow of the deep hydrosphere is extremely important because

fluid flow alters the physical and chemical properties of the Earth's crust, which in turn

affects the ocean and the atmospheric chemistry that is vital for human existence

(COMPLEX, 1999). At active plate margins, fluid flow can influence movement along

faults and thus the nature of the earthquake cycle. Several research projects have recently

been focused on studying fluid flow along active plate margins. The Ocean Drilling

Program (ODP) has contributed valuable information on fluid behavior by sampling the

sediments and oceanic crust at shallow ends of the subduction zones.

At convergent margins, the incoming sediments and lithosphere are fed into the

subduction factory where processes such as compaction and dewatering, diagenesis,

dehydration, metamorphism, melting, melt migration and mantle convection result in

hazardous seismicity, explosive volcanism as well as the formation of ore deposits and

new continental crust (Moore, 1998). A large number of the world's greatest earthquakes

are associated with subduction zones. A small portion of the plate contact, known as the

seismogenic zone, is responsible for generating these large earthquakes. Understanding

the processes of the seismogenic zone provides valuable information on earthquake

generation, but requires studying many aspects of geology.

Shallowly dipping subduction zones provide a large fault surface that is accessible

to study by allowing sampling of the incoming sediments. In such localities, accretionary

complexes are formed if sediments are scraped off the subducting oceanic plate and









accreted on the upper plate. Many accretionary complexes grow above sea level and

even until they form mountain belts. One such example of a partially exposed

accretionary complex is the Barbados Island located in the Caribbean. Examples of

ancient complexes include the Shimanto Belt of southwest Japan, the Franciscan complex

of California, and the Kodiak accretionary complex of Alaska.

Even though more than half of the world's convergent plate boundaries are forming

accretionary prisms (von Huene and Scholl, 1991) there are margins where all of the

sediments riding on the oceanic plate are underthrust beneath the upper plate of the

subduction zone (e.g., Costa Rica). Particularly in the western Pacific, trenches lack

accretionary sediments because the terrigenous sediment supply reaching the trench floor

is insufficient to accrete (von Huene and Scholl, 1991). Thus, typically non-accreting

margins are bordered by sediment-starved trenches such as the Mariana and Tonga.

Deformation Processes of Subduction Zones

At convergent boundaries, sediments can be either offscraped as a series of thrust

sheets at the frontal edge of the accretionary prism or underthrust with the subduction

plate to great depths (Moore, 1989). Accreted sediments form a series of imbricate thrust

sheets that extend from the surface to the basal detachment fault or the decollement.

Accretionary prisms grow volumetrically by two main processes (Figure 1-1). On the

surface, accretionary prisms grow by frontal accretion while on the subsurface they grow

by underplating (von Huene and Scholl, 1991). According to von Huene and Scholl

(1991) the division of these two processes refers to the seaward position of the margin's

resistive rock structure (backstop), which also acts as the mechanical backstop of the

seaward part of the margin. Frontal accretion takes place in front of the backstop by

offscraping the upper part of the oceanic sediment while the lower parts of the oceanic










sediment are underthrust. Sediment subducted beneath the backstop is subsequently

accreted by underplating or transported to greater depths (von Huene and Scholl, 1991).

During frontal accretion, thrust slices will detach the upper part of the incoming

sediments. When sediments move from the oceanic plate to the accretionary prism, the

state of stress changes from gravitational to that of a thrust belt (Moore, 1989). On the

oceanic plate the maximum principal stress is oriented vertically and as sediments get

accreted the maximum principal stress gradually inclines (Davis et al., 1983). With the

stacking and upward rotation of these thrust slices, the accreted material will thicken and

shorten (von Huene and Scholl, 1991).


<-Accretionary prism->

Active Margin sediment
accretion
Trench wedge a to ..-

_> A A<--- Accreled by underplating
I '- Decollement
Underthrust sediment Subducted sediment

Figure 1-1. Diagram of accretionary prism and the processes of frontal accretion and
underplating (modified from von Huene and Scholl, 1991).

Role of the Decollement Zone

The decollement zone is the principal boundary that separates the upper and lower

converging plates (von Huene and Scholl, 1991). The sediments above the decollement

are highly deformed while the sediments below remain coherent (Moore et al., 1982).

Change in structural style across the decollement suggests that this zone marks a major

shift in the orientation of the stresses (Moore, 1989). This has been supported by the

presence of extensional veins observed in the underplated sediments that are interpreted

as hydrofractures with near vertical orientation of the maximum principal stress (Fisher

and Byrne, 1987). An important question that has not yet been answered is how the









decollement initiates. It has been proposed by many studies that elevated fluid pressures

are necessary for the initiation and sliding of the decollement (Davis et al., 1983;

Westbrook and Smith, 1983). If the decollement steps down within a packet of

subducted sediments, then the material of the lower plate can be transferred to the upper

plate, effecting underplating and volumetric growth of the prism (von Huene and Scholl,

1991). Furthermore, if the decollement moves upward, accreted material will be

transferred to the lower plate promoting subduction erosion (Charlton, 1988).

At some accretionary prisms such as Cascadia, southern Chile and eastern Alaska,

the decollement lies at the base of the incoming sedimentary section suggesting that all

incoming sediments are frontally accreted (von Huene and Scholl, 1991). Furthermore,

Davis and Hyndman (1989) inferred that large accretionary prisms such as Barbados or

the Makran prism of southern Pakistan have achieved their exceptional size due to

efficient offscraping favored by slow convergence and thick incoming sediments,

although the decollement is located well above the igneous basement.

Critical Taper Theory

The critical taper theory has been widely used to explain the shape of accretionary

prisms as well as to estimate excess pore pressures. Davis et al. (1983) used Coulomb

failure theory to demonstrate that homogeneous wedges reach a stable critical taper that

remains constant as long as the controlling parameters do not change. Once it reaches the

maximum thickening and shortening, the accretionary prism will maintain its taper by

adding material either by underplating or by new thrust faults that cut the accretionary

prism at shallow angles, which are known as the out of sequence thrusts (von Huene and

Scholl, 1991). The critical taper is defined by a +3, where a and 3 are the topographic









slope and the decollement dip respectively (Figure 1-2). The parameters that control the

taper angle are the internal (4) and basal friction angles (4b) and the pore fluid pressure

ratios of the wedge (k) and the base (kb). The k is given by

Pf Pseo oo
htho seafloor

where Pfis the pore pressure in the sediments, P is that pressure in the water column

above the seafloor, and Phtho is the total pressure of the overlying water column and

sediments. The internal ([t) and basal friction (tlb) coefficients are given by

[t=tan) and lbb=tan4b

According to critical wedge theory (Davis et al., 1983; Dahlen, 1984) the wedge

taper gives an indication of either the material properties of the wedge or the friction at

the base of the wedge. A large critical angle indicates either a weak material, which

needs to deform before stable sliding could occur, or high basal friction (Davis et al.,

1983). In contrast a small critical angle indicates either a strong material, which need not

deform for stable sliding to occur, or very little basal friction (Davis et al., 1983).





-"o \ H o riz o n ta l






Figure 1-2. Schematic of an ideal coulomb wedge (modified from Hatcher, 1995).

Using the concepts presented by Davis et al. (1983), Bernstein-Taylor et al. (1992)

interpreted that a large change in basal friction could result from a change in fluid









pressure from an overpressured decollement zone to hydrostatic fluid pressure beneath

the toe. If the material is different from the toe to arcward, then the material at the toe

can be relatively weak and can deform until a large critical taper is achieved. Arcward of

the toe, stronger material need not deform internally for stable sliding and thus, will have

a small critical angle (Davis et al., 1983).

Role of Fluid Flow at Convergent Margins

Approximately 40% of the sediment section entering the world's subduction zones

is composed of water in pore spaces (Moore et al., 2001). Fluid flow at accretionary

complexes is due to a number of different driving forces, including gravitational loading,

tectonic compression, fluid density gradients, and dehydration reactions. During both

gravitational loading and tectonic compression, sediments will generally compact.

However, if the rate of loading or compression is sufficiently high, then fluids cannot

escape fast enough, which causes pore pressures to rise. These localized excess pore

pressures generate a pressure gradient and hence fluid flow. Density variations can result

from differences in solute concentrations or due to the introduction of heat sources.

Mineral dehydration is another method where fluids are released during the reaction,

hence increasing the fluid pore pressures. A common example of a dehydration reaction

found in accretionary complexes is the smectite to illite transition.

Evidence for Fluid Flow and Pressures

At many active accretionary prisms (e.g., Japan, Barbados, Oregon-Washington,

Marianas) evidence for fluid expulsion has been observed, including the presence of

biological communities, heat flow and geochemical anomalies, and mud volcanoes and

diapirs (Peacock, 1990). Each of the evidence types is summarized in the following

paragraphs.









One of the most interesting indicators of fluid expulsion from accretionary prisms

is the presence of biological communities (Le Pichon et al., 1987). Studies have shown

that the range of biomass found in colonies is related to the fluid chemistry and flow rate

(Sibuet et al., 1990). Thus, studying the nature and distribution of the colonies provide

valuable information on qualitative fluid flow. Such examples of ancient seep

communities have been identified in accretionary prisms. These biological communities

were formed as concentrations of macrofossils in deepwater rocks (Moore and Vrolijk,

1992). If flow continues at seeps for a longer time the biomass can grow beyond the

limits of the calcite compensation depth (depth below which calcite is dissolved in the

deep-sea) developing into reef like structures that eventually could transform into

hydrocarbon reservoirs (Hovland, 1990).

Fluid transport has been inferred from fluid inclusions from veins that show

anomalously high temperatures at shallow depths in ancient accretionary complexes

(Vrolijk et al., 1988). At modem accretionary complexes, the age of the oceanic crust

can be used to predict conductive heat flux and therefore, anomalies from the conductive

heat flux that might be due to fluid flow can be identified. The best evidence for deeply

derived warm fluids comes from the Northern Barbados accretionary prism (Davis and

Hussong, 1984). It is interpreted that the observed heat flow anomalies were caused by

advection of heat during channelized fluid flow along faults (Moore and Vrolijk, 1992).

At the Barbados accretionary complex, both borehole temperatures and marine heat flow

measurements demonstrate thermal gradients approximately twice that expected for the

age of the subducting crust (Fisher and Hounslow, 1990). Heat flow values that are

lower than predicted by conductive cooling of the oceanic crust have been reported









offshore of Peru (Langseth and Silver, 1996). One possible explanation for this low heat

flow is the rapid sedimentation at Peru, which may prevent equilibrium (Moore and

Silver, 2004). Similarly, heat flow measurements at Cocos Plate off the Nicoya

Peninsula have revealed significantly low heat flow values (Langseth and Silver, 1996).

It has been suggested that hydrothermal cooling of the oceanic basement occurs from the

flow of seawater into the upper crust of the subducting plate (Langseth and Silver, 1996).

Furthermore, evidence for possible fluid migration comes from observed low

chloride anomalies (concentrations less than seawater), found in many modern prism

sediments (Kastner et al., 1991). One of the widely used explanations for the observed

low chloride anomalies is the dehydration reaction of smectite to illite. Smectite is a

common and abundant type of clay found in subduction zones (Moore and Vrolijk,

1992). The dehydration of smectite to illite is a kinetic reaction that depends on

temperature and time (Elliott et al., 1991). The reaction takes place in temperatures

above 60C and the amount of water released during the reaction is estimated to be 20%

by weight (Bekins et al., 1995). Smectite is replaced by fluid plus illite, which has

greater volume than smectite, thus resulting in an increase in pore pressures (Bekins et

al., 1995). At Barbados, low-chloride anomalies were observed along the decollement

(Kastner et al., 1993) and it is inferred that pore fluids generated from smectite

dehydration migrate toward the toe of the prism mainly through faults or fractures

lowering the chloride concentration at the toe of the prism (Bekins et al., 1995). A broad

low chloride anomaly was also observed above and below the decollement at the Nankai

accretionary complex. Because the smectite contents at the Nankai sites are low, it has

been inferred that freshening of pore fluids may be related to the in situ dehydration









caused by high temperature regimes (Brown et al., 2001). The high temperature regimes

are believed to be related to the fossil spreading ridge (Kinan Seamount) that ceased

spreading 15 Ma ago located on the Philippine Sea Plate (Shipboard Scientific Party,

2001).

Mud volcanoes and diapirs are also common features found in accretionary

complexes. They form due to increased pressure at depth and transport overpressured

mud to the surface. In areas where serpentine diapirs or volcanoes are present, it requires

fluids to hydrate basalt in order to form serpentine and these structures become conduits

of fluid flow from depth (Fryer et al., 1990). During diapirism and mud volcanism rock

sequences are disrupted and melanges are formed (Brown and Westbrook, 1988). Mud

diapirs and mud volcanoes are often observed along thrust faults, suggesting thrusting as

a mechanism that triggers mud diapirism (Behrmann, 1992).

All the evidence for fluid flow from accretionary complexes implies fluid pressure

gradients in excess of hydrostatic because pressures gradients drive fluid flow. Indirect

evidence for elevated fluid pressures includes the presence of extensional veins (Moore

and Vrolijk, 1992). Crack-seal textures in veins indicate repeated pulses of high fluid

pressure (Vrolijk, 1987; Fisher and Byrne, 1990). Repeated events of vein growth

suggest that fluid pressure evolve throughout the growth of the accretionary complex

(Labaume et al., 1997). Fluid pressure in accretionary prisms can vary from hydrostatic

(equivalent to the weight of the overlying column of water) to nearly lithostatic

(equivalent to the weight of the overlying column of sediments). Variation in fluid

pressures from hydrostatic to lithostatic has been observed in wells at the eastern

Aleutian Trench where the fluid pressures are at hydrostatic at 2-3 km from the surface









and increasing to more than 80% of lithostatic pressures at total depth (Moore and

Vrolijk, 1992).

Basics of Fluid Flow Related to Subduction Zone

Although sediments found in accretionary prisms are highly deformed during

subduction, it is assumed that the high density and interconnectedness of the fractures in

accretionary prisms approximate Darcian flow (Moore and Vrolijk, 1992). The fluid

produced during the accretionary process can be evaluated using the two principles that

govern fluid flow in the subsurface. They are the principle of conservation of mass and

Darcy's law. The principle of conservation of mass states that for an arbitrary control

volume, the rate of mass accumulation within the volume plus the net mass flux out of

the volume must equal the rate of mass generation within the volume (Bird et. al., 1960).

If we consider a very small volume of the aquifer known as a control volume, we can

approximate the flow through the matrix using Darcy's law. The most basic form of

Darcy's law states

Q/A=-K dh/dl

where Q/A is flow per area or linear velocity [LT-1], K is hydraulic conductivity [LT1],

and dh/dl is hydraulic gradient. The hydraulic conductivity (K), which is a

proportionality constant, represents both properties of the fluid and the porous media. It

is given by

K= kpg/[t

where k is intrinsic permeability [L2], which is representative of the properties of the

porous medium, p is the fluid density [ML-3], [[ is the fluid viscosity [L-1 T-1] and g is the

gravitational constant [L T-2]. Intrinsic permeability depends on variables such as grain









size, sorting, and roundness of the sediment through which fluid flowing. Hereafter,

intrinsic permeability will be referred to as permeability.

Permeability

Sediment permeability is the most important factor that controls pore pressures as

permeability can vary by many orders of magnitude (e.g., Bruckmann et al., 1997; Saffer

and Bekins, 1998). Therefore, the use of a systematic relationship between porosity and

permeability is valuable for approximating the permeability structure in an accretionary

wedge fluid flow model (Saffer and Bekins, 1998). Processes in subduction zones such

as loading, compaction and cementation of sediments lead to reductions in permeability.

When permeability is reduced and reaches some critical value, rocks can no longer

transmit significant amounts of fluid, but new permeability can be created in the form of

a fracture or a fault. If fractures are formed then they can significantly affect the

permeability of the accretionary complex and, thus, large-scale field measurements of

permeability would be more appropriate than core scale permeability measurements to

determine large-scale fluid flow (Moore and Vrolijk, 1992). Unfortunately only a few

large-scale field measurements of permeability have been made in accretionary settings.

