|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004605
Variable-density flow models of saltwater intrusion in coastal landforms in response to climate change induced sea level rise and a chapter on time-frequency analysis of ground penetrating radar signals
h [electronic resource] /
by Swagata Guha.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains X pages.
Dissertation (Ph.D.)--University of South Florida, 2010.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
ABSTRACT: Populations residing on and near the world's coasts have become increasingly dependent on coastal groundwater for their supply of freshwater. Under the conditions of predicted climate changes, the expected rise in global sea level can adversely affect the quality and quantity of freshwater resources in coastal areas as a result of saltwater intrusion. In this study, a suite of two- and three-dimensional variable-density groundwater flow models of major coastal landforms (e.g. deltas, estuaries and small islands) has been constructed to assess the effects of sea level rise (SLR), using different SLR rates of 0.5 m, 1m and 1.5 m over the next 90 years, from 2010-2100. The model results indicate that in natural coastal systems the extent of saltwater intrusion is significantly controlled by the stratigraphy of the depositional environments. Among deltaic aquifers, wave-dominated deltas are more prone to saltwater intrusion than river- and tide-dominated deltas. In case of a partially mixed, microtidal estuary, SLR can cause extensive porewater salinity increases, especially within estuarine sand deposits. Simulations of atoll and barrier islands reveal that carbonate atoll islands with high conductivity units, are severely affected by SLR, resulting in significant reduction of the volume of freshwater lens. In contrast, migrating sandy barrier islands could retain their freshwater resources with rising sea level under conditions of increased recharge, assuming the barriers can migrate in response to SLR. The freshwater lens of barrier island aquifers would reduce in size due to increased evapotranspiration caused by change in vegetation pattern. When examined for anthropogenic impacts of groundwater withdrawal through pumping, all the coastal aquifers show evidence of saltwater intrusion, with varying degrees of impact. Wave-dominated deltas are more affected by groundwater withdrawal than river- and tide-dominated deltaic aquifers. Saltwater intrusion in atoll islands is further enhanced by pumping withdrawal. It is evident from the results of the simulations that, the potential effects on coastal aquifers of groundwater withdrawals for potable water can easily exceed the adverse effects of SLR in terms of salinity increase.
Advisor: Mark Stewart, Ph.D.
t USF Electronic Theses and Dissertations.
Variable-density Flow Models of Saltwater Intrusion in Coastal Landforms in Response to Climate Change induced Sea Level Rise a nd a Chapter on Time-Frequency Analysis of Ground Penetrating Radar Signals by Swagata Guha A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Geology College of Arts and Sciences University of South Florida Major Professor; Mark T. Stewart, PhD H. L. Vacher, PhD M. Rains, PhD P. Wang, PhD K. Grote, PhD Date of Approval June 10, 2010 Keywords: Groundwater Modeling, Coastal Aquifer, SE AWAT, Near-surface Geophysics, Thin Beds Copyright @ 2010, Swagata Guha
ACKNOWLEDGEMENTS I would like to thank the Department of Geology, U niversity of South Florida, for supporting this research. I am indebt to my thesis committee members, Dr. Mark Stewart, Dr. Henry L. Vacher, Dr. Ping Wang, Dr. Mark Rains and Dr. Katherine Grote for their advice and suppor t. I would also like to thank Dr. Jim Rumbaugh at Environmental Simulations Inc. for providing a professional version of Groundwater Vistas that was used for the groundwater modeling. During the early years of my PhD program that were dedicated to geophysical studies, I received valuable assistance from several USF graduate and undergraduate students in my experimental ventu res with the Ground Penetrating Radar. My sincere gratitude goes to all those who helped me and to the funding support I received from the Geological Survey of America Student Research Grant and the Gulf Coast Association of Ge ological Societies Student Grant.