These include shipboard packer tests and submersible-based tests conducted at a sealed

borehole at the Oregon accretionary complex (Screaton et al., 1995) and shipboard packer

tests (Fisher and Zwart, 1996) and submersible slug tests and discharge tests (Screaton et

al., 1997) in the decollement of the Barbados accretionary complex. Because of

difficulties in conducting large-scale field measurements, permeability measurements are

primarily from core samples that are retrieved from the frontal part or shallow depths of

the accretionary complex. Even though these core sample measurements do not represent

the large-scale variations in permeability due to faults, they provide valuable estimates









for the matrix permeability that is critical in approximating permeability structures in the

accretionary complex for modeling studies.

Hydrogeologic Modeling

Mathematical models used in hydrologic modeling are derived from the governing

principles of fluid flow and specifications such as formation geometry, boundary

conditions and initial conditions. These models help quantify conceptual models of sub-

seafloor hydrogeologic flow system. These models can be extremely useful and cost

effective in providing possible explanations for known or hypothesized conditions. They

can also be used to assess whether or not a conceptual model is feasible. As a starting

point, with limited data it is best to use one or two-dimensional analytical solutions

derived from simple well-defined boundary problems (Anderson and Woessner, 1992).

However, in sub-seafloor settings, numerical models are often necessary in order to

account for parameters such as complex geometry, variable density fluid flow, and

variations in heat flow.

Due to limited access to convergent margins, models are essential for integrating

the field observations with laboratory results. It has also been recognized that numerical

models are required in order to extend observations made at shallow parts of the

subduction system to greater depths such to the seismogenic zone (COMPLEX, 1999).

Most previous modeling studies have focused on coupled compaction-fluid flow and

diffusion-advection models of pore fluid chemistry and heat for Barbados, Nankai and

Cascadia accretionary prisms (e.g., Bekins et al., 1995; Saffer and Bekins, 2002; Screaton

and Ge, 1997). However, recent data collection allows significantly improved

characterization of permeability, which is a major component that affects modeled fluid

pressures in accretionary complexes. Furthermore, previous modeling studies at Nankai









have largely focused on estimating pore pressures with little focus on examining the

causes of excess pore pressures (Le Pichon and Henry, 1992; Screaton et al., 2002;

Saffer, 2003). At Barbados, modeling studies were focused on pore pressure generation

either at the toe of prism (Henry and Wang, 1991; Shi and Wang, 1994; Stauffer and

Bekins, 2001) or in a steady-state approach instead of examining pore pressure

generation through the subduction process.

Statement of Problem

The objective of this investigation was to expand the knowledge of fluid flow and

the development of pore pressures based on both laboratory measured permeability

values and numerical models at selected accretionary complexes. This study benefits our

current understanding of fluid flow in accretionary complexes in several ways. One of

the primary benefits of this research is the contribution and synthesis of permeability

measurement of subduction zone sediments, which provides new insight to flow

simulations in convergent margins. This study also provides valuable information on

fluid flow paths, areas of excess pore pressures, degree of importance of lithology and

sediment thickness in fluid flow, the initiation of decollement, and factors that contribute

to the initiation of the decollement. These results further provide valuable information

for future drilling projects such as the Seismogenic Zone Experiment (SEIZE) that is

focused on understanding the relationship between earthquakes, deformation, and fluid

flow. The following were the specific objectives of this study:

* To synthesize permeability data and predict permeability-porosity relationships at
four major convergent margins. The four locations are the Northern Barbados,
Costa Rica, Nankai and Peru subduction zones, which represent a variety of marine
sediments.

* To investigate the effects and magnitudes of parameters such as bulk permeability,
lateral stress, and the presence of a low-permeability barrier at the decollement on









the generation of excess pore pressures at the toe of the Nankai accretionary
complex.

* To investigate the evolution of pore pressures and implications for episodic fluid
flow at the Barbados accretionary complex using a two-dimensional growth and
flow model.

The following chapters include detailed methodology and discussions of the results

of the proposed research. Chapter 2 is titled "A comparative study of permeability

measurements from the subduction zones of northern Barbados, Costa Rica, Nankai and

Peru" and a modified version of this chapter will be submitted for publication to Marine

Geology. This chapter is a contribution to the permeability data of marine sediments at

subduction zones. The major results of this chapter include relationships among

permeability and porosity for different types of marine sediments. The constraints

provided by these relationships will allow realistic estimation of fluid flow and pore

pressures in marine settings. Furthermore, the laboratory measured permeability data

from Nankai and Peru contributes to the general knowledge of marine sediments. The

permeability data of this work has been published as two data reports (Gamage and

Screaton, 2003; Gamage et al., 2005). A modified version of Chapter 3, titled

"Characterization of excess pore pressures at the toe of the Nankai accretionary complex,

Ocean Drilling Program sites 1173, 1174, and 808: Results of one-dimensional

modeling" has been accepted for publication by the Journal of Geophysical Research,

authored by Gamage and Screaton. This chapter contributes to the understanding of the

development of pore pressure at the toe of the Nankai accretionary complex. This study

was based a one-dimensional model and uses the permeability-porosity relationship

developed for Nankai hemipelagic sediments in the previous study. The sensitivity of

pore pressures to bulk-permeability, lateral stresses within the prism, and a low-









permeability barrier at the decollement was also tested. The results of this simplified

model assess parameters that significantly affect pore pressures in subduction zones.

Furthermore, the results of this study provide insight on the initiation of the decollement

zone. Chapter 4 contributes to the understanding of the development of pore pressures

through space and time at the Barbados accretionary complex. This study is based on a

two-dimensional model that allows tracing the development of pore pressures as

sediments subduct beneath the prism. The model allows examination of the effect of

hydrofracture and a decollement with varying permeability based on a relationship of

bulk permeability -vertical effective stress. Results indicate the spatial and temporal

variations of excess pore pressures and provide insight to possible mechanics for episodic

fluid flow. This chapter will be adapted for submittal to the Earth Planetary Science

Letters. Chapter 5 summarizes the principal findings of chapters 2 through 4.














CHAPTER 2
A COMPARATIVE STUDY OF PERMEABILITY MEASUREMENTS FROM THE
SUBDUCTION ZONES OF NORTHERN BARBADOS, COSTA RICA, NANKAI,
AND PERU

Introduction

Marine sediments have been widely studied for their physical properties both in

academic and industrial research. With the introduction of the Deep Sea Drilling Project

(DSDP) and the Ocean Drilling Program (ODP), a new level of understanding has been

added to the knowledge of marine sediments during the past few decades. Physical

properties of submarine sediments have been studied largely through recovered cores,

down-hole logging, and also by in situ instrumentation. Permeability is one such

physical property that has been closely studied for its importance in fluid flow and pore

pressures in the subsurface. Previous studies based on permeability measurements of

marine sediments have suggested that correlation between permeability and porosity

could provide insight to a large range of sediments in nature (Bryant, 2002).

Investigations based on numerical modeling have shown that permeability is a

crucial parameter in accretionary complex hydrology (e.g., Bekins et al., 1995;

Bruckmann et al., 1997; Saffer and Bekins, 1998). According to Saffer and Bekins

(1998) sediment permeability is the most important factor that controls modeled pore

pressures because it can vary by several orders of magnitude. Thus, using a systematic

relationship between porosity and permeability is a powerful way to approximate the

permeability structure in an accretionary wedge model (Saffer and Bekins, 1998).

Results from modeling studies have shown that pore pressures are highly sensitive to the









permeability-porosity relationship (e.g., Gamage and Screaton, 2006). Prior to the

availability of core samples of marine sediments, many studies extrapolated

permeabilities from fine-grained terrigenous sediments found on-shore and in many cases

these values produced ambiguous results (Bryant et al., 1981). With the availability of

more samples, the quantity of permeability data has significantly increased. However,

difficulties in laboratory measurements and finding undisturbed cores have limited the

amount of permeability measurements representative of different lithologies and

structural domains at subduction zones.

The focus of this study is to synthesize available permeability data from four

different subduction zones with the aim of predicting permeability-porosity relationships

for a number of sediment types found in modern accretionary complexes and to examine

what parameters affect the relationship between permeability and porosity. The samples

used in this study represent sediments from Northern Barbados, Costa Rica, Nankai, and

Peru. Samples representative of the Northern Barbados and Nankai subduction zones

mainly consist of fine-grained clays and silts that are commonly grouped as

hemipelagics. Samples from Costa Rica consist of both hemipelagics and calcareous

oozes while samples from Peru consisted of calcareous oozes and siliceous oozes. Using

existing permeability data, permeability-porosity relationships were developed based on

depositional environment, grain size distribution and structural domain. These

relationships were compared and examined to evaluate the relative importance of each

variable.











Background

Barbados

The Barbados accretionary complex is located in the Caribbean where the north


American Plate (Figure 2-1) is being subducted beneath the Caribbean Plate at a rate of


about 2 cm/yr in an east-west direction (DeMets et al., 1990). Active accretion of


sediments at the Barbados accretionary complex takes place at the eastward end of the


complex. The complex is partially exposed above sea level at Barbados Island (Figure 2-


1). At the location of DSDP and ODP drilling, the incoming sediments are


predominantly clay and claystones.


A w **









SDSDP'ODP dril T.1e.


\\ \ ^ area
,o









Depth
A45. .45
1 Site (in She 1047 Ste k10
--..---. I I I 1 3. .- -
4S-- -" -45
SI5 1O4* S*n $401041 (tss4





Bend in seclon
1 5 km
Figure 2-1. A) Map of large-scale regional setting location of the drill sites at the
Barbados accretionary complex. B) Cross section from the seismic depth









section extending from west of Site 949 (ODP Leg 156) to Site 672 (ODP Leg
110), (Shipboard Scientific Party, 1998).

This study used core permeability measurements from ODP Leg 156 Sites 948 and

949 and ODP Leg 110 Sites 671, 672 and 676. Site 948 is located 4.5 km west of the

deformation front and coincides with the location of Site 671 while Site 949 is located 2

km northeast of Site 948 (Shipley et al., 1997). Site 676 was drilled 0.25 km arcward of

the deformation front and Site 672 was drilled 6 km east of the deformation front to

provide an undeformed reference site (Mascle et al., 1988).

Costa Rica

The Middle American Trench (MAT) is formed by the eastward subduction of the

Cocos Plate beneath the Caribbean Plate (Figure 2-2) at a rate of about 8.8 cm/yr (Silver

et al., 2000). At Costa Rica the incoming sedimentary sequence is about 380 m thick and

consists of approximately 160 m of siliceous hemipelagic sediments overlying about 220

m of pelagic carbonates (Silver et al., 2000). As indicated by drilling on DSDP Legs 67

and 84, this stratigraphy is regionally continuous between the Leg 170 transect and

offshore Guatemala (Aubouin and von Huene, 1985; Coulboum, 1982). During ODP

Leg 170, two locations penetrated the decollement zone. Site 1043 is located 0.5 km

landward of the trench and Site 1040 is located 1.6 km seaward of the trench. The

incoming sediments at Site 1039, which is located at 1.5 km seaward of the deformation

front, were also drilled during ODP Leg 170. In a more recent visit to the MAT, ODP

Leg 205 drilled Sites 1253, 1254 and 1255 (Figure 2-2). Site 1253 is located 0.2 km

seaward of the deformation front while Sites 1254 and 1255 are located coincident with

Sites 1040 and 1043, respectively.












A


SNicoya Peninsula

SCosta Rica
10 N





Legs 17
LagsODP 7 ..... '
Mexico
0 Carib-
9 30 N 0I
ean
tat
Study Area
Cocos Plate

0 0
86W 85 30' W

B
4.5


5.0
hemipelagic 1040

z125v w





0_' ..ocean crust -' ...:
7.0

Figure 2-2. A) Map showing location of ODP drilling along the Costa Rica subduction
zone. B) Cross-section indicating ODP Leg 170 and 205 drilling sites used
for this study (Silver et al., 2000).


Nankai

The Nankai accretionary complex is formed by the subduction of the Shikoku


Basin on the Philippine Sea Plate beneath the southwest Japan arc on the Eurasian plate


(Figure 2-3) at a rate of about 4 cm/yr (Seno et al., 1993).








21



A
34'




yp .. -. _
33' r-- 1 4





1173 Muroto
) Transect



\ Transect

132'E 133 134 135' 13'


B
NW SE
X Line 900 800 700 600 500 400 300 200
ProtoihNrs
zone
S ite 808 Deformation Site 1173
i &-i^S^- ie i 1174 ron
Fron Trencl,-Wedge Facie5
E 4- i 4 -ppe' Sri kclu Basn Fa-Ies
.I' Lower Shikoku Basin Facies
Seco t.emeni zone Volcaniclastic Facies
Ocean Crus r I



C
NW SE
3 Pitkhorust
3 P u NT-2 Ashizuri Transct
Zone
Apit. S | 1177583




a,_'Z


5kmrn

Figure 2-3. A) Location map of ODP Leg 190 (solid circles) and previous ODP/DSDP
drill sites (solid squares) in the Nankai Trough. B) Seismic reflection profile
through the Muroto Transect reference (Site 1173) and prism toe sites (Site
1174 and 808). C) Seismic reflection profile through Ashizuri Transect
showing the reference Site 1177 (Shipboard Scientific Party, 2001).


During ODP Leg 190, Sites 1173 and 1174 were drilled along the Muroto Transect


while Site 1177 was drilled approximately 100 km west of Muroto along the Ashizuri


Transect (Figure 2-3). Site 1173 was drilled 11 km seaward of the deformation front and









provides an undeformed reference site of the incoming sedimentary sequence (Shipboard

Scientific Party, 2001). Site 1174 is located about 1.8 km landward of the deformation

front and penetrates the decollement within the proto-thrust zone (Figure 2-3). Site 1177

was drilled approximately 18 km seaward of the deformation front as the reference site

for the Ashizuri Transect (Shipboard Scientific Party, 2001). At both Sites 1173 and

1174, the hemipelagic sediments of the upper and lower Shikoku Basin are overlain by

the turbidite-rich trench-wedge facies, which was not tested for permeability. At Site

1177, the trench-wedge facies was not cored.

Peru

The Peru accretionary complex is formed by the northeastward subduction at

approximately 6.1 cm/yr of the Nazca plate (Hampel, 2002) below the Andean

continental margin along the Peru Trench (Figure 2-4). During ODP Leg 201, seven sites

were drilled into a wide range of subsurface environments in both open-ocean (Sites

1225, 1226 and 1231) and ocean-margin provinces (Sites 1227 and 1230) of the eastern

tropical Pacific Ocean. These subsurface environments include carbonates and siliceous

oozes typical of the equatorial Pacific, clays and nannofossil-rich oozes of the Peru

Basin, biogenic and terrigenous-rich sediments of the shallow Peru shelf, and clay-rich

deepwater sequences of the Peru slope (Shipboard Scientific Party, 2003).









A









1i3 ,e tang. a--~ne L
,"0 500 000












map of equatorial Pacific primary sites. ODP site designations are in
S P 2003).)



Laboratory PAmermeability Data
5,. ,* ,, ,










FigLaborure 2-4. A) Map showing genermeablity dlocations of drill sites occupia were provided by several souringces for eachgs
138 (rectangle B) and 112 (rectangle C) at Peruof the four subduction zones. Two widely used methods for permeability measurements
mare through direct flow tests (e.g., falling or constant head, constant flow) are in


pconsolidation tests are one order of magnitude less thargin from diret measu. Previoments using
DSDP/direct flow methods. Becausite Brydesignationst et arel. (1981) obin parentheses (Shipboard Scientific Party,
2003).

Laboratory Permeability Data122







Figure 2-4. consolidation tests. Bryant et al. (1981) cited that results of calculatpied dupermeability framing ODP Legs


138direct flow methods. Because Bryant et al. (1981) atobs Peru subducvation was based one. B) Locationan











individual direct flow method, it cannot be confirmed that all direct flow methods are

incompatible with consolidation tests. However, to be consistent, I limited the


permeability data only to those obtained from direct flow methods and excluded data









obtained from consolidation tests. Permeability data that were reported without porosity

or void ratio information were also excluded from this study, which examines

permeability as a function of porosity. Methods of permeability measurements used in

this study are briefly discussed in the following section. Where only hydraulic

conductivity was provided, permeability was calculated using (Fetter, 1994)

k = K /pg (1)

where k is intrinsic permeability [L2], K is hydraulic conductivity [L T-1], p is fluid

density [M L-3], g is the gravitational constant [L T2] and [t is the kinematic viscosity [M

L-1 T1]. Values of fluid density and viscosity were determined based on temperature

values reported for the experiment and the salinity of the permeant used. In cases where

temperature and/or permeant used were not reported, I assumed a temperature of 250 C

and a salinity (for permeant) of 35 kg/m3.