i TABLE OF CONTENTS LIST OF TABLES v LIST OF FIGURES vii ABSTRACT xv CHAPTER 1: INTRODUCTION 1 Climate change and sea level rise 1 Coastal aquifers and saltwater intrusion 4 Saltwater intrusion: a global concern 5 The United States 6 Europe 7 Africa 8 Asia 9 South America 10 Australia 10 Small islands of the Pacific 11 Objective of present study 11 CHAPTER 2: METHOD 19 Variable-density saltwater intrusion 19 Early works on saltwater intrusion 20 Numerical Modeling 21 Modeling saltwater intrusion with SEAWAT 22 CHAPTER 3: IMPACT OF SEA LEVEL RISE ON DELTAIC AQUI FERS 24 Introduction 24 Characteristics of deltas 26 Groundwater flow models 28 Model 1: River-dominated delta 29 Model description 29 Model design 29 Time discretization 29 Hydrologic units 30 Hydraulic parameters 30
ii Boundary conditions 31 Model simulations 32 Base case scenario 32 Relative sea level rise scenarios 32 Groundwater withdrawal scenario 33 Results and discussion 33 Model 2: Wave-dominated delta 35 Model description 35 Model design 35 Hydrologic units 35 Hydraulic parameters 36 Model simulations 36 Base case scenario 36 Relative sea level rise scenarios 37 Groundwater withdrawal scenario 37 Results and discussion 37 Model 3: Tide-dominated delta 38 Model description 38 Model design 38 Boundary conditions 39 Hydrologic units 39 Hydraulic parameters 39 Model simulations 39 Base case scenario 39 RSLR and groundwater withdrawal scenarios 40 Results and discussion 40 Conclusions 41 CHAPTER 4: POREWATER SALINITY CHANGES IN A PARTIALL Y MIXED, MICROTIDAL ESTUARY DUE TO SEA LEVEL RISE 60 Introduction 60 Characteristics of estuaries 62 Groundwater flow models 63 Model description 63 Model design 64 Boundary conditions 64 Hydrologic units 64 Hydraulic parameters 64 Model assumptions 65 Model simulations 65 Base case scenario 65 Sea level rise scenarios 66 Results and discussion 66 Conclusions 68
iii CHAPTER 5: MODELING SEA LEVEL RISE INDUCED SALTWATE R INTRUSION IN A MIGRATING BARRIER ISLAND 72 Introduction 72 Geology of barrier islands 7 3 Migration 74 Hydrogeology of barrier islands 75 Factors controlling freshwater lens morphology 75 Generic model of a barrier island 76 Introduction 76 Geologic setting 78 Hatteras Island hydrogeology 79 Freshwater-saltwater interface 80 Factors affecting freshwater resources in Hatter as Island 81 Recharge 81 Groundwater withdrawal 82 Surface drainage 82 Tidal and storm effects 83 Groundwater flow models 83 Model description 84 Model design 84 Hydrologic units 84 Time discretization 85 Boundary conditions 85 Hydraulic properties 86 Model assumptions and limitations 86 Model simulations 87 Base case scenario 87 Sea level rise scenarios 87 Additional scenarios 88 Reduced recharge 89 Withdrawal 89 Results and discussion 89 Conclusions 93 CHAPTER 6: VARIABLE-DENSITY FLOW MODELS OF SALWATER INTRUSION IN ATOLL ISLAND AQUIFER 108 Introduction 108 Hydrogeology of atoll islands 1 09 Generic model of atoll island 1 11 Introduction 111 Geologic setting 112 Hydrogeology of Laura, Majuro Atoll 113 Groundwater flow models 114 Model description 115
iv Model design 115 Hydrologic units 115 Boundary conditions 115 Hydraulic parameters 116 Tidal effect simulation 116 Model simulations 117 Base case scenario 117 Sea level rise scenarios 117 Results and discussion 118 Conclusions 121 CHAPTER 7: SUMMARY OF RESULTS 13 4 CHAPTER 8: TIME-FREQUENCY ANALYSIS OF GROUND PENETR ATING RADAR SIGNALS 137 Introduction 137 Synthetic model examples 138 Example 1: Thin beds in coastal deposits 141 Example 2: Contacts bounding layered packages 143 Conclusions 145 LIST OF REFERENCES 151 ABOUT THE AUTHOR END PAGE
v LIST OF TABLES Table 3.1: Hydraulic parameters used for SEAWAT mod el simulation of river-dominated delta. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. 44 4 Table 3.2 : Hydraulic parameters used for SEAWAT mo del simulation of wave-dominated delta. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. 45 Table 3.3: Hydraulic parameters used for SEAWAT mod el simulation of a tide-dominated delta. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. 46 Table 4.1: Hydraulic parameters used for SEAWAT mod el simulation of partially mixed estuary 69 Table 5.1: Hydraulic parameters used for SEAWAT mod el simulation. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. 95 Table 6.1: List of some Pacific Island Countries w ith populated atolls and volcanic islands. Islands with freshwater lens are shown in italics (adapted from White and Falkland, 2010). 12 2
vi Table 6.2: Hydraulic parameters used for SEAWAT mod el simulation. Layer thickness is given by b, Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. 123
vii LIST OF FIGURES Figure 1.1 Ground-water flow patterns and the zone of dispersion in an idealized, homogeneous coastal aquifer ( Cooper, 1959 ). 13 Figure 1.2 a. A) a fully confined coastal aquifer, B) partly leaky confined coastal aquifer formed by low permeability, thinnin g out, layer causing upward discharge of freshwater, C) a leaky coastal aquifer where a thin semipermeable layer creates a confined aquifer that shows no freshwater discharge to the s ea (modified after (Custodio, 2005). 14 Figure 1.2. b. A multilayered coastal aquifer with low permeability layers forming confined aquifers with different positions of mixing zones (modified after Custodio, 2005 15 Figure 1.2 c. Coastal deltaic aquifers. An unconfi ned aquifer (upper figure) with a river that allows freshwater dischar ge to the sea (similar situation in Lower Llobregat river valley and delta, Barcelona, Spain). Holocene seawater intruded withi n the deep aquifer but was washed away by freshwater disc harge from the river. In the lower figure (similar to Low er Ebre river valley and delta, Southern Catalonia, Spain) the ri ver intersects the upper unconfined aquifer, thus no fr eshwater head is present to allow freshwater discharge. Holo cene seawater within the lower deep aquifer slowly flows upstream (modified after Custodio, 2005). 16 Figure 1.2d. Island aquifers. In volcanic islands ( upper figure) the presence of saltwater could be limited to narrow co astal strips, coastal plains in recent volcanoes, alluvium or slo pe deposits. In a low heterogeneous island formed by carbonate a nd marl (middle figure) a central freshwater lens is presen t while saltwater wedges intrude the more permeable layers. In lower figure, a freshwater lens floats on top of saline w ater present within highly permeable lower units (modified after Custodio, 2005). 17
viii Figure 1.2. e. Freshwater-saltwater distribution in a near-shore dune and intermediate low land environment (e.g. as in Nethe rlands, Belgium, Mar del Plata of Argentina). The lowlands may represent evaporation ponds, externally fed freshwa ter lakes, artificially drained areas (as in Netherlands) (mod ified after Custodio, 2005). 18 Figure 3.1 a) Plan view of top layer of model doma in of river-dominated delta showing different hydrologic units and bounda ry conditions. b) shows cross-sectional view of model with hydrologic units, layer thickness (m) vertical dist ribution of hydrologic units and their thicknesses. (X and Y ax es are column and row numbers) 47 Figure 3.2 Comparison of salinity distributions i n a river-dominated delta at a depth of 1m under initial and RSLR conditions. 0.25 kg/m3 isochlor marks the extent of potable water. 4 8 Figure 3.3 Comparison of salinity distributions i n a river-dominated delta at a depth of 7 m under initial and RSLR conditions. 0.25 kg/m3 isochlor marks the extentd of potable water. 49 Figure 3.4 Comparison of salinity distributions in a river-dominated delta with steady-state groundwater withdrawal, shown at different depths for RSLR =1.18 m. 50 Figure 3.5 a) Plan view of model 3 (wave-dominated delta) showing hydrologic units on top layer and boundary conditio ns b) Cross-sectional view of model 3 showing vertica l distribution of hydrologic units and their thicknes ses. (X and Y axes are column and row numbers). 51 Figure 3.6 Comparison of the extent of the freshw ater-saltwater interface in a wave-dominated delta shown by 1kg/m3 isochlors at different depths. a) base case and b) RSLR =0.68 m. 52 Figure 3.7a.Comparison of the extent of the saltwat er intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for RSLR=1.18m 53 Figure 3.7b.Comparison of the extent of the saltwat er intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for RSLR=1.68m 54