Barbados

Vrolijk et al. (unpubl. data, cited in Zwart et al., 1997) used a sample from ODP

Leg 156 Site 948 to measure permeability using a constant-head permeameter at an

effective stress of 241 kPa. A sample with 2.5 cm in diameter and about 5 cm high was

contained in a triaxial cell. The sample was backpressured at 350 kPa to dissolve any

trapped air in the system. The permeant was saline whereas the confining fluid was oil.

Bruckmann et al. (1997) measured permeability using three whole round core samples of

size 6.2 cm in diameter and about 2 cm high from Leg 156, Site 949. Permeability was

measured at individual load steps using low gradient flow tests as described in Olsen et

al. (1985). Fresh water was used as both the permeant and the confining fluid. Maltman

and others (presented in Zwart et al., 1997) measured permeability of a cylindrical









subsample from Leg 156, Site 949 using low gradient (- 25 kPa), constant-rate flow tests.

The sample was 3.8 cm in diameter and 7.6 cm in height. Two sets of permeability

measurements were obtained. During the first set, the effective stress was varied by

maintaining a constant confining pressure while varying the pore pressures within the

sample. In the second set, effective stress was varied by maintaining a constant pore

pressure and varying the confining pressure.

Taylor and Leonard (1990) used the falling head method to measure permeabilities

of samples from Leg 110. Samples were backpressured to ensure complete saturation

according to Lowe et al. (1964). Permeabilities were obtained at least 24 hr after the

application of each new load, yielding a distribution of permeabilities at incremental void

ratios. Although both vertical and horizontal permeabilities were measured, I only used

the vertical permeability values in this study for consistency with the other samples.

Permeabilities were estimated based on hydraulic conductivities reported in Taylor and

Leonard (1990).

Nankai and Peru

Gamage and Screaton (2003), Gamage et al., (2005), and Hays (unpublished data)

used the constant flow method, which induces a hydraulic gradient across the sample

where the measurements of the pressure difference allow determination of permeability.

The samples measured by this method were from ODP Leg 190 Sites 1173, 1174 and

1177 at the Nankai margin and ODP Leg 201 Sites 1225, 1226, 1227, 1230 and 1231 at

the Peru margin. Samples had a minimum diameter of 5.84 cm and a height that ranged

from ~ 5.84 tolO cm. Samples were backpressured prior to flow tests. Several

consolidation steps were run with the confining fluid pressure used for the consolidation.

Permeability was measured at the end of each consolidation step. An idealized solution









of seawater was used as the permeant while deionized water was used as the confining

fluid in the cell. Corresponding porosities for estimated permeability were calculated

using the change in volume of fluid contained in the cell after each consolidation step.

Masters and Christian (1990) used constant head tests on two Peru samples from

ODP Leg 112 Sites 679 and 681. Whole-round samples of 10 cm in height were

backpressured to ensure complete saturation. Constant head tests were performed at

different hydraulic gradients at varying stress levels. All flow tests were performed in the

upward vertical direction. De-aired, filtered seawater with a salinity of 35 kg/m3 was

used as the permeant.

Permeability measurements of Nankai sediments conducted by Taylor and Fisher

(1993), Byrne et al. (1993), Bourlange et al. (2004) and Adatia and Maltman (2004) were

not used in this study due to insufficient data presented and/or inconsistencies found in

the methodology. For example, Byrne et al. (1993), Bourlange et al. (2004) and Adatia

and Maltman (2004) failed to report porosity/void ratios. Although Bourlange et al.

(2004) include amounts of void ratio decrease at certain confining pressures the data did

not include an initial porosity to estimate porosities at each effective stress. Taylor and

Fisher (1993) used air in permeability testing and also failed to backpressure their

samples prior to flow tests. According to Saffer and Bekins (1998) samples that are not

reconsolidated could overestimate permeabilities due to fabric expansion, and the use of

air in permeability testing would further overestimate the permeability measurements.

Costa Rica

Saffer et al., (2000) performed constant flow tests on samples measuring 6.25 cm in

diameter and 1.5 to 1.6 cm in length from ODP Leg 170, Sitesl039 and 1040.

Permeabilities were measured at several stages during the sample consolidation to









acquire permeability values at varying void ratios. Multiple flow tests were conducted at

each void ratio. McKieman and Saffer (2005) performed flow through permeability tests

on samples that were 2cm tall and 5cm in diameter of ODP Leg Sites 1253, 1254 and

1255. During flow tests, fresh water was pumped into the top of the sample at a constant

rate while pressure was maintained at the cell base. The pressure difference was

determined by monitoring the pressures at the top of the cell during each flow rate.

Varying flow rates were used to produce varying pressure difference across the sample.

Distilled, de-aired water was used both as the permeant and confining fluid.

Screaton et al. (2005) used constant flow permeability tests and constant pressure

difference tests on samples from ODP Leg 170 Sites 140 and ODP Leg Sites 1253 and

1255. Testing conditions were the same as described in Gamage and Screaton (2003) and

Gamage et al. (2005). The only exception to this method was using a constant pressure

difference to induce flow through the sample rather than applying a constant flow rate for

several of the samples.

Permeability-Porosity Relationship

Bryant et al. (1975) and Neuzil (1994) observed that permeability of argillaceous

sediments follows a log-linear relationship with porosity. The log linear relationship is

given by

log (k)= log (ko)+bn (2)

where ko is the projected permeability at zero porosity, b is a parameter describing the

rate of change of the logarithm of permeability with porosity, and n is the porosity.

Description of Statistical Methods

The coefficient of correlation (R2) of the regression equation describes the

variability of the estimates around the mean. However, it inherits the problem of small









sample size. Thus, in such situations the derived statistics are not necessarily the best

indicator of "goodness of fit". Examination of residuals helps to reaffirm the "goodness

of fit" of the regression equation in conjunction with the R2 value. This examination

involved plotting the residuals vs. the dependent variable. If the residuals exhibit a

random distribution and have a more or less even split above and below the zero line then

it is possible to say that the equation describes the relationship well (Kirkup, 2002)

The other test requires formulating a null hypothesis and an alternate hypothesis,

which are then tested using ANOVA and t-statistic. This test helps to access the

suitability of the best-fit equation that describes the relationship between the variables

(Kirkup, 2002). The hypothesis test utilize in this analysis is a one-tailed ANOVA. In

this case the null and alternate hypotheses are,

Ho: the equation has a zero slope;

Ha: the equation has non-zero slope.

Since the linear regression equation relates porosity to permeability using the slope

and the intercept of the equation, the null hypothesis is that of zero slope, which yields a

constant function. The hypothesis tests were performed at the 95% confidence interval.

Results

Permeability values were plotted on an outline of Neuzil's (1994) compiled range

of permeabilities as a function of porosity for argillaceous sediments (Figure 2-5). The

majority of permeability values were enveloped within Neuzil's (1994) plot. However,

several samples from Site 1231 of Peru and Sites 1039 and 1040 of Costa Rica plotted

outside Neuzil's (1994) plotted area. These include non-argillaceous sediments of

calcareous oozes and few samples containing siliceous oozes.



















18 +


-20 -+ Barbados
Costa Rica
x Nankai
-22 A Peru


-24 1
0 02 04 06 08 1
Porositl

Figure 2-5. Plot of laboratory derived permeability measurements from Barbados, Costa
Rica, Nankai, and Peru subduction zones superimposed on outline of Neuzil's
(1994) plot for argillaceous sediments. As compared to Neuzil's (1994) paper,
the axes have been transposed.

Effects of Depositional Environment

According to Boggs (2001) there is little agreement regarding the classification of

deep-sea sediments and thus, suggested classifications range from those that are largely

genetic to those that are largely descriptive. Unfortunately there is no single

classification that take into account both genesis and descriptive properties of all kinds of

deep-sea sediments (Boggs, 2001). Thus, here I used generalized descriptions used by

Boggs (2001) to categorize hemipelagic and pelagic sediments. Hemipelagic muds are

defined as mixtures of fine-grained terrigenous mud with biogenic remains that are

deposited under very low current velocities. According to Stow and Piper (1984),

hemipelagic muds contain more than 5% biogenic remains and a terrigenous component

of more than 40%. The terrigenous component of the hemipelagic muds are commonly

composed of fine terrigenous quartz, feldspar, micas, and clay minerals while biogenic









remains include siliceous organisms such as diatoms and calcareous organisms such as

foraminifers and nannofossils (Boggs, 2001).

Following Boggs (2001), I subcategorized the pelagic sediments into two basic

groups based on the abundant type of biogenic remains present in the sediment. Pelagic

sediments may be composed mainly of clay-size particles of terrigenous or volcanogenic

origin or it may contain significant amounts of silt to sand-size planktonic biogenic

remains (Boggs, 2001). Pelagic sediments that contain significant amounts ofbiogenic

remains are called oozes. However, little agreement exists with regards to the amount of

biogenic remains required to qualify a sediment as an ooze (Boggs, 2001). Boggs (2001)

suggested that oozes have more than two-thirds of biogenic components. Thus,

depending on which biogenic component is dominant, a sediment can be classified either

as a calcareous or a siliceous ooze. The pelagic sediments that are predominantly of

calcium carbonate tests were grouped as calcareous oozes while sediments that are

predominantly of diatom tests were grouped as siliceous oozes.

In situations where similar lithological descriptions based on depositional

environments are used, sediments from different locations were grouped together. For

example, Northern Barbados, Nankai and Costa Rica all provide samples of hemipelagic

sediments. Because even slight differences in permeability-porosity relationships could

affect results of fluid flow models (e.g., Gamage and Screaton, 2006), it is worthwhile

comparing the permeability-porosity relationships based on depositional environment

between individual locations.

Based on the depositional classification described above all samples from Northern

Barbados and Nankai were grouped as hemipelagic sediments. Samples from Costa Rica









represented both hemipelagic sediments as well as calcareous oozes. Samples from Peru

consisted of both calcareous and siliceous sediments. Unfortunately the only quantitative

data available for the biogenic component of these sediments are from smear slide

analyses. As a result it was difficult to determine a representative percentage of the

biogenic component. The nannofossil rich samples were grouped as calcareous oozes.

The most problematic was to classify the diatom-rich sediments. Lithological

descriptions and smear slide analysis favored them in the "siliceous ooze" category, and

quantitative data on the components were lacking. However, it should be noted that these

samples could contain considerable amounts of biogenic and terrigenous components and

may fall in between hemipelagic and pelagic sediments.

Peru samples from Site 1227 were rich in organic carbon and did not fit either of

the pelagic sediment categories identified in this study. Thus they were excluded from

this study. It should be noted that the samples from Sites 1225 and 1226 represents the

sediments of the equatorial Pacific where the direction of plate movement carries them

away from the trench. Thus, these two sites do not represent the sediments of the Peru

subduction zone. However, they consist dominantly of calcareous oozes, which can be

compared to the calcareous oozes of Costa Rica, and for that purpose were included in

this study. The two main lithological groups of hemipelagics and pelagics were plotted

separately in Figures 2-6.

Based on the calculated R2 value for the Barbados hemipelagics, the equation

explains only 20% of the correlation between permeability and porosity. Although the

analysis of residuals showed randomness, the hypothesis testing did not support the

existence of a statistically significant correlation between permeability and porosity for









the Barbados hemipelagics. When six of the data points, which were obtained from

Taylor and Leonard (1990), were removed from the regression, the R2 value of the

remaining 19 samples increased from 0.20 to 0.50, predicting a permeability-porosity

relationship of log (k)=-24.24 + 11.3 In. Although statistically this makes the six data

points outliers, a closer look at the lithological descriptions, grain size data and CaCO3

percentages suggest that the hemipelagics samples used by Taylor and Leonard (1990)

are likely to be different from the rest of the hemipelagics sediments representative of

Barbados because samples used by Taylor and Leonard (1990) represented calcareous

muds whereas other samples represented claystones.

Even though many of the samples from Barbados, Nankai, and Costa Rica fall in

the depositional classification of"hemipelagics", they are not well represented by the

same permeability-porosity relationship (log (k) = -19.91+4.9n, R2 = 0.5). The Barbados

permeability-porosity relationship predicts similar values of permeability to Nankai and

Costa Rica relationships at porosities between 0.55-0.70 (Table 2-1). However, at lower

porosities the Barbados permeability relationship predicts lower values than those of

Costa Rica and Nankai. Because Barbados permeabilities are constrained by a small

range of porosities, one should be cautious outside the porosity range of the laboratory

results. The log-linear relationships for Nankai and Costa Rica plot roughly parallel to

each other with slightly lower permeabilities at Costa Rica for a given porosity than at

Nankai (Figure 2-6). The R2 value obtained for Nankai was 0.79 while for Costa Rica it

was 0.70 suggesting reasonable correlation between permeability and porosity at these

two locations. The R2 value obtained for the log linear relationships for the calcareous

oozes were less than 0.5 while for the siliceous oozes it is greater than 0.5 (Figure 2-6).











The calcareous oozes of Costa Rica plot approximately in the same permeability range as


those from Peru, near the upper boundary of Neuzil's (1994) plot.


Table 2-1. Log linear permeability-porosity relationships predicted for varying
depositional environments at Barbados, Costa Rica, Nankai and Peru.
Location and Depositional environment Permeability-porosity relationship
Barbados hemipelagics log (k) = -22.02 + 8.25n (R2 0.20)
Costa Rica hemipelagics log (k) = -20.84 + 6.27n (R2= 0.70)
Nankai hemipelagics log (k) = -19.80 + 5.37n (R2= 0.79)
Costa Rica calcareous oozes log (k) = -18.09 + 4.83n (R2= 0.33)
Peru calcareous oozes log (k) = -20.87 + 7.79n (R2= 0.40)
Peru siliceous oozes log (k) = -18.64 + 3.55n (R2= 0.67)

-12 I -12 I
A B

-14 -14 -


-16 16 -




SCosta Rica calcareous oozes
Barbados hemipelagics -20 -- =-18 9 + 83rn, R2 33)
-20 -1-. log k=-22 02 + B25n, (R2 =020)
Costa Rica hemipelagics / Peru calcareous oozes
logk =-2084 + 27n, (R2 = 70) -- log k -20 87+7 79n, (R2 = 040)
x Nankai hemipelagics r Peru sliceous oozes
-22 log k=-19BO +537n, (R2= D79) 22,
-- -- log k =-18 64 + 355n, (R2 =U 67)
Allheipelaic All Calcareous oozes
g k =-1 91 +4 90n. (R2 = 50) log k =-1827 1421n, (2 =0 14)
1 1 I I I I I -24 1 1 1
-24 T 02 04 06 08 1 0 02 04 06 08
Porosity Porosity

Figure 2-6. Permeabilities classified based on depositional environment and location. A)
Predicted log-linear permeability-porosity relationships for hemipelagic
samples. B) Permeabilities and predicted log-linear permeability-porosity
relationships for pelagic samples.

Peru calcareous oozes lie at a higher porosity range (0.55-0.80) while Costa Rica


calcareous oozes represent porosities between 0.45 and 0.70. The Peru siliceous oozes


plot between the lower permeability values of calcareous oozes and the higher


permeability values of hemipelagic sediments for porosities ranging from 0.45-0.8.


Although the predicted R2 values were fairly low for the two calcareous oozes, a linear


relationship between the logarithm of permeability and porosity for those two groups was









not rejected based on the residual distribution and hypothesis testing. Relative to

relationships predicted for hemipelagic sediments, the permeability-porosity relationship

predicted for the siliceous oozes show similar permeabilities at porosities >0.7 and higher

permeability values at porosities <0.7.

Effects of Grain Size

Bryant et al. (1981) stated that grain size is the most important characteristic of a

sediment that determines permeability as it affects the porosity of the sediment. This

contention has been supported by studies such as by Koltermann and Gorelick (1995), in

which grain size distribution was used to predict permeability and porosity in various

sediment mixtures. Ninety percent of their predicted hydraulic conductivities matched

the hydraulic conductivities estimated from field tests within one order of magnitude

(Koltermann and Gorelick, 1995). Accordingly, I categorized the sediment samples in

this study based on a grain size classification described by Bryant (2002). In contrast to

Bryant's (2002) study, which exclusively used sediments from the Gulf of Mexico, I

combined samples from all four subduction zones due to limited variation in grain size

data found at any single location.

Bryant (2002) noted that fine-grained marine sediments with low amounts of

carbonate did not affect the permeability-porosity relationship, but did not specify what

percent of carbonate was considered. Based on our calcareous pelagic samples I used 45

CaCO3 wt% as the limit between the high and low carbonate content. The 45 wt% was

used because majority of the samples contained either distinctively high (>> 45 wt%

CaCO3) or low (<<45 wt% CaCO3) amounts of CaCO3. For samples that did not have

CaCO3 wt% data, I used information from other methods such as X-ray diffraction and









inorganic carbon percentages to infer whether samples were likely to contain greater than

45 wt%/ of carbonates. Although grain size data were available for samples from ODP

Leg 112 Sites 679 and 681 (Peru), the scale used for the particle size criteria (Masters and

Christian, 1990) was different from the rest of the grain size data used in this study.

Thus, I excluded grain size data from Sites 679 and 681 in this part of the study. All

samples used in the grain size classification (marked with an asterisk) and values of

available weight percentages of CaCO3 are tabulated in Appendix A.

All samples were categorized into the following four groups based on their grain

size distribution:

* Group 1, sediment containing more than 80% clay size material.

* Group 2, sediment containing 60-80% clay-size material.

* Group 3, sediment-containing silty-clays with less than 60% clay and less than 5%
sand.

* Group 4, sediment-containing sandy-silts with less than 60% clay and more than
5% sand.

The following particle size criteria were used for classification: sand (>63 tpm), silt

(63-4 tpm) and clay (<4 tpm). The log linear relationships for each of the groups obtained

by least squares regression fit are given in Table 2-2. Group 4 was excluded from the

grain size classification as it only contained a total of four samples.

Table 2-2. Permeability-porosity relationships based on grain size analyses.
Group Description Permeability-porosity Permeability-porosity relationship
relationship for Gulf of Mexico (Bryant
2002)*
1 > 80% clay log (k) = -24.28 + 11.32n log (k) = -20.9 + 6.54n
(R2 = 0.53)
2 60-80% clay log (k) = -19.73 + 4.49n (R2 log (k) = -20.53 + 6.16n
= 0.56)
3 Silty-clays with log (k) = -19.91 + 5.45n (R2 log (k) = -20.59 + 6.77n
<60% clay and = 0.43)
<5% sand










Bryant (2002) used hydraulic conductivity instead of intrinsic permeability. Thus,

the hydraulic conductivities were converted to permeability (Table 2-2) using a viscosity

of 0.000966 Pa-s and density of 1023 kg/m3 at a temperature of 25C and a salinity of 35

kg/m3.

Based on the grain size classification, Group 1 consists only of hemipelagic

samples that are representative of Barbados. Group 2 consists of hemipelagic samples

from Barbados, Nankai and Costa Rica while Group 3 consists of Nankai and Costa Rica

hemipelagics. In general the permeability-porosity relationship obtained from grain size

classification suggests an increase in permeability with a decrease in clay size particles.

The R2 values predicted from the permeability-porosity relationships show R2 > 0.5 for

Groups 1 and 2 and 0.48 for Group 3 (Figure 2-7). It should be noted that the 95%

confidence intervals for the grain size groupings show an overlap, suggesting that the

permeability-porosity relationships are not statistically significantly different from each

other. This overlap is probably caused by the difficulties in permeability measurements,

and the likelihood that sediments with slightly varying grain size percentage (e.g., 79%

clay versus 81% clay) may not have distinguishable values of permeability, despite being

classified in different categories. Although the R2 values predicted from grain size

classification were higher than the ones predicted by depositional environment only, it

predicted lower R2 values than those predicted from classification based on location

except for Barbados. This may suggest that varying deformational processes at different

locations may affect sediment properties at varying degrees.








37



-12 II -12
A B

-14 14







1 + 18

-20 2 -2 0 F


Group 1 Group2
> 80% Clay 60-80%Clay
-22 + Barbados logk=-2428+1132n (R2 053)- -22 logk=-1973+449n(R2=056)
Costa Rica --- Bryant (2002) -- Brynt (2002)
Sx Nnkai log k=-20 90 +654n -U k og k =-20 53 +61 n

-24 1 1 1 1 -24
0 02 04 06 08 1 0 02 04 06 08
Porosity Porosity
12



-14






S-18






Group 3
22 Silty-clay with<60%clay and <5%sand
log k =-19 91 +5 45n,(R2 =043)
Bryant (2002)
log k -2059 + 6 77n
24 I I I
0 02 04 06 08 1
Porosity


Figure 2-7. Permeabilities classified based on grain size distribution. Solid lines

represent permeability-porosity relationship predicted for samples used in this
study. Dashed line represents permeability-porosity relationships of Bryant

(2002).


In order to understand the general trend of permeability-porosity relationships


based on grain size distribution, I compared my data with that of Bryant's (2002) Gulf of


Mexico data, which were obtained through consolidation tests. The dashed lines in


Figure 2-7 represent the relationships of Bryant (2002) and solid lines represent predicted


relationships from this study. For Group 1, the predicted permeability from this study


matched the permeabilities of Bryant's (2002) at porosities around 0.6 to 0.7. As









porosities decrease our relationship for Group 1 predicted lower values of permeability

than Bryant's (2002) permeability-porosity relationship. The large discrepancy between

Bryant's (2002) and this study's relationship in Group 1 might be caused by the narrow

range of porosities (0.50-0.70) represented by the Group 1 data. In contrast Groups 2 and

3 represent a larger range of porosities between approximately 0.25-0.70. Group 3 shows

fairly similar permeability-porosity relationships to those of Bryant's (2002) while our

Group 2 relationship crosses Bryant's (2002) permeability-porosity relationship at a

porosity of 0.5.

Effects of Structural Domain

Samples that were used for the grain size analyses were further categorized based

on the structural domains of each sample to test the effect of deformation on the

permeability-porosity relationship. Samples that represented the underthrust sediments

and the incoming sediments at reference sites were grouped together as they represented

the undeformed or minimally deformed sediments of the subduction complex. Prism

sediments and sediments that represented the decollement zone were grouped together as

these samples generally are highly deformed during the accretionary process. Only grain

size Group 2 of our samples had a considerable amount of samples that represent both the

underthrust/incoming sediments and the prism/decollement sediments. Using the data

from Group 2, I fitted linear regression lines for both the underthrust/incoming sediment

and the prism/decollement sediment groups (Figure 2-8).










-12 I I I


-14


-16
-14 -







20
-20- -


< Group 2 (underthrustlincoming)
--- logk =-19 B6 +446n; (R2 D 64)
-22 Group 2 (prismidecollement)
log k = -21 DO + 9 33n, (R2 = D 91)


-24 I I I I 1
0 02 04 06 08 1
Porosity
Figure 2-8. Permeabilities classified based on structural domain. Permeability-porosity
relationships only shown for underthrust/incoming samples (dashed line) and
prism/decollement samples (solid line) of Group 2.

The prism/decollement sediments group only contained eight samples and thus,

gave a relatively high R2 value of 0.91. The underthrust/incoming sediment group

contained twenty-two samples and showed moderate correlation between permeability

and porosity with a R2 value of 0.64. Even with the limited data available, the two

relationships exhibited similar permeabilities at lower porosity values (0.2-0.3). With

increasing porosities the two relationships predicted permeabilities that diverge from each

other. For example, the permeability-porosity relationship of prism/decollement group

predicted a permeability value of 1 x 10-16 m2 at a porosity value of 0.50 while at the

same porosity level the underthrust/incoming group predicted a lower permeability value

of 1 x 10-18 m2. The differences exhibited in these two permeability-porosity

relationships, while inconclusive, suggest that it might be useful to study more samples

representing each group, particularly the prism/decollement group, to test the predicted

relationships.









Discussion

The hemipelagic samples used in this study were well confined within the limits of

Neuzil's (1994) plot of argillaceous sediments. Although hemipelagic samples plotted in

the same region, the predicted permeability-porosity relationships varied with location.

In the case of Barbados, the permeability-porosity relationship showed scatter even

within an individual location due to variations found in the hemipelagic sediments.

These varying relationships suggest that samples categorized as "hemipelagics" could

have different permeability-porosity relationships at different locations. Thus, combined

classification of both depositional environment and location may provide better

correlation between permeability and porosity.

In general the observed differences in the log permeability-porosity relationship

appears correlated to the amount of clay and silt size particles present in the sample.

Except for Group 1, our predicted permeability-porosity relationships based on grain size

distribution are in good agreement with the relationships of Bryant (2002). Groups 2 and

3 exhibited similar trends as those predicted by Bryant's (2002). This similarity suggests

that even though samples were taken from different subduction zones, samples classified

based on their grain size distribution exhibit similar trends between permeability and

porosity compared to those samples that were taken from a single location (e.g., Bryant's

(2002) samples from Gulf of Mexico). Thus, it could be concluded that the permeability-

porosity relationships obtained based on grain size distribution are generally applicable to

samples from marine settings. However, additional data, particularly in Group 1, would

further test this conclusion.

At high porosities, the relationship predicted for the siliceous oozes suggests

similar permeabilities to those predicted for hemipelagic sediments. However, there were









no samples that represent lower porosities (less than 0.45) of the siliceous oozes and

therefore the predicted relationship should be used with caution at lower porosities.

The lack of correlation exhibited between permeability and porosity in calcareous

oozes suggest that one should consider other variables such as depth, consolidation rates

and relative age of the sediment to obtain meaningful relationships for pelagic sediments.

According to Bryant et al. (1981), unlike grain size, the influence of calcium carbonate is

more pronounced with increasing depth because as burial increases less reduction of

porosity is observed for calcareous sediments compared to non-calcareous sediments.

Mechanical compression tests conducted by Terzaghi (1940) and Robertson (1967)

demonstrated that calcareous muds compact less than non-calcareous muds. A similar

study by Bryant et al. (1981) demonstrated that under a similar load, carbonate sediments

do not consolidate to as low a void ratio as non-carbonates. Based on this finding they

speculated that the resistance towards consolidation in carbonates could be a result of the

relative age of the sediment or the differences in particle shape or the structural strength

of the individual particles.

Even though comparison of underthrust/incoming and prism/decollement structural

domains from Group 2 suggested the possibility of different varying permeability-

porosity relationships, it is recommended to further investigate these relationships using

more samples; the results presented here are based on a limited number of samples

especially those represented by the prism/decollement group. It would be worthwhile to

further test the effects of structural domain on permeability-porosity relationship, as this

information will allow future studies based on permeability-porosity relationships to

more realistically represent fluid flow.









This study used a large number of data from many different sources to investigate

the relationship between permeability and porosity. Assembling data for this study

highlighted the importance of documenting detailed methodology including critical data

such as the temperature during permeability testing and the type of permeant used in the

flow test. It is also important to have porosity and grain size data available for

documented permeability values, as these parameters are valuable estimating

relationships between permeability and porosity. This study also identified gaps in

available data. For example, only a few sediment samples were available from the

structural domain representing the prism/decollement and few samples from Barbados

were tested at low porosities. Thus, future data collections should focus on collecting

samples representing the prism and the decollement as well as deeper parts of the

underthrust sediments, where much remains to be learned.

Conclusions

I examined permeability-porosity relationships for sediments from four different

subduction zones based on their depositional environment, grain size distribution, and

structural domain. Greater correlation was observed between permeability and porosity

for hemipelagic sediments and for siliceous oozes while relatively low correlations were

predicted for calcareous oozes. Based on the hemipelagic samples used in this study, it is

clear that permeability-porosity relationships vary among hemipelagics at different

locations and thus, classification based on depositional environment should be used with

caution when applied at different locations. Grain size predicts more meaningful

correlation between permeability and porosity than depositional environment only, and

these relationships are generally consistent with results from other marine settings.

However, grain size classification shows less correlation between permeability and






43


porosity relative to the relationships predicted by location. Due to lack of data, the effect

of structural domain on permeability-porosity relationship could not be evaluated. To

predict meaningful relationships for permeability of carbonaceous sediments, one should

consider other factors such as depth, consolidation rates, and relative age of the sediment.














CHAPTER 3
CHARACTERIZATION OF EXCESS PORE PRESSURES AT THE TOE OF THE
NANKAI ACCRETIONARY COMPLEX, OCEAN DRILLING PROGRAM SITES
1173, 1174, AND 808: RESULTS OF ONE-DIMENSIONAL MODELING

Introduction

Examining the fluid flow of the deep hydrosphere is extremely important because

fluid flow alters the physical and chemical properties of the Earth's crust, affecting the

ocean and the atmospheric chemistry that is vital for human existence (COMPLEX,

1999). At active plate margins, fluid pressures can influence movement along faults and

thus, the nature of the earthquake cycle (Moore and Vrolijk, 1992). Pore fluid pressures

build up in areas where sediment permeabilities are low enough to prevent pore fluid

escape at a rate comparable to the rate of loading due to tectonic and gravitational

stresses. These elevated fluid pore pressures play an important role in the development

of accretionary complexes. It is speculated that high pore pressures control the formation

of the decollement, which separates highly deformed overlying wedge sediments from

slightly deformed underthrust sediments (Westbrook and Smith, 1983). Furthermore,

pore pressures affect deformation within and the taper angle of the accretionary wedge

(Davis et al., 1983).

At the Nankai accretionary complex, a thick terrigenous sequence rapidly deposits

over a low-permeability hemipelagic sequence. Previous studies have used indirect

estimates to document the development of pore pressures within the underthrust

sediments at the toe of the Nankai accretionary complex. For example, Screaton et al.

(2002) estimated depth-averaged excess pore pressures using porosity-depth profiles.









The estimated overpressures within the underthrust sediments suggest insufficient

permeability for fluid escape at a rate comparable to sediment loading. Saffer (2003)

used shipboard observations of porosity and laboratory measurements of consolidation to

estimate pore pressures and evaluate their variations down section. His results indicate

undrained conditions in the underthrust sediments that may have been caused by rapid

sedimentation and loading due to underthrusting. A modeling study conducted by Le

Pichon and Henry (1992) suggested that trench sedimentation and rapid accretion at the

toe of the prism could generate excess pore pressures in the underthrust sediments at and

even seaward of the deformation front. However, that study did not have permeability

data, and used estimates based on lithologies.

In this study, I used laboratory permeability data to investigate porosity reduction

and the generation of excess pore pressures at the toe of the Nankai accretionary

complex. Based on measured permeabilities I developed a permeability-porosity

relationship for hemipelagic sediments at the Nankai accretionary prism. I then used this

permeability-porosity relationship in a one-dimensional numerical model to simulate the

excess pore pressures and porosities that would result from sedimentation and loading by

the prism at the toe of the accretionary complex. Finally I tested the sensitivity of excess

pore pressures and porosities to bulk permeability, lateral stresses within the prism, and a

hypothetical low-permeability barrier at the decollement.

Background

Geologic Setting

The Nankai accretionary complex is formed by the subduction of the Shikoku

Basin on the Philippine Sea Plate beneath the southwest Japan arc on the Eurasian plate

(Figure 3-1) at a rate of 4 cm/yr (Seno et al., 1993). This study focused on the Muroto









Transect (Figure 3-1) where the thickness of the complex varies from 750 m at the toe to

-4500 m at 50 km arcward. Along the Muroto Transect, the prism toe has a taper angle

of 40-5 (Shipboard Scientific Party, 2001). Based on this low taper angle, it has been

inferred that the Muroto Transect has high decollement pore pressures or low intrinsic

decollement strength (Saffer and Bekins, 2002).

During ODP Leg 131, Site 808 was drilled on the Muroto Transect approximately 3

km landward of the deformation front. A 560 m thick sequence ofturbidites was found

above the hemipelagic muds of the upper and lower Shikoku Basin facies. The

decollement zone was identified by intense brittle deformation at 945 to 964 meters

below seafloor (mbsf), and develops from a homogeneous interval of hemipelagic

mudstones within the lower Shikoku Basin facies (Shipboard Scientific Party, 2001).

During ODP Leg 190, Sites 1173 and 1174 were drilled seaward of Site 808 along the

Muroto Transect (Figure 3-2).

34'N








808
O L i 1174
1173 Muroto
32.' e4 Transect




132"E 133" 134' 135" 136'

Figure 3-1. Location map of the study area in the Nankai accretionary complex and sites
used for this study (modified from Moore et al., 2001).









Site 1174 penetrated the decollement within the proto-thrust zone (Figure 3-2).

The decollement was observed between 808 and 840 mbsf and was marked by fractures

and brecciation in the lower Shikoku Basin facies (Shipboard Scientific Party, 2001).

The turbidites extended 431 m above the hemipelagic muds of the upper and lower

Shikoku Basin facies (Shipboard Scientific Party, 2001).

NW SE
X Line900 800 700 600 500 400 300 200


Protothrust
I- zone nefnrmatinn


"s Site 1174 FrontSite 1173
E FZ! tl Trench-Wedge Facies
n Upper Shikoku Basin Facies
~ -. Lower Shikoku Basin Facies
^ V emen zone .. Volcaniclastic Facies
_.-- :-- ---_ Ocean Crust .
1


Figure 3-2. Schematic interpretation of the Muroto Transect showing tectonic domains
and location of Leg 190 drill sites used in this study (modified from Moore et
al., 2001).

Site 1173 was drilled 11 km seaward of the deformation front providing an

undeformed reference site of the incoming sedimentary sequence (Shipboard Scientific

Party, 2001). The turbidite layer at Site 1173 is much thinner (-102 m) than at Site 1174,

because it is farther away from the trench. At Site 1173, the age equivalent of the Site

1174 decollement zone occurs between 390 and 420 mbsf, within the lower Shikoku

Basin facies (Shipboard Scientific Party, 2001).

At the Nankai accretionary complex, heat flow values ranging from 180 mW/m2 at

Sites 1173 and 1174 to 130 mW/m2 at Site 808 have been estimated (Shipboard Scientific

Party, 2001). These measured high heat flow values are related to the fossil spreading

ridge represented by the Kinan Seamount on the Philippine Sea Plate which ceased

spreading 15 Ma ago (Shipboard Scientific Party, 2001). Previous studies have









suggested that high temperatures at the Muroto Transect could affect in situ dehydration

reactions such as the smectite to illite transition (Kastner et al., 1993) as well as observed

cementation by authigenic clays (Ujiie et al., 2003).

Previous Hydrologic Studies

Previous workers have examined several aspects of fluid flow and pore pressure

development at the Nankai accretionary complex. Le Pichon and Henry (1992) used a

one-dimensional model of sedimentation representing the stratigraphic sequence at Site

808. Their model consisted of a coarser terrigenous sediment layer with permeability

values of 10-16 to 10-17 m2, rapidly depositing over a less permeable hemipelagic sequence

(10-19 to 10-17 m2). Because permeability data for Nankai sediments was not available,

they used permeabilities that are representative ofuncompacted terrigenous and

hemipelagic sediments. Their modeling results showed that this stratigraphic succession

could potentially form a low mechanical resistance horizon within the upper portion of

the low permeability sediments, but testing of their model requires permeability data.

Taylor and Fisher (1993) performed permeability measurements on subsamples of

Site 808 cores using two methods. One method directly measured flow under known

head values and the other method indirectly measured flow from the rate of consolidation

due to a known applied stress. Their measured permeabilities ranged from 10-14 to 10-19

m2 in horizontal and vertical directions.

Saffer and Bekins (1998) used a log-linear permeability-porosity relationship and

an assumed porosity distribution in a two-dimensional numerical model to match pore

pressures estimated from the critical taper theory of Davis et al. (1983). The critical taper

is the shape for which the wedge is on the verge of failure under horizontal compression

and to maintain a small critical taper it requires high basal pore pressures (Davis et al.









1983). For varying assumptions about permeability and porosity distribution within the

underthrust sediments, Saffer and Bekins (1998) simulated X*values between 0.2-0.4 in

the region of Sites 1174 and 808. It should be noted that these simulations assumed a

constant decollement permeability of 10-16 m2

The estimated permeability-porosity relationships of Saffer and Bekins (1998) fall

within the range for argillaceous rocks defined by Neuzil (1994), but was approximately

two orders of magnitude lower than the direct permeability measurements given by

Taylor and Fisher (1993). Saffer and Bekins (1998) suggested that direct permeability

measurements by Taylor and Fisher (1993) were overestimates due to fabric expansion,

because these measurements were performed on samples that were not reconsolidated

under pressure. Moreover, Saffer and Bekins (1998) argued that the use of air in the

Taylor and Fisher (1993) permeability tests would tend to further overestimate

permeabilities.

Screaton et al. (2002) used porosity measurements of core samples from Sites 808

(Shipboard Scientific Party, 1991), 1174, and 1173 (Shipboard Scientific Party, 2001) to

estimate pore pressures in the underthrust sediments. In this study, Screaton et al. (2002)

assumed that the solid volume is constant thus relating the change in volume to change in

porosity. Site 1173 was assumed to provide a reasonable proxy for conditions of the

Sites 1174 and 808 sediments when they were seaward of the trench. The average

porosities of the underthrust (or equivalent) sediments of the lower Shikoku Basin facies

decrease landward from 0.42 at Site 1173 to 0.34 and 0.33 at Sites 1174 and 808,

respectively (Screaton et. al., 2002). The porosity profile from Site 1173 shows steadily

decreasing porosity with increasing depth (Figure 3-3).









Site 808 Site 1174 Site 1173
0 .0 0

200 200 o 200

400 '- 400 0 400
E E E
S600 600 600

800 n 800 O 800

1000 1000 1000

1200 1200 1200
0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6 0.8
Porosity Porosity Porosity

Figure 3-3. Porosity profiles of Sites 808, 1174, and 1173. Light shading shows lower
Shikoku Basin facies. Darker shading represents the decollement zone at
Sites 808 and 1174 and at Site 1173 the age-equivalent level of Site 1174
decollement zone.

At Sites 1174 and 808 the porosities below the decollement increase relative to the

porosities above. Comparison between the porosity profiles of the underthrust sediments

from Sites 1174 and 808 with the reference site (Site 1173) suggest excess pore pressure

ratio (X*) of 0.42 at Site 1174 and 0.47 at Site 808 (Screaton et al., 2002), where X*

represents the magnitude of excess fluid pressure relative to sediment overburden

pressure and is defined as

= (P PhPPh)/(P P (1)

where P is pore fluid pressure [M L1T-2], Ph is hydrostatic pressure [M L1T-2], and Pi is

lithostatic pressure [M L1T-2]. The excess pore pressure ratio removes the effect of the

overlying water column, and describes excess fluid pressures within the sedimentary

sequence relative to the overburden load minus hydrostatic pressure. For lithostatic pore

pressures, A* = 1, whereas at hydrostatic pressures, A* = 0.









Saffer (2003) integrated measurements of logging-while-drilling (LWD), physical

properties data, and laboratory consolidation tests to compare pore pressure development

and progressive dewatering in the underthrust sediments. Calculations of k* based on

Saffer's (2003) results indicate values of 0.30 and 0.44 at 15 m below the decollement at

Sites 1174 and 808, respectively. These values are slightly lower than the average values

over the entire underthrust profile inferred from porosity data by Screaton et al. (2002).

The estimated k* values calculated from results of Saffer (2003) generally decrease with

depth below the decollement at both Sites 1174 and 808.

Evidence concerning possible lateral fluid flow is ambiguous. Broad low chloride

anomalies were observed within the lower Shikoku Basin facies (Kastner et al., 1993;

Gieskes et al., 1993; Spivack et al., 2002). Controversial ideas have been suggested as

possible explanations for the observed low chloride anomalies. Kastner et al. (1993) and

Underwood et al. (1993) suggested that smectite dehydration is the most likely

mechanism for the observed low-chloride anomaly. On the basis of geochemical

analyses, Kastner et al (1993) suggested lateral fluid flow along the decollement or along

a deeper conduit within the underthrust sediments while Spivack et al. (2002) suggested

continuous lateral fluid flow between Sites 808 and 1174 within the underthrust

sediments. Brown et al. (2001) suggested that in situ dehydration of-10-15 wt% of

smectite could generate most of the freshening observed at Site 808 and therefore,

provides a possible explanation for pore fluid freshening. However, if the initial amount

of smectite is less than 10-15 wt% then the low-chloride anomaly could reflect the

combined effects of both in situ and lateral flow at depth (Brown et al., 2001). This in

situ dehydration hypothesis has been further supported by a recent study by Henry et al.









(2003) using cation exchange capacity. Henry et al. (2003) showed that there are

sufficient amounts of smectite to explain the chlorinity anomalies by in situ reactions, and

thus, lateral fluid flow is not required. Although fluid from dehydration greatly affects

pore water chemistry, Saffer and Bekins (1998) suggest that fluid sources from

dehydration of smectite are 10-1000 times smaller than the compaction fluid sources.

Thus, dehydration fluid sources are unlikely to be significant in generating pore fluid

pressures at the toe of the prism.

Laboratory Permeability Measurements

Although previous studies have provided evidence for excess pore pressures at the

Nankai accretionary complex, the causes of it were not sufficiently characterized due to

lack of well-constrained data in sediment properties. Because sediment permeability is

the most important factor in controlling modeled pore pressures, it is important to

establish a systematic relationship between permeability and porosity to approximate

permeability in the accretionary complex (Saffer and Bekins, 1998). Thus, as the first

step of the study I used laboratory measured permeability data from nine core samples at

varying depths to establish a permeability-porosity relationship. Gamage and Screaton

(2003) provide detailed information about test methods. Vertical permeability tests were

performed using the constant flow method on nine core samples taken from the Shikoku

Basin facies at Sites 1173 and 1174. The constant flow permeability method induces a

hydraulic gradient across the sample, and measurement of the pressure difference allows

determinations of permeability. To determine porosities for the permeabilities presented

by Gamage and Screaton (2003), I calculated change in porosity during consolidation

based on the change in volume of fluid [L3] contained in the cell at the end of each










consolidation step, and used shipboard porosity measurements for the initial porosities


(Table 3-1).


Table 3-1.


A summary of laboratory measured permeabilities for samples from ODP Leg
190 Sites 1173 and 1174. Permeability values were recalculated (as compared
to Gamage and Screaton, 2003) based on a viscosity value of 0.0008 kg/s.m
and a density of 1020 kg/m3, based on laboratory temperatures during testing
(Lide, 2000). Values for Sample 1174B-84R-3, which are included in
Gamage and Screaton (2003) are not presented here due to insufficient data to
calculate norosities at measured nermeabilitv values.


-----------I --- ^ ^__ __ __ I- - -_._ _--------------------
Sample Effective Stress Porosity Permeability
(MPa) (m2)
1173A-22H-2, 199.9 mbsf, Silty clay 0.24 0.57 5.14 x 1017
0.42 0.56 4.02 x 10-17
0.54 0.55 3.90 x 10-17
1173A-31X-1, 284.59 mbsf, Silty claystone to 0.27 0.62 1.98 x 1017
0.42 0.60 1.28 x 10-17
1173A-39X-5, 367.07 mbsf, Silty claystone, 0.26 0.41 2.46 x 10-8
0.39 0.36 2.18 x 1018
0.83 0.33 1.51 x 1018
1173A-41X-cc, 388.75 mbsf, Silty claystone, 0.29 0.45 2.48 x 1018
0.41 0.43 1.66 x 1018
0.55 0.43 1.18 x 1018
1173A-46X-1, 428.59 mbsf, Silty claystone 0.25 0.45 1.90 x 1018
0.40 0.43 1.52 x 1018
0.51 0.41 1.24 x 1018
1174B-42R-3, 538.23 mbsf, Silty claystone, altered 0.19 0.37 7.84 x 1018
0.48 0.34 2.00 x 1018
0.62 0.32 1.52 x 1018
1174B-59R-5, 704.95 mbsf, Silty claystone 0.55 0.32 5.18 x 10-19
0.69 0.30 3.53 x 10-19
0.83 0.28 2.94 x 10-19
1174B-69R-2, 795.17 mbsf, Silty claystone, 0.55 0.29 8.48 x 10-19
0.69 0.28 5.78 x 10-19
0.83 0.26 2.83 x 10-19
1174B-74R-1, 842.75 mbsf, Silty claystone- 0.55 0.30 6.40 x 10-19
0.69 0.28 6.16 x 10-19
0.83 0.27 4.70 x 10-19

For clay-rich sediments, Bryant et al. (1975) and Neuzil (1994) established that

logarithm (base 10) of permeability (k) decreases systematically with decreasing porosity


(n) and thus, k can be represented as a function of porosity,










log (k) = log (ko)+pn (2)

where ko is the projected permeability at zero porosity and yp is a parameter describing the

rate of permeability change with porosity.

With respect to Neuzil's (1994) compilation of permeability data for argillaceous

sediments, most of the Nankai core permeability values fall in the central portion of the

data range (Figure 3-4), while a few permeability values with porosities between 0.2 and

0.3 fall closer to the higher-permeability margin of the data range. The measured

permeability values are best fit by the following relationship:

log (k)= -19.82+5.39n (3)
-14

-16

-18


0 -20

-22 -


0 0.2 0.4 0.6 0.8 1
Porosity

Figure 3-4. Permeability data measured for samples from ODP Leg 190 (solid diamonds)
superimposed on a large-scale permeability versus porosity outline for
argillaceous sediments compiled by Neuzil (1994). Solid line represents the
best-fit line for the measured permeabilities. The dashed line with + mark
represents the permeability-porosity relationship for the lower range of
measured permeability values. The shaded area represents the range of
permeability-porosity relationships established by Saffer and Bekins (1998),
with the dashed line (log (k)=-20+5.25n) indicating a mid-range value.

The permeability-porosity relationship obtained from this study has a similar slope

to the relationships used by Saffer and Bekins (1998) (log (k) = -20+5.25 n) but predicts

slightly higher permeability values (Figure 3-4). The lower range of permeability values









are near the bottom of the range used by Saffer and Bekins (1998) relationship, and can

be represented by a line with the relationship, log (k) = -20.15 + 5.39n (Figure 3-4).

Modeling Methods

Theoretical Background

As sediments are loaded, the sediment column beneath will compact if the pore

fluid can escape at a rate comparable to the loading rate. However, if the permeability of

the sediment column is low, dewatering will be inhibited, excess pore pressures will

build, and compaction will be prevented. Darcy's Law expresses the relationship

between pore pressures, sediment properties, and fluid velocities. For constant density

saturated flow, Darcy's Law can be written in terms of the hydraulic head (Voss, 1984).

v =K Vh (4a)


P
where h = -- + h (4b)
Pfg

where v is pore fluid velocity [L T-], Kis hydraulic conductivity [L T-], n is porosity, h

is hydraulic head [L], pfis fluid density [M L-3], g is gravitational acceleration [L T-2],

and hz is elevation head [L]. Hydraulic conductivity is defined by K=kpfg /, where k is

intrinsic permeability [L2] and p is dynamic viscosity [M L-1 T-1]. Intrinsic permeability

is a function of the porous medium while fluid density and viscosity depend on the

properties of the fluid and may change with temperature, salinity, and to a lesser extent

with pressure.

Combining the mass conservation of fluid with Darcy's law (Eq. 4a), the following

equation can be written for diffusion of head in porous media for saturated fluid flow.

?h
V(KVh) = S + Q (5)
-at









where S, is specific storage [L-1], and t is time [T]. Specific storage is a measure of the

volume of formation fluid per unit volume of media per unit head of a saturated

formation that is stored or expelled from storage due to compressibility of the matrix and

change in fluid pressure. Specific storage is defined as Ss=pfg(a+nl) where a is matrix

compressibility [M-1 L T2], and fl is the compressibility of the formation fluid [M-1 L T2].

Matrix compressibility (a) is defined by a = [1/(V)](dV /doe) where V is the bulk

volume [L3] and oe is the effective stress [ML-1T-2]. The second term on the right hand

side is a source term representing changes to fluid volume or pressure, such as due to

loading, per time per unit volume of media [T-1].

During loading, the stress added from the sediment load is partitioned between the

pore fluid and the matrix. The loading efficiency, y, denotes the fraction of the stress

added to the pore fluid pressure and, assuming the sediment grains are incompressible

relative to the matrix, is defined as follows (after Neuzil, 1986)

y= a/(a+nf ) (6)

For highly compressible sediments, the loading efficiency is near 1.

The increase in pore pressure caused by loading drives fluid flow. When initial

conditions, boundary conditions and the controlling hydrogeologic parameters k and a

are known, Equation 5 can be solved for hydraulic head at any point at any time, and

thus, pore fluid velocity relative to the sediment framework can be calculated using

Equation 4a.

The pore pressure is related to effective stress (Ge) by


o O P









where ct is the total stress. As fluid escapes, pore pressure decreases and effective stress

increases, causing porosity to decrease. For sediments that are normally compacting

(sediments at hydrostatic pore pressure), porosity has been observed to decrease

exponentially with depth, as suggested by Athy (1930)

n = noexp[-bz] (8)

where no is initial porosity, b is a constant [L-1], and z is burial depth [L]. For hydrostatic

conditions, the change in effective stress, oe, with depth is given by

(do-, / dz) = (po,- ,of) (l-n) g (9)

where p, is grain density. Combining Equations 8 and 9 results in

(dn / do ) = (-bn) / [(p, f) (1- n) g] (10)

Equation 10 is only applicable for sediments undergoing compaction. For this

investigation, it is assumed that sediments cannot decompact or expand when the

effective stress is reduced. Thus, porosities will remain constant unless the effective

stress exceeds the previous maximum effective stress value. The change in volume (dV)

can be related to (dn) by

dV/V= (dn) / (1 n) (11)

In one dimension, the volume change represents only the change in vertical

thickness and thus the horizontal dimension of the sediment column stays constant. As

porosity decreases with depth, matrix compressibility (a) is also reduced. Based on

Equations 10 and 11, the matrix compressibility is calculated as

a= (bn)/[(p, pf) (1-n)2 g] (12)









Model Implementation

This modeling study focused on Sites 1173, 1174, and 808 to examine the

development of pore pressures along the Muroto Transect during early subduction.

Modeling is one-dimensional, and thus, lateral fluid flow cannot be included. However,

as discussed above, evidence for lateral fluid flow at this Transect is inconclusive. A

method previously described by Screaton and Ge (2000) and Kemerer and Screaton

(2001) was modified for this investigation to add the permeability-porosity relationship

(Eq. 2). The modeling method combines a loading program to simulate pore pressure

increases due to sedimentation and initial subduction with an existing fluid flow and

transport code, SUTRA (Voss, 1984).

Based on the rate of sedimentation or thickening of the overriding prism for each of

the segments, the loading program calculates the additional thickness of each added

sediment layer. As new sediment layers are added to the top of the model, the layers

beneath were moved down one row, and their pore pressures were incremented with the

additional pore fluid pressure due to the load of the new layers multiplied by y (Eq. 6).

The additional load is calculated from the thickness and bulk density of the new layer,

with the hydrostatic pressure subtracted.

The updated pore pressures were used as inputs to SUTRA as initial conditions to

perform transient fluid flow simulations. Once the pore pressures (P) at the end of each

loading step were calculated by SUTRA, they were transferred back into the loading

program, in which effective stress was calculated using Equation 7. If effective stress

increased relative to the previous loading step, the porosity decrease was determined

using an iterative method to solve Equation 10 for change in porosity. Vertical spacing









was reduced to maintain constant solid volume and to ensure mass balance (Eq. 11). The

new porosity values were then used to calculate the matrix compressibility for the next

sedimentation step (Eq. 12). SUTRA was modified to calculate specific storage for each

node based on the compressibility and porosity calculated by the loading program.

Model Dimensions, Boundary Conditions, and Initial Conditions

The geometry and hydrogeology of the Nankai accretionary complex were

simplified into three hydrogeologic units including the turbidites, upper Shikoku and

lower Shikoku Basin hemipelagic sequences. The model domain was discretized into

200 rows. Because the model is one-dimensional, lateral flow along the decollement

could not be simulated and thus, special physical properties such as higher permeabilities

were not assigned to the decollement.

The upper and lower Shikoku Basin layers were assigned similar porosity and

permeability parameters as the two layers were mainly composed of hemipelagic muds.

The physical properties for hemipelagic units were assigned based on geological

observations and laboratory measurements. The lithology of the turbidite unit at Nankai

is composed of a variety of thick to thin bedded sand and silt turbidites with some

hemipelagic muds (Shipboard Scientific Party, 2001). Values of no (0.77) and b (1.1 x

10-3 m-1) for hemipelagic sediments were obtained from Screaton et al. (2002) while for

turbidites (no=0.65, b = 7 x 10-4 m-1), the porosity-depth relationship of Bekins and Dreiss

(1992) was used. To test the sensitivity of the b coefficient for hemipelagic sediments, I

used the porosity data from Site 1173 to estimate the standard deviation for the b values

as 3 x 10-4 m-1. Using the standard deviation, I checked the sensitivity ofb values at the

minimum possible value of 1.4 x 10-3 m-1 and a maximum possible value of 8 x 10-4 m1.

For the turbidites, I used b values of 1.0 x 10-3 m-1 and 4 x 10-4 m-1 as the maximum and









minimum based on the b values used by Bekins and Dreiss (1992). The estimated k*

values were only increased by 0.04 at the maximum b value while k* values decreased by

the same amount at the minimum b value, suggesting the sensitivity to the b value is

small.

The permeability-porosity relationship for the hemipelagic sediments was obtained

from the laboratory permeability measurements (Equation 3). Generally, the sandy

portions of the turbidites would be expected to have higher permeabilities than the

hemipelagic sediments. However, because the lowest permeability layer generally

controls vertical fluid flow (e.g., Bear and Verruijt, 1987), I used the same permeability-

porosity relationship for both turbidites and hemipelagics. I examined the effects of the

permeability-porosity relationship by varying the log (ko) value for both hemipelagics and

turbidites during the sensitivity analyses.

The top boundary of the model was set as hydrostatic to simulate the effect of the

water column above the complex. I assume that the hydraulic connection between the

sediments and the oceanic crust is low due to an ash layer that would be expected to alter

to low-permeability clays (Saffer and Bekins, 1998). Thus, permeability of 10-23 m2 was

assigned to all elements below the lower Shikoku basin facies. The ocean crust consisted

of sufficient rows at the beginning of the simulation so that only "ocean crust" rows

would be dropped from the base of the model as the sediment layers are added, to

maintain a constant number of elements throughout the simulation. Matrix

compressibility of the ocean crust was set at 1.0 x 10 11Pa-.

To approximate the effects of in situ temperature on viscosity, I applied

temperature and heat flow boundary conditions, and used SUTRA to calculate









temperature distribution. Both the seafloor and oceanic crust boundaries were treated as

specified temperature boundaries. The seafloor boundary was set at 2 C while the heat

flow at the base of the model was assigned to produce temperatures consistent with those

obtained from shipboard measurements. Thermal conductivities and specific heat values

for both fluid and solid matrix are given in Table 3-2. SUTRA then calculated fluid

viscosity as a function of temperature using the following relationship (Voss, 1984)

248.37
p(T) (239.4*10 10 0 (T+133"15) (13)

where p is pore fluid viscosity [kg m-s-1] and T is temperature in C.

Table 3-2. Fluid and solid matrix properties used for numerical simulations.
Parameter Value
Fluid compressibility [Pa-1] 4.40E-10
Fluid density [kg m-3] 1035
Fluid specific heat [J kg-OC-'1] 4180
Fluid thermal conductivity [J s-' m-1 "C-1] 0.7
Solid grain density [kg m-3] 2650
Solid grain specific heat [J kg-'lC -] 1000
Solid grain thermal conductivity [J s-1 m-1 C-1] 3.0

Sedimentation and Prism Thickening Rates

Sedimentation seaward of the deformation front and loading due to the over-riding

prism were applied separately in two different phases (Table 3-3). During phase one,

sedimentation rates calculated from the biostratigraphy were used to build the sediment

columns at Sites 808, 1174, and 1173. At Site 808 and 1174, the sediment columns were

built using 137 time steps of 100,000 years and at Site 1173, 140 time steps were used.

The average initial sedimentation rate for each unit was based on the initial thickness of

that unit and its corresponding deposition time as provided by the Shipboard Scientific

Party (2001). The initial thickness of each unit (i.e., the thickness of the sediment layer

prior to consolidation) was calculated based on an assumed initial porosity of 0.77 and









0.65 for hemipelagics and turbidites respectively, and a final porosity based on core

measurements of shipboard porosity (Shipboard Scientific Party, 2001). During phase

two, prism thickening rates for Sites 1174 and 808 were calculated using change in prism

thickness and a convergence rate of 4 cm/yr. The convergence rate was used to calculate

the time needed for the incoming sediments to underthrust to Sites 1174 and 808. In

phase two, each stress step was 50,000 years, and Site 1174 was loaded for 6 stress steps

while Site 808 was loaded for 7 stress steps.

Table 3-3. Summary of sedimentation rates calculated from biostratigraphy at Sites 1173,
1174 and 808 and prism thickening rates calculated from prism geometry and
convergence rate.
Site Unit Thickness Time Vertical loading
(m) (Ma)
Initial Prism
sedimentation rate thickening rate
(m/yr) (m/yr)
1173 Turbidites 120 Present-0.25 5.07 x 10-4
upper Shikoku 220 0.25-2.85 7.42 x 10-5
lower Shikoku 50 2.85-3.75 1.21 x 10-4
Proto-decollement 25 3.75-4.95 4.55 x 10-5
lower Shikoku 290 4.95-14.00 8.08 x 10-5
1174 Turbidites 420 Present-0.30 1.9 x 10
Trench-wedge faces 50 0.3-0.90 1.91 x 10-4
upper Shikoku 190 0.90-2.5 2.61 x 10-4
lower Shikoku 149 2.5-6.25 1.12 x 10-4
Decollement 25 6.25-7.15 7.85 x 10-5
lower Shikoku 256 7.15-14.30 1.20 x 10-4
808 Turbidites 550 Present-0.35 3.6 x 10-3
Trench-wedge faces 75 0.35-0.95 4.28 x 10-4
upper Shikoku 188 0.95-2.85 2.20 x 10-4
lower Shikoku 125 2.85-5.35 1.46 x 10-4
Decollement 25 5.35-6.35 7.30 x 10-5
lower Shikoku 265 6.35-14.35 1.03 x 10-4

Results

Model Results

Results of the base run using the one-dimensional loading and fluid flow model are

summarized in Table 3-4. For the base run, values of ko and (p were assigned according









to the permeability-porosity relationship determined from laboratory measurements

(Equation 3). The simulated porosity profiles at all three sites decrease smoothly with

depth in the lower Shikoku Basin facies. At Site 1173, the simulated porosity profile

indicates a small overestimate of porosities compared to the observed porosity profile

(Figure 3-5). This reflects that Site 1173 is slightly overpressured (k*=0.04) in the

simulation (Figure 3-5), whereas the porosity-depth relationship assumed hydrostatic

conditions at Site 1173 (Screaton et al., 2002). This would imply that k* estimated based

on the assigned b value would be an underestimate.

It should be noted that results of Screaton et al. (2002) and Saffer (2003) make a

similar assumption of hydrostatic conditions at Site 1173. As a result, their pore pressure

results may also underestimate actual pore pressure ratios. At both Sites 1174 and 808,

the simulated porosities reasonably match the observed porosities above the decollement

while underestimating the porosities below the decollement (Figure 3-6). The estimated

k* values below the decollement at Site 1173 slightly increase with depth while at Sites

1174 and 808, k* values slightly decrease with depth below the decollement. Simulated

excess pore pressure ratios, k*, observed at 15 meters below the decollement increase

landward from 0.04 at Site 1173, to 0.14 and 0.22 at Sites 1174 and 808 respectively

(Figure 3-5 and Figure 3-6), but are less than that inferred by Screaton et al. (2002) and

Saffer (2003). Thus, using the permeability-porosity relationship from Equation 3, the

simulations suggest that sedimentation and prism loading alone is not sufficient to

generate excess pore pressures as large as predicted from previous studies.











Table 3-4. Summary of simulation runs at Site 1174 and 808. Values of X* and porosity
for each simulation are given at 15 m below the decollement (855 mbsf at Site
1174 and 979 mbsf at Site 808). At these depths, shipboard measurements
indicate porosities of 0.37 at Site 1174 and 0.36 at Site 808 (Shipboard
Scientific Party, 1991; 2001). Italics indicate parameters that were changed
from the base run.
Run Unit log

(ko) eral Low k 1174 1174 808 808
Stress Barrier X* n ,* n
Ratio (m2)
Base Turbidites -19.82 5.39 0 NA 0.14 0.31 0.20 0.28
run Hemipelagics -19.82 5.39
Saffer and Turbidites -20.00 5.25 0 NA 0.20 0.33 0.31 0.33
Bekins Hemipelagics -20.00 5.25
(1998)
Raise Turbidites -16.82 5.39 0 NA 0.10 0.30 0.15 0.28
turbidite k Hemipelagics -19.82 5.39
Raise Turbidites -18.82 5.39 0 NA 0.06 0.29 0.10 0.25
bulk k Hemipelagics -18.82 5.39
Lower Turbidites -20.82 5.39 0 NA 0.47 0.45 0.47 0.42
bulk k Hemipelagics -20.82 5.39
Lower Turbidites -20.15 5.39 0 NA 0.23 0.34 0.29 0.32
bulk k Hemipelagics -20.15 5.39
Vary Turbidites -19.82 5.39 0.1 NA 0.13 0.31 0.20 0.29
lat.stress Hemipelagics 19.82 5.39
Vary Turbidites -19.82 5.39 0.3 NA 0.15 0.31 0.21 0.29
lat.stress Hemipelagics 19.82 5.39
Vary Turbidites -19.82 5.39 0.5 NA 0.18 0.32 0.23 0.30
lat.stress Hemipelagics -19.82 5.39
Vary Turbidites -19.82 5.39 0.7 NA 0.21 0.32 0.26 0.30
lat.stress Hemipelagics -19.82 5.39
Vary Turbidites -19.82 5.39 0.9 NA 0.24 0.32 0.29 0.31
lat.stress Hemipelagics -19.82 5.39
Lower Turbidites -20.15 5.39 0.3 NA 0.26 0.35 0.32 0.34
bulk k + lat Hemipelagics -20.15 5.39
stress
Low k Turbidites -19.82 5.39 0 10 0.15 0.31 0.21 0.29
barrier Hemipelagics -19.82 5.39
Low k Turbidites -19.82 5.39 0 1 0.28 0.35 0.34 0.32
barrier Hemipelagics -19.82 5.39
Low k Turbidites -19.82 5.39 0 10 0.45 0.40 0.52 0.39
barrier Hemipelagics -19.82 5.39









Site 1173 b Site 1173
a 350 350


400 400

0 ++a+ + 1
S450 450 -
C-

1 500- 500 -

550 550 -
+

600 I 600
0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3
Porosity ;*

+ Observed Porosities
--- Base Run


Figure 3-5. Simulated porosity and X*profiles for base run at Site 1173. A) Comparison
of observed porosity to simulated porosity at Site 1173. B) Simulated '*
profile at Site 1173. Shading represents the age-equivalent level of Site 1174
decollement zone.

Sensitivity to Bulk Permeability

Because sediment permeability is the most important factor that controls modeled

pore pressures (Saffer and Bekins, 1998), here I conducted a sensitivity analysis to test its

effects on porosities and excess pore pressures. Bulk permeabilities of hemipelagics and

turbidites were changed separately. Results indicate that simulated k* values were not

very sensitive to an increase in turbidite permeability. For example, when turbidite

permeability was increased 3 orders of magnitude by changing log (k0) from -19.82 to -

16.82, the simulated k* decreased by only 0.05 at Site 808 and by 0.04 at Sites 1174

(Figure 3-6).










a Site 1174 b Site 1174 C Site 808 d ite 808
600 600 700 700 I
+ ++ +

700- 700 800- + 800-
4 0
800- E 8-E E
a 09900

100 900 1000 o1000


91100 *1200 + 1200

2 0.3 0.4 0.5 0 0.2 0.4 0.6 0.2 0.3 0.4 0.5 0 0.2 0.4 0.6
Porosity ,* Porosity
+ Observed Porosities -- log(ko)=-20.82
-- Base Run (log(ko) =-19.82) x--log (ko) =-20.15
-- Turbidite log(ko) =-16.82 X* based on Saffer (2003)
-- k-n relationship from
Saffer and Bekins (1998)

Figure 3-6. Simulated porosity and k* profiles with varying bulk permeability at Sites
1174 and 808. A & C) Comparison of observed porosity to simulated porosity
at Sites 1174 and 808. B & D) Simulated '* profiles at Sites 1174 and 808.
Shading represents the decollement zone. The. values estimated by Saffer
(2003) were based on LWD (Site 1174) and Shipboard data (Site 808).
Vertical lines represents estimate by Screaton et al. (2002).

When the bulk permeability of both turbidites and hemipelagics was lowered by an

order of magnitude (from log (ko) =-19.82 to -20.82) the pore pressures below the

decollement significantly increased compared to the base run (Table 3-4, Figure 3-6).

The simulated k* values reached above 0.45 at both Sites 808 and 1174 and the simulated

porosities above and below the decollement were overestimated compared to the

observed porosity values (Figures 3-6). As bulk permeability was lowered, the steepness

of the slope of the profile gradually increased (Figure 3-6).

The profile for log (ko) = -20.15 represents the relationship shown in Figure 3-4

for the lower range of permeability values. The profile of this simulation shows higher

values for k* compared to the results obtained for the base run. The estimated porosities









below the decollement at both sites were less than observed values. The simulated

porosities above the decollement were higher at Site 1174 than observed, while at Site

808 they reasonably matched the observed porosities.

I also examined the sensitivity of porosity and X* to a permeability relationship (k=-

20+5.25n) in the range used by Saffer and Bekins (1998) (Table 3-4, Figure 3-6). The

simulated X* values using this relationship predicts greater X* values than those predicted

from the base run suggesting that simulated X* values are very sensitive to even slight

changes in the log-linear permeability-porosity relationship (Eq. 2). The predicted X*

using this relationship were greater by 0.06 (Site 1174) and 0.11 (Site 808) compared to

the X* values predicted from the base run (Table 3-4).

In comparison to the k* profile of Saffer (2003), the values of k* predicted for log

(ko) = -20.82 is in the same range as the k* values predicted by Saffer (2003). The

general trend of the k* profile matches values predicted by Saffer (2003) best in the upper

(Site 808) and mid (Site 1174) portions of the underthrust and with depth the k* profile

gradually deviates to lower values than predicted by Saffer (2003).

Sensitivity to Lateral Stress

In reality, the prism sediments are structurally deformed by lateral stresses caused

by tectonic compression. To examine the effects of tectonic compression on excess pore

pressures and porosities, I included lateral stress within the prism sediments as an

additional fraction of the vertical loading (Jaeger and Cook, 1969). Following Domenico

and Schwartz (1998) the ratio of horizontal to vertical stress becomes greater than one in

areas of tectonic compression. I varied the lateral stress factor from 0.1 to 0.9 (i.e., the










total stress at any point in the prism was assumed to be 1.1 to 1.9 times the vertical stress,

Table 3-4).

a te1174 b Site 1174 Site 808 d Site 808
600 600 700 700o
*l
700 700 800 800-











+ Observed Porosities ---* Lateral Stress Factor =0.5
D D+
8 117 8008 80g. S900




1000 the decollement zone. The1000 110 values estimated by Saffer (2003) 1100

9100 1100 1200 +1200
u2 0,3 04 05 0-6 0.2 0.4 0.6 0.2 03 04 0.5 0.6 0 02 0.4 06
Porosity on LWD (Site 1174) and Shi Porositya (Site 808). Vertical lines
reprise Obsents ed Porosities -estimateLateral Stress Factor 0.5
Base Run X* based on Saffer (2003)
Whenlog(ko)20.15 and lateral stress facto.was added to the prism sediments, the porosities smoothly


Figure 3--7. Simulated porosity and ki profiles with added lateral stress at Sites 1174 and
808. A & C) Comparison of observed porosity to simulated porosity at Sites
1174 and 808. B & D) Simulated k* profiles at Sites 1174 and 808. Shading
represents the ddcollement zone. The maximum values estimated by Saffer (2003)
were based on LWD (Site 1174) and Shipboard data (Site 808). Vertical lines
represents estimate by Screaton et al. (2002).

When lateral stress was added to the prism sediments, the porosities smoothly

decreased with depth above the d~collement, and abruptly increased to a maximum at the

d~collement (Figure 3-7). Below the decollement, the porosities gradually decreased

with depth. The sharp change in porosity at the d~collement is due to the added lateral

stress increasing effective stress and lowering porosities within the prism sediments. The

simulated k* values with added lateral stress show a gradual increase of pore pressures

above the d~collement while reaching the maximum k* within the d~collement (Figure 3-

7). Below the d~collement, k* values gradually decreased with depth.

Even though lateral stress is added only in the prism, the k values beneath the

prism increased because the vertical fluid flow from the underthrust is restricted by the









elevated pressures within the prism. Thus, when the lateral stress was increased, both k*

and porosity values were slightly increased below the prism (Table 3-4). However, the

effect of prism lateral stress on pore pressures in the underthrust is relatively small

compared to the sensitivity to bulk permeability. Although it may be possible to have

greater values of the lateral stress factor than 0.9, simulations suggest that they would

have little additional affect on pore pressures in the underthrust sediments.

Increasing the lateral stress factor results in a greater offset in porosity across the

decollement zone. For example, the simulated porosity profile with a lateral stress factor

of 0.5 underestimated observed porosity values both above and below the decollement

(Figure 3-7). Raising the lateral stress factor increases porosities in the underthrust

sediments (Table 3-4), yielding a better match to observed, but decreases porosities in the

prism, yielding a poorer match. Thus, by itself, changing lateral stress (with the porosity-

permeability relationship obtained in this study) cannot generate porosities that matched

observed porosities.

When lateral stress was combined with lower bulk permeability, the pore pressures

were significantly increased with respect to lateral stress alone. For example, when the

bulk permeability of the hemipelagic sediments were lowered to a log (ko) of -20.15 (the

lower boundary to permeability data in Figure 3-4) with a lateral stress factor of 0.3, the

'* at 15 m below the decollement at Site 808 reached 0.32. The simulated porosity

profile for this combination well matched the observed porosity profile above the

decollement while the porosities below the decollement were very slightly

underestimated at both Sites 808 and 1174 (Figure 3-7).









Sensitivity to a Low-Permeability Barrier

Previous accretionary complex studies have suggested the idea of a low-

permeability barrier at the decollement as a possible cause of excess pore pressures. For

example, at the Barbados accretionary prism, Bekins et al. (1995) simulated a 15-m thick

low-permeability cap above the top of the decollement based on an anomalously low

permeability value (6.5 x 10 21 m2) measured by Taylor and Leonard (1990). In a more

recent study based on inferred porosity variations obtained from inverted seismic

reflection data at Nankai, Bangs and Gulick (2005) argued that consolidation of the

uppermost lower Shikoku Basin strata forms a barrier blocking the fluid flow from

below. Because the barrier lies just above the projected level of the decollement, they

suggest that higher-porosity, underconsolidated, and overpressured sediment below forms

a surface of potential decollement propagation.

To test the effects of a low-permeability barrier at the decollement at Nankai, I ran

simulations with a 10-m thick low-permeability barrier above the top of the decollement

zone. This barrier was implemented after the initial sedimentation steps and prior to the

loading by the prism (Table 3-4). As expected, when the low-permeability was added,

the k* values below the barrier abruptly increased to a maximum because the fluids in the

underthrust could not migrate through the barrier fast enough to keep pace with loading.

The simulated k* values at both Sites 1174 and 808 gradually decrease with depth below

the decollement (Figure 3-8).










a Site 1174 b Site 1174
600 600

700 700
.0 0
800 800
5 I-" !1'? ',
a 900 a 900

1000 1000-

1100 i I1100
0.2 0.3 0.4 0.5 0.6 0 0.2 0.4 0
Porosity X*
+ Observed Porosities
-- Base Run


C t Site 808 d ite 808
700 8 700

800 800-

E 900 JS 1 3 900

O 1000 I0 1000

1100 1100 ,

1200 1200
.6 0.2 0.3 0.4 0.5 0.6 0 0.2 0.4 0.6
Porosity V*
Low-permeability Barrier (k=1021m2)
h* based on Saffer (2003)


Figure 3-8. Simulated porosity and X* profiles with added low-permeability barrier at
Sites 1174 and 808. A & C) Comparison of observed porosity to simulated
porosity at Sites 1174 and 808. B & D) Simulated k* profiles at Sites 1174
and 808. Shading represents the decollement zone. The k* values estimated
by Saffer (2003) were based on LWD (Site 1174) and Shipboard data (Site
808). Vertical lines represents estimate by Screaton et al. (2002).

The simulated porosities smoothly decrease both above and below the decollement

while reaching a maximum at the decollement (Figure 3-8). The porosity profiles

generated by the low-permeability barrier with a permeability of 1 x 10 21 m2 were

slightly higher below the decollement at both Sites 808 and 1174 than the observed

porosities. Above the decollement, the porosities closely matched the observed values at

Site 1174 while slightly underestimating the observed values at Site 808. The simulated

X* values at both sites were similar to the values presented by Screaton et al. (2002). In

comparison to the k* profile of Saffer (2003), the values of k* predicted for the low-

permeability barrier is roughly in the same range as those predicted by Saffer (2003).

However, the general trend of the k* profile of the low-permeability barrier is different

from that of Saffer (2003). The low-permeability barrier predicts a sharp decrease in









k* with increasing depth while the profile of Saffer (2003) predicts a gradual decrease of

k* with increasing depth.

Although modeling results indicate high k* values at and below the decollement

due to the added low-permeability barrier, core samples immediately above the

decollement are not available for laboratory-measurements to assess the presence of a

low-permeability barrier. Although I do not have any independent evidence suggesting

that the permeability is low in the direction perpendicular to the plane of decollement,

previous studies have suggested the formation of a low-permeability barrier during the

accretionary process due to shearing of sediments below the wedge as it moves above the

underthrust sediments. For example, Maltman et al., (1992) suggested that at the Nankai

accretionary complex, high stresses at the depth of the decollement could seal fractures

lowering its permeability despite the presence of highly brecciated and fractured material.

This idea is further strengthened by lack of evidence for fluid flow such as veins,

mineralized surfaces or plastic dykes at the decollement (Maltman et al., 1992). If a low-

permeability barrier does exist, then it is more likely to form at the base of the wedge

where the high shearing is present increasing the pore pressures below the barrier. The

presence of a low-permeability barrier is likely to generate excess pore pressures below

the barrier, as shown in Figure 3-8.

The affects of lateral fluid flow on pore pressures are difficult to predict. Lateral

fluid flow could allow pore pressures to be lowered, if fluids can escape to the seafloor

near the toe of the prism. However, if the decollement transmits fluids to near the toe of

the prism, and these fluids cannot escape, pore pressures will be increased at the updip

end of the lateral flow path. Lateral fluid flow within the decollement could decrease the









significance of the low-permeability barrier, if fluid could escape along the decollement

zone. Similarly, lateral fluid escape along the decollement may decrease the impact of

lateral stresses within the prism on k* values within the underthrust sediments. Because

the decollement has been speculated as a possible, yet controversial, pathway for lateral

flow it will be valuable to address in future investigations.

Implications

Both simulations using the added lateral stress and low-permeability barrier show a

sharp increase in k* near the decollement zone at both Sites 1174 and 808 (Figure 3-7

and Figure 3-8). However, each scenario generated a distinct k* profile. The k* profile

generated by the added lateral stress demonstrate a gradual increase in pore pressures

both above and below the decollement, and the k* value peak at the decollement. The

maximum k* corresponds to the maximum simulated porosity found near the top of the

decollement zone (Figure 3-7). In contrast, the k* profile generated by the low-

permeability barrier shows an abrupt increase of pore pressures at the lower part of the

decollement. Above the decollement, the k* profile shows a very slight decrease with

depth while below the decollement k* decreases more gradually. As observed in both

scenarios, the maximum k* is located near the decollement, consistent with previous

inferences (e.g., Hubbert and Rubey, 1959) that pore fluid pressures play a major role in

the mechanics of thrust faulting.

The excess pore pressures observed near the decollement can also be related to the

sliding of the decollement as proposed by the critical wedge theory (Davis et al., 1983).

Critical wedge theory predicts that pore pressures significantly greater than hydrostatic

are needed to maintain small taper angles such as at the Muroto Transect of Nankai.









Generally a small critical taper is an indication of very little basal friction, due either to

low intrinsic strength or elevated pore pressures, or that the wedge consists of a strong

material, which need not deform for stable sliding to occur (Davis et al., 1983). Thus, it

can be speculated the maximum k* values observed in the profiles at the decollement

could represent the level of least mechanical resistant that promotes stable sliding. It is

possible that a permeability barrier would also affect fault zone initiation, if elevated pore

pressures are transmitted seaward of the deformation front. A two-dimensional model

would be needed to fully examine this possibility.

Many previous studies based on the critical taper theory have assumed the pore

pressure ratios (pore fluid pressure/lithostatic pressure) at the base of the wedge and

within the wedge to be equal. The results from simulations with the low-permeability

barrier demonstrate a mechanism for basal pore pressure ratios to be significantly greater

than those in the wedge which could increase taper angle for a given basal pore pressure.

Lateral stresses in the prism may also result in basal pore pressure ratios to be higher than

in the wedge, but the difference is not as great as with the low-permeability barrier.

Conclusions

I examined the effects of different parameters on excess pore pressures at the toe of

the Nankai accretionary complex. Using a permeability-porosity relationship based on a

best fit to laboratory data, simulations suggest that sedimentation and prism thickening

generate excess pore pressures, but not as high as predicted from previous studies.

Results from this study demonstrated significant increase in pore pressures at the

decollement with lower bulk permeability, such as obtained by using the lower boundary

of permeability-porosity data. Because the lowest permeabilities generally control









vertical flow, this relationship may be more appropriate for the simulations than the best

fit equation. However, if the high excess pore pressures suggested by Screaton et al.

(2002) or Saffer (2003) are correct, permeabilities must be even lower, requiring either a

bulk permeability represented by log (ko) = -20.82, a low permeability (<10-20 m2) barrier

above the decollement, or a combination of lower bulk permeability and permeability

barrier. Alternatively, other factors must contribute, such as recent prism growth rates

greater than the time-averaged rates simulated here.

The magnitude of pore pressures in the underthrust sediments demonstrated only

slight sensitivity to added lateral stresses in the prism, although the profile of the pore

pressure ratio is affected. Results further illustrated that the simulations with low-

permeability barrier and lateral stress both produced a sharp increase in porosities below

the decollement zone, as is observed in the measured values. Furthermore, in both

scenarios, maximum excess pore pressure ratios were found at the decollement, which

could contribute to stable sliding of the decollement zone.














CHAPTER 4
EVOLUTION OF HIGH PORE PRESSURES AND IMPLICATIONS FOR EPISODIC
FLUID FLOW AT THE NORTHERN BARBADOS ACCRETIONARY COMPLEX

Introduction

Variations in fluid flow pressures and its distribution within subduction zones

regulate the mode of deformation, affecting the evolution of the accretionary complex. It

has been speculated that excess pore pressures play a major role in the mechanics of

thrust faulting (Hubbert and Rubey, 1959), thus, allowing the weak semilithified

sediments in accretionary complexes to glide over the subducting plate along a low-angle

detachment surface (Davis et al., 1983). Excess pore fluid pressures are also responsible

for increasing sediment permeability associated with reduction in effective stress (Yeung

et al., 1993; Fisher and Zwart, 1996). In addition, pore pressure has been claimed to

influence seismogenic faulting, through its control on effective stress and consolidation

state (e.g., Moore and Saffer, 2001; Scholz, 1998). Thus, an understanding of the

development of excess pore pressures will provide valuable insight to the evolution of

accretionary complexes as well as to subduction zone processes such as fault mechanics.

The abundant evidence for excess pore pressures at Barbados accretionary complex

provides an excellent opportunity for the study of evolution of pore pressure generation.

The distribution of mud volcanoes (Gretener, 1976) and the overall shape of the

accretionary prism (Davis et al, 1983) suggest the existence of excess pore pressures at

the Barbados accretionary complex. In addition, elevated pore fluid pressures have been

inferred at the Barbados accretionary complex from fluid flow modeling (e.g., Shi and









Wang, 1988; Screaton et al., 1997; Bekins et al., 1995; Henry, 2000) and direct borehole

measurements along the decollement (Foucher et al., 1997; Becker et al., 1997).

Studies have speculated that fault zones play a major role in focusing fluid

expulsion (e.g., Bekins et al., 1995; Moore et al., 1998; Henry, 2000). Indirect evidence

for transient fluid flow along the decollement is provided by geochemical and thermal

anomalies (e.g., Bekins et al., 1995; Fisher and Hounslow, 1990). A widely held

hypothesis relating elevated pore pressures and evidence for fluid flow is that flow must

be episodic, although the mechanisms producing the episodic fluid flow events are not

fully understood. Field evidence for episodic fluid flow includes the presence of multiple

episodes of fracturing and vein filling in accreted sediments (Labaume et al., 1997).

However, the time scales of these events are difficult to estimate (Knipe et al., 1991).

Several mechanisms have been put forward to explain the time-variable

permeability in the fault zone. One mechanism is that the permeability in the

decollement is enhanced by the episodic opening of horizontal hydrofractures when pore

pressures reach values above lithostatic (e.g., Behrmann, 1991; Moore and Vrolijk, 1992;

Brown et al., 1994). Similarly, numerical modeling by Bekins et al. (1995) predicted

results with observed chloride anomalies within the decollement of the Barbados

accretionary complex by assuming an instantaneous permeability increase within the

decollement. Another mechanism for episodic fluid flow is based on in situ bulk

permeability measurements that were made at a variety of fluid pressure conditions.

According to Fisher and Zwart (1996), this relationship between bulk permeability and

effective stress may explain the dynamics of fluid-fault interactions and the transient

nature of hydrologic processes at convergent margins. Additional hydrologic tests that









were conducted at a sealed borehole penetrating the decollement at the Barbados

accretionary complex support the conclusion by Fisher and Zwart (1996) that significant

permeability increase can occur within the decollement at pore pressures below lithostatic

pressures (Screaton et al., 1997). Studies based on fluid budgets shows that fluid flux

varies with both arcward distance and through time (Saffer and Bekins, 1999). In

addition, fluid budget studies suggest that initiation of connected flow conduits is delayed

with respect to the time of accretion and may be related to burial below a critical depth,

where channelized fluid escape is more efficient than diffuse flow to the sea floor or

where sediments may behave brittlely (Saffer and Bekins, 1999). Even though previous

modeling studies have investigated the production of overpressures, these models did not

indicate when the overpressures are generated in the evolution of the complex and thus,

the connectivity between excess pore pressures and episodic fluid flow is not well

understood. Models used by Henry and Wang (1991), Shi and Wang (1994), and

Stauffer and Bekins (2001) focused on processes that take place at the toe of the prism

during initial offscraping. Bekins et al. (1995) focused on steady-state pore pressures and

transient fluid flow assuming instantaneous decollement permeability after the entire

prism had grown. However, to fully understand the development of pore pressures and

thus, hypothesized episodic fluid flow, one should examine the development of pore

pressures both at the toe and deeper parts of the accretionary complex through both space

and time. Thus, in this study I modeled 50 km of the accretionary complex as a time-

dependent evolving prism. A combined prism growth and fluid flow model was used to

examine the development of pore pressures. Mechanisms for episodic fluid flow were

examined during the evolution of the accretionary complex by including hydrofracture or









a decollement with varying permeability based on a relationship of bulk permeability -

vertical effective stress.

Background

The Barbados accretionary complex is located in the Caribbean where the

Atlantic Plate (Figure 4-1) is being subducted beneath the Caribbean Plate at a rate of 2

cm/yr in an east-west direction (DeMets et al., 1990). At Barbados, active accretion of

sediments takes place at the eastward end of the complex as the more stabilized portion

lies westward, where the complex is partially exposed above sea level at the Barbados

Island (Figure 4-1). The complex varies in thickness from 200 km south of Tiburon Rise

at 140N, to approximately 10 km north of Tiburon Rise at 160N (Bangs and Westbrook,

1991). The variation in thickness of the complex is related to the distance from the

terrigenous sediment source from South America (Underwood and Deng, 1997), as well

as the affect of local barriers such as the Tiburon Rise (Figure 4-1), which slows the

influx to the complex.

The northern Barbados accretionary complex is rich in hemipelagic sediments

(Bekins et al., 1995), whereas the southern part is dominated by turbidites (Langseth et

al., 1990). The age of the sediments at the Barbados accretionary complex varies from

Late Eocene to Late Cretaceous (Underwood and Deng, 1997). Mud volcanoes and mud

diapirs, which indicate excess pore pressures, are abundant in the southern part of the

complex where the sediment sequence is thicker (Moore et al., 1982).

Based on seismic reflection images the decollement was estimated to be -14 m

thick at the northern Barbados accretionary complex (Shipley et al., 1994). It was

inferred as a high-porosity zone with undercompacted sediments and high-fluid pressures











(Moore et al., 1995). During Deep Sea Drilling Project (DSDP), Leg 78A high pore

pressures were encountered at the toe of the northern Barbados accretionary complex

while attempting to drill through the decollement (Biju-Duval et al., 1984). The seismic

profiles show prism sediments being subjected to high lateral strain while sediments in

the underthrust remain undeformed (e.g., Westbrook et al., 1988), possibly due to the

presence of excess pore pressures on the decollement.


A p
62 61 600 5 58W C b..n PIm


*1- ?r t


L-'!. \ ,i_ .... 16,
.. DSDP/ODP drill sites
992 3-D survey area




4 1'4-0- i
I I -., .

r t 19 i. I, l -. .' ,
I3 N


B
Depth











Figure 4-1. Location map and cross-section of the Barbados accretionary complex. A)
45. -45
t 1046 O Sio 1 1,iT S1 0ow,






15km E

Figure 4-1. Location map and cross-section of the Barbados accretionary complex. A)
Map of the eastern Caribbean showing the deformation front. B) Cross-
section of the Barbados accretionary complex (Shipboard Scientific Party,
1998).









Ocean Drilling Program (ODP) Leg 110 revisited northern Barbados and three sites

were drilled (Sites 671, 672 and 676). Sites 671 and 676 (Figure 4-1) were drilled

arcward of the deformation front at a distance of 4.5 km and 0.25 km respectively

(Mascle et al., 1988). Site 672 was drilled 6 km east of the deformation front to provide

an undeformed reference site (Mascle et al., 1988). The core samples from Leg 110 were

analyzed for structural features, chemical signals, permeability, consolidation behavior,

and bulk composition. Taylor and Leonard (1990) inferred near-lithostatic pore pressures

directly above the proto-decollement zone based on consolidation tests on core samples.

In contrast several other studies (e.g., Shi and Wang, 1985; Screaton and Ge, 2000)

concluded that sedimentation rates alone were not sufficient to produce excess pore

pressures.

During Leg 156 Logging While Drilling (LWD) was performed at Site 948, which

coincides with the location of Site 671 (Shipley et al., 1997). Site 949 was drilled 2 km

northeast of Site 948 (Shipley et al., 1997). During Leg 156, a hydrologic borehole

packer was successfully deployed and fluid flow experiments were conducted at Sites

948 (671) and Site 949 (Leg 171A, Site1046) (Fisher and Zwart, 1996). Results of

packer experiments further supports the presence of excess pore pressures at the

decollement, although perturbations from drilling and testing were difficult to separate

from natural pore pressures. Estimated values for fault zone permeability from in situ

packer tests vary from 10-12 m2 when fluid pressure is at lithostatic to 10-18 m2 when fluid

pressure is at hydrostatic (Fisher and Zwart, 1996). Results ofhydrogeologic tests that

were performed at a sealed borehole (949C) penetrating the decollement support the idea









of significant permeability increase within the decollement with increasing pore pressure

(Screaton et al., 1997).

Fluid Flow and Pore Pressures

Evidence for fluid flow at the Barbados accretionary complex comes from the

presence of low-chloride anomalies observed along the decollement (Kastner et al.,

1993). It has been suggested by Bekins et al. (1995) that smectite dehydration is the most

likely mechanism for low-chloride anomalies. In contrast, Fitts and Brown (1999)

suggested that low-chloride anomalies occur as a result of artificial squeezing of

sediments released from smectite interlayers during pore water sampling. However, even

after accounting for the effects of sample squeezing process, the low-chloride anomaly is

still 12% fresher than seawater (Fitts and Brown, 1999) supporting the clay dehydration

as a possible explanation for low-chloride anomalies. The clay dehydration reaction

takes place at temperatures between 60-160 C (Perry and Hower, 1970). Based on

kinetic modeling of clay dehydration in the Barbados accretionary complex, Bekins et al.

(1995) estimated the peak reaction window to be at 50 km arcward of the deformation

front. Thus, if the fluid released from smectite dehydration is responsible for the

observed low-chloride anomalies at the toe of the prism (Site 948), then pore fluids must

migrate over 46 km to reach Site 948 (Bekins et al., 1994).

Heat flow anomalies have also been observed at the Barbados accretionary

complex from temperature measurements (Fisher and Hounslow, 1990) and surface heat

flow surveys (Foucher et al., 1990). Seafloor heat flow values near the toe of the

complex range from 96 to 192 mW m-2 (Fisher and Hounslow, 1990) and are well above

the 53-55 mW m-2 expected for 90 Ma oceanic lithosphere (Ferguson et al., 1993). One

of the possible explanations for the observed heat flow anomalies is fluid flow along the









decollement (Langseth et al., 1990). In contrast, Fisher and Hounslow (1990) and Moore

et al. (1998) suggest possible lateral fluid flow along turbidites in the underthrust

sequence. Muller and Smith (1993) argued that the uniformly high background heat flow

in the ODP drilling area could be due to crustal thinning of the Tiburon Rise. Based on

both steady state and transient state models, fluid fluxes needed to explain the observed

heat flow anomaly are approximately one order of magnitude higher than the fluxes

needed to explain the low-chlorinity anomaly (Henry, 2000). Similarly several other

studies (Foucher et al., 1990; Saffer and Bekins 1999; Cutillo et al., 2003; Bekins and

Screaton, 2006) have noted that the outflow from the underthrust necessary to create a

thermal anomaly (greater than 60 mW m-2) seems excessive compare to the in flow thus,

suggesting that the hypothesis on crustal thinning under Tiburon Rise should be further

explored.

In addition to low-chloride and heat flow anomalies, the presence of mineralized

veins supports the idea of transient fluid flow along the decollement. Mineral veins were

found in the upper part of the decollement (Labaume et al., 1997). The orientations of

the veins suggest that pore fluids around the fault zones is at near lithostatic pressure

during vein formation (Labaume et al., 1997). Veins that were formed during several

growth phases reflect the episodicity of fluid flow along the decollement (Labaume et al.,

1997).

Bekins et al. (1995) used low chloride anomalies as constraints on a transient

model. In their model they raised permeabilities from 10-12 -10-15 m2 along the entire

decollement zone to match observed low-chloride anomalies. To justify this, they

hypothesized that pressures in the accretionary complex build until it reaches values that









are close to lithostatic. These high pore pressures lower the effective stress along the

decollement creating fractures or dilating existing fractures, raising the permeability

along the decollement. To simulate this hypothesized scenario based on a steady -state

model, they assigned varying values of decollement permeability (kd) until sublithostatic

pressures were simulated. Once the appropriate kd value was determined they assumed

that the pressures estimated from this solution represent those immediately before a slip

event that increases kd and used these pressures as initial conditions in the transient

simulation. When decollement permeability was suddenly raised the pore pressures

along the decollement reached values that are closer to lithostatic during the first few

thousand years supporting the concept of pressure build-up and release cycles or episodic

fluid flow.

In another study, Henry (2000) modeled the episodicity of fluid flow in a slightly

different manner. In his model he started a pressure pulse at the arcward boundary of the

complex and allowed it to propagate along the decollement. The decollement was

assigned a permeability that increased continuously as pore pressure increased following

the relationship of Fisher and Zwart (1996). Also it is assumed that all fluid flow occurs

either along the decollement or along sandy layers at 200 m below the decollement within

the underthrust sequence. Thus, a bulk permeability of 3 x 10-13 m2 is assigned either to

the decollement or to the sand layer to simulate fluid flow. According to this simulation,

significant fluid flow persists for several thousands to 10,000 years and it is in agreement

with a diffusion-advection model of the chlorinity anomaly (Henry, 2000).

Based on long term monitoring at Holes 948D and 949C during ODP Leg 156, the

estimated k* values (where k* = (P-Ph)/(PI-Ph), P = pore pressure, Ph = hydrostatic









pressure and P = lithostatic pressure) were 0.50 (Foucher, et al., 1997) and 0.36 (Becker

et al, 1997), respectively. These results were similar to those previously obtained using a

steady state model with a permeability-depth relation for clay-rich sediments (Bekins et

al; 1995). Moreover, the increase of ~* by 0.15 between Sites 949 and 948 over a lateral

distance of 2.2 km was also estimated by Stauffer and Bekins (2001) based on inferred

consolidation state.

In addition, constraints for pore pressure distribution also follow from an analysis

of the mechanical force balance in accretionary wedges presented by Davis et al. (1983).

According to Davis et al. (1983), in order for the sediments of the wedge to move over

the underthrust sequence along a low-angle decollement, high pore pressures (k = 0.92

for the overall taper, where X is the ratio of pore fluid pressures to the vertical normal

traction exerted by the lithostatic overburden) must be present along the decollement.

The presence of lower pressures would result in a steeper taper angle than that observed

at Barbados (Bekins et al., 1995).

Hydrofractures

Behrmann (1991) suggested that hydrofracturing enhances permeability in

argillaceous rock sequences. According to Behrmann (1991) the capability of rocks to

hydrofracture depends on the mode of faulting and the effective mean stress. He also

noted that the depth to which hydrofracture could occur is a function of both the faulting

mode and the ratio of fluid and lithostatic pressures. Thus, wrench and normal faults

hydrofracture even at ratios of fluid and lithostatic pressure is less than one. In contrast

thrust faults always require ratio of fluid and lithostatic pressure to be greater than one to

hydrofracture. Vertical fluid flow has been indicated by 3 -D seismic images at the









northern Barbados accretionary complex which shows at least 50 m offset in the

turbidites along the normal faults extending upward from the basement (Zhao and Moore,

1998) suggesting that vertical hydrofracture could occur in the underthrust. According to

Behrmann (1991), the criteria requires X* to be > 0.8 for vertical hydrofracturing in soft

rocks. Vertical hydrofractures occur perpendicular to the least principle stress axis.

According to Price (1975), when sediments hydrofracture, the hydraulic properties are

expected to change dramatically.

Modeling Methods

Model Implementation

A model developed by Screaton and Ge (2000) was modified to simulate the effects

of subduction beneath a prism. This model builds the accretionary complex in segments

through time. The prism growth/flow model consists of two sub programs. The first sub

program is a modified loading program (Gamage and Screaton, 2006), which builds the

initial sediment column that enters the accretionary complex at the deformation front.

The second sub program (prism loading) advances the accretionary complex over the

subducting sediments in segments through time at a convergence rate of 2 cm/yr. It is

assumed that the taper of the prism is constant. The prism-growth/flow model calculates

the time necessary for prism to advance one column by dividing the horizontal dimension

of the column by the convergence rate. The vertical dimension is calculated using the

prism thickening rates. As the prism advances seaward the loading program adds

sediments on each advanced column according to the assigned prism thickening rates.

Based on the calculated increase in overburden and sediment properties, the prism-

growth/flow model calculates the pore fluid pressures. These pore pressures are then

input into SUTRA (Voss, 1984), a finite-element code that simulates transient two-









dimensional fluid and thermal transport. SUTRA simulates the fluid flow and the pore

pressure changes with time. Once the pore pressures were calculated in SUTRA they

were transferred back into the prism-growth/flow model, which calculates effective

stress, porosity, and permeability before beginning the next loading step.

During hydrofracture, the model checks for regions that have reached the criteria

for vertical hydrofracture after every seaward advancement of the prism. If pressures

meet the vertical hydrofracture criteria, then the model increases vertical permeability

from the point of lithostatic pressure up to the decollement. If pressures are less than the

assigned criteria for hydrofracture the permeability values are assigned according to

permeability-porosity relationship.

SUTRA uses a backwards finite-difference scheme which enhances numerical

stability. The large permeability contrast during the introduction of hydrofractures into

the model was challenging for the iterative solver, especially for the thermal transport

simulations. However, by reducing the contrast between the highest and the lowest

permeabilities and increasing upstream weighting for the transport, convergence was

obtained.

Model Equations

Combining the mass conservation of fluid with Darcy's law, the following equation

can be written for two-dimensional transient flow:

K h K a'h Sh
K X X K2 a Q2 --d (1)


where Ss is specific storage [L-1], h is hydraulic head [L], x and y are spatial

coordinates, Q is a source term reflecting processes such as loading that affect fluid

volume or pressure [T-1]. The left hand side term accounts for fluid flow into and out of a