ix Figure 3.8 Comparison of the extent of the freshwat er-saltwater interface in a wave-dominated delta under groundwater withdr awal conditions, shown by 1kg/m3 isochlors at different depths. 55 Figure 3.9 a) Plan view of model 3 (tide-dominated delta) showing hydrologic units on top layer and boundary conditio ns b) Cross-sectional view of model 3 showing vertica l distribution of hydrologic units and their thicknes ses(X and Y axes are column and row numbers). The different col ors represent hydraulic conductivity zones associated w ith the hydrologic units 56 Figure 3.10 Comparison of salinity distributions in a tide-dominated delta at a depth of 0.75m. 0.25 kg/m3 (red contour line) isochlor represents the extent of potable water. X and Y axe s are in meters. 57 Figure 3.11 Comparison of salinity distributions in a tide-dominated delta at different depths for RSLR =1.18m under condition s of groundwater withdrawal. 0.25 kg/m3 (red contour lin e) isochlor represents the extent of potable water. X and Y axe s are in meters. 58 Figure 4.1 Plan and cross-sectional view of model with hydrologic units, boundary conditions and layer thicknesses. Colored zones represent different hydraulic conductivity values o f the hydrologic units. 70 Figure 4.2 Salinity changes in model estuary at a depth of 4.5 m (estuary bottom) under different conditions of sea level ris e. X and Y axes represent distance in meters. 71 Figure 5.1. Schematic showing landward migration/re trogration of a barrier island by the process of overwash deposition (adapt ed from Rosati, 2009) 96 Figure 5.2 Location map of Hatteras Island. Geol ogical and hydrogeological data available from previous studie s have been used for modeling purpose in the present study 96 Figure 5.3a Schematic of Hatteras Island (cross-se ction) with hydrologic units as used in model simulations (from Anderson, 2000) 97
x Figure 5.3b. Cross-section of 2D model showing mode l geometry, boundary condition and hydrogeologic units designed for present study based on Andersons model (Figure 5.3 a). 98 Figure 5.4 Salinity distributions from model simu lations. a) Base case scenario (Model1) where freshwater lens extends upt o a depth of ~18 m. The lens shape is slightly asymmetric, in dicating stratigraphic control over lens formation. The tran sition zone (defined by 1kg/m3 and 10kg/m3 isochlors) is wider along the Sound side. The land-sea boundary position is shown by the asterisk. (All figures are vertically exaggerated). 99 Figure 5.5 Salinity distributions from Model 2a (S LR =0.5m in 90 years, island migration rate =5m/yr) shown at different ti mesteps 100 Figure 5.6 Salinity distributions from Model 2b (S LR =1.0 m in 90 years, island migration rate =10 m/yr) shown at different timesteps 101. Figure 5.7 Salinity distributions from Model 2c (S LR =1.5 m in 90 years, island migration rate =15 m/yr) shown at different timesteps 102 Figure 5.8a Changes in water table elevations as th e barrier island migrates with rising sea-level. The head distributi ons generated are due to increased recharge and rise in sea level along with new migrated positions of the island ove r the underlying, low conductivity semi-confining unit. 1 03 Figure 5.8b Comparison of water table elevation cha nges under conditions of increased recharge, constant recharge as the isl and migrates and hypothetical scenario with constant re charge but no migration. All models consider 1m rise in sea-le vel in 90 years. The scenario with increased recharge in a mi grating island (i.e. Model 2b) produces the maximum head va lues in the central part of the island. When recharge is ke pt the same as in initial conditions, the head value drops in dicating the effect of 10% increased recharge. The effect of isl and migration dominates on the head distribution, as se en in the case of a static island with a much lower head valu e. This happens due to the new migrated position of the isl and with the semi-confining unit underlying the central part The low conductivity of this unit retards infiltration due to recharge and raises the water-table. 104 Figure 5.9 Salinity distribution obtained from Mode l 3a. a) Base case scenario shown for comparison purpose. b) Result fr om Model
xi 4a, where recharge was reduced by 10% from the base case while sea-level rise of 1m in 90 years and island m igration rate of 10m/yr were considered (i.e. similar to Model 2b ). The depth of the freshwater lens decreases to ~12 m by 6m from base case (18m) and by 13 m from sea-level scenario Model 2b (25 m). This shows that recharge plays a very im portant role in maintaining the freshwater resource. 105 Figure 5.10 Salinity distributions (kg/m3) obtained from Model 3b shown at different timesteps. The model is similar to sea le vel rise scenario, Model 3b, with a pumping well extracting groundwater from the upper aquifer at 100 m3/day causing saltwater intrusion 106. Figure 5.11 Comparison of changes in head distribut ion over the barrier island under conditions of increased recharge (by 1 0% from base case), reduced recharge (by 10% from base case ) and withdrawal through pumping. The most influencing fa ctor is recharge as also in figure 5.8b. 107 Figure 6.1 Atoll islands with different sizes wi th enclosed or semienclosed lagoons. A) Pingelap Atoll, Federal states of Micronesia B) Diego Garcia Atoll, Indian Ocean and C) Majuro Atoll, central-western Pacific Ocean. The di rection of prevailing winds influence the size of the islands, the size larger along the leeward side than the windward sid e where wind forces cause destruction of the atoll margins (from Bailey et al., 2008). 124 Figure 6.2 Typical hydrogeology of an atoll island showing a shallow freshwater lens with a transition/mixing zone withi n Holocene formations separated by the Thurber discontinuity f rom the underlying Pleistocene deposits. The arrows at the bottom indicate the direction of tidal propagation (adapte d from Bailey, 2009) 125 Figure 6.3 Relationship between atoll island width and maximum thickness of freshwater lens base on their mode of occurrence s within the atolls (from Bailey, 2009). 126 Figure 6.4 Model results of potable water (2.6% sal inity) depth at island centers with respect to island width for different atoll islands under different recharge (R) conditions (from Under wood, 1992) 127
xii Figure 6.5 Hydrogeologic units of Laura area, Majur o Atoll (from Anthony et al., 1989. 128 Figure 6.6 Extents of freshwater lens in Laura, Maj uro Atoll in 1984 (upper figure) and in 1998 (lower figure) across the centr al part of the island. (adapted from Presley, 2005) 129. Figure 6.7 a) Model plan view showing the position of well sites. Well site 1 consists of wells 1,2 and 3 and well site 2 has w ell numbers 4,5 and 6. Pumping rates are given in Table 6.3. b) Cross-sectional view of the model along AA sho wing the hydrologic units and boundary conditions. 130 Figure 6.8 Salinity distribution map of Model 1 sh own at different depths. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens boundary. At a depth of 1.5 m (uppe r figure) and depth of 4.5 m (lower figure) 131 Figure 6.9 Salinity distribution maps of Models 2a (upper figure), 2b (middle figure) and 2c (lower figure) shown at a de pth of 1.5 m. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens boundary. For a rise o f 0.5 m sea level, saltwater intrusion occurs from the ocean si de, increasing salinity over the reef flat plate area o f the island. Pumping induces saline intrusion along the lagoon s ide. Overall lens extent has reduced. For sea level rise of 1m, saltwater intrusion increases due to pumping close to the lagoon side, creating an irregular shape of the len s. Lens extent reduces from the first scenario. For a sea l evel rise of 1.5 m, saltwater intrusion continues along the lago on side, while fresh and salt water mixing widens the transi tion zone along the ocean side. Details given in text. 132 Figure 6.10 Salinity distribution maps of Models 2a (upper figure), 2b (middle figure) and 2c (lower figure) shown at a de pth of 4.5 m. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens boundary. For a rise o f 0.5 m sea level, only a freshwater nucleus exists at the isla nd centre with saline water elsewhere. At high sea level of 1m, th e freshwater nucleus reduces to pockets and diminishes with high er sea level rise of 1.5m. Saltwater intrusion due to pump ing is evident under all three sea level rise conditions. 133
xiii Figure 6.11 Changes in water table elevations under different rates of sea level rise. X axis represents distance in meters ac ross the model and y-axis is water table elevation above MSL (m) 134 Figure 8.1. Schematics of the fi rst four models to illustrate the bed thickness distribution within each sequence. Note s equences are longer than shown in schematics. 147 Figure 8. 2 JTFA of 1D FDTD model traces of layered sequences with input frequency of ~200 MHz. (The first 10 ns recor d the outgoing pulse.) Beds in all sequences are separate d by a 1mm layer representing a contact between successive beds. Upper panel of each figure shows the model result o ver layered sequences. Lower panels represent correspon ding Stransform of the model results. Results described i n text. 148 Figure 8.3 a) GPR reflection profile over hurrican e washover deposit in Santa Rosa Island, Florida (modified from Wang and Horwitz, 2007). Black box indicates trench position. Red lin e (Iv) demarcates the Ivan washover deposits (Iw) from the underlying pre-Ivan sediments. Blue lines indicate the trace positions that are extracted for analysis. (b) Phot o showing packages of dipping beds in homogeneous quartz sand Close inspection reveals internal layers composed of thin beds (~2 cm). (c) Amplitude spectra of the downgoing GPR pul se shows a dominant frequency of 290 MHz. All S-transf orm results of GPR traces are compared with respect to this dominant frequency. (d) Traces extracted from GPR p rofile and their S-transforms. Dashed red lines mark the c ontact of Ivan deposits (Iw) and the pre-Ivan deposits. Dashe d blue lines indicate the water-table reflector. Results o f JTFA discussed in text. 149 Figure 8.4. (a) GPR profi le over tephra deposit at Cerro Negro volcano, Nicaragua. Data at center of profile were collected with 100MHz frequency. Data at both ends were collected wit h 200MHz frequency. The deposit represents a sequence of eruptive events. The base 1850 reflector is believe d to be underlain by lava flows and older tephra. (b) Top p anel shows 100-MHz trace from reflection profile. Bottom panel shows its S-transform. Contacts marking the base of major dep osits, such as the 1992 and 1850 events, are characterized by lower frequencies than returns from within depositional u nits. Th in beds within the tephra deposit correspond to higher frequencies. (c) Map of relative energy at 125 MHz on GPR
xiv profile (4A). Zones with internal layers have stron ger returns at 125 MHz (red) than does the base 1850 reflector (we aker blue response), 150
xv Variable-density Flow Models of Saltwater Intrusion in Coastal Landforms in Response to Climate Change induced Sea Level Rise a nd a Chapter on Time-Frequency Analysis of Ground Penetrating Radar Signals Swagata Guha ABSTRACT Populations residing on and near the worlds coast s have become increasingly dependent on coastal groundwater for t heir supply of freshwater. Under the conditions of predicted climate changes, the expected rise in global sea level can adversely affect the quality and quan tity of freshwater resources in coastal areas as a result of saltwater intrusion. In this study, a suite of twoand three-dimensio nal variable-density groundwater flow models of major coastal landforms (e.g. deltas, estuaries and small islands) has been constructed to assess the e ffects of sea level rise (SLR), using different SLR rates of 0.5 m, 1m and 1.5 m ov er the next 90 years, from 2010-2100. The model results indicate that in natur al coastal systems the extent of saltwater intrusion is significantly controlled by the stratigraphy of the depositional environments. Among deltaic aquifers, wave-dominated deltas are
xvi more prone to saltwater intrusion than riverand t ide-dominated deltas. In case of a partially mixed, microtidal estuary, SLR can c ause extensive porewater salinity increases, especially within estuarine san d deposits. Simulations of atoll and barrier islands reveal that carbonate atoll isl ands with high conductivity units, are severely affected by SLR, resulting in signific ant reduction of the volume of freshwater lens. In contrast, migrating sandy barri er islands could retain their freshwater resources with rising sea level under co nditions of increased recharge, assuming the barriers can migrate in resp onse to SLR. The freshwater lens of barrier island aquifers would reduce in siz e due to increased evapotranspiration caused by change in vegetation p attern. When examined for anthropogenic impacts of groundw ater withdrawal through pumping, all the coastal aquifers show evid ence of saltwater intrusion, with varying degrees of impact. Wave-dominated delt as are more affected by groundwater withdrawal than riverand tide-dominat ed deltaic aquifers. Saltwater intrusion in atoll islands is further enhanced by p umping withdrawal. It is evident from the results of the simulations that, the poten tial effects on coastal aquifers of groundwater withdrawals for potable water can easil y exceed the adverse effects of SLR in terms of salinity increase.
1 CHAPTER 1 INTRODUCTION Climate change and sea level rise Successful adaptation to climate change includes p roper maintenance and management of natural water resources and is one of the greatest challenges of the 20th century. Abrupt variations in climatic conditions will have inevitable impacts on the hydrogeologic cycle. According to ob served data as well as climatic model predictions, changes in atmospheric temperature will cause changes in precipitation, soil moisture, runoff, in crease in dry land area and increased melting of ice as well as global ocean wa rming, leading to rise in global sea level (Arnell, 1995; IPCC Climate Change report 2007). Thus the sources and sinks within the hydrological regime will be af fected, thereby modifying the availability and distribution of natural water reso urces. In addition to climatic changes, increase in popul ation, economic growth and irrigational needs will create additional deman ds for water usage (e.g. van Dam, 1999; Vorosmarty, 2002). This could result in further stressed situations in some of the worlds currently stressed water basins such as those in northern Africa, Mediterranean regions, Middle East, Souther n Asia, Northern China, Australia, the U.S., Mexico, parts of Brazil and W est coast of South America (IPCC Climate Change report, 2007; Arnell 1995, 20 04).
2 An important factor associated with global climate change is the state of global sea level. Fluctuations in global sea level have been recorded in geologic past (Harvey, 2006). Over the 20th century, acceler ation in the rate of sea level rise (SLR) has been noted (Church and White, 2006; Woodworth, 2008; Jevrejeva et al., 2008). Measurements from differe nt sources as such geologic indicators, tide-gauges and altimeters indicate tha t SLR has been increasing over the past 130 years (Church et al., 2008). It is und erstood that global warming could lead to thermal expansion of ocean waters thu s elevating sea level (e.g. Cabanes, 2001). Furthermore, there remains the thre at of additional sea level rise through glacial melting (Church, 2001; Church et al., 2008). Melting of the Greenland ice sheet alone could contribute to the r ising sea level by 7 m while that of the West Antarctic Ice Sheet (WAIS) could b e responsible for 6 m of SLR (Shindell, 2007). Accurate measurement and prediction of SLR are com plex as the phenomenon is influenced by several other factors s uch as tectonic process, coastal subsidence and glacio-isostatic movements ( Harvey, 2006). With respect to the locally operating geologic and hydrogeologic processes, SLR is termed as relative. Prediction of SLR and its impacts is a matter of o ngoing debate, leading to different views regarding the rate of the process ( Edward, 2008). For example, a rise of 20-30 cm in global sea level ( Meehl, et al ., 2005) can be projected by 2100, if the polar ice sheets remain stable or the rise could be by 2 meters by 2100, if ice sheets become unstable (Pfeffer et al. 2008). Considering the IPCC
3 future warming scenarios, Rahmstorf (2007) presente d a projection of 0.5 m to 1.4 m of SLR by 2100 by relating historical SLR to temperature. Although the exact rate of SLR is debatable, it can be said for certain that the future sea level will be higher than at present. Rising sea level is of great concern due to the vu lnerability of a large portion of the human population residing in coastal areas worldwide. Coastal zones are economically vital locations that exhibit a strong trend in urbanization (Nicholls and Lowe, 2004). In 1990, 1.2 billion peo ple were reported to live within 100 km of coasts (Small and Nicholls, 2003). It is estimated that by 2010, 20 mega-cities e.g.New York, Tokyo, London and Mumbai will be near coasts (Nicholls, 1995). Already susceptible to natural ha zards and anthropogenic modifications, coastal areas will be strongly affec ted by the accelerating rise of global sea level (Nicholls, 2008; Harvey and Nichol ls, 2008). Even a rise of 2 m could displace 60 million people in mega-cities lik e Calcutta, India, or 40 million in Shanghai, China (Shindell, 2007). Densely popula ted countries of South America (Argentina, Brazil, Columbia) rely on groun dwater for 25 -50 % of potable supplies while in countries such as Chile, Peru, Venezuela, Surinam 50100% of drinking water is derived from groundwater (Bocanegra, 2010). Inhabitants of some Pacific small islands with population densities of 12,000 people/km2, depend on island aquifers as their only source of freshwater (White and Falkland, 2010). African countries like Algeria and Tunisia use groundwater for 67-95% of their water supply (Steyl and Dennis, 2010). Elevated sea level could extend zones of salinization through flooding cause saltwater intrusion into
4 fresh groundwater resources (IPCC Climate Change re port, 2007 and Climate Change and Water report, 2008), as well as acceler ate coastal land loss as a result of shoreline retreat (Pilkey, 2004), resulti ng in highly stressed water resources in many parts of the world. From the hydrogeological point of view, it is impo rtant to understand the impact of SLR on coastal aquifers in order to asses s the future availability of fresh groundwater resources. Coastal aquifers and saltwater intrusion In coastal areas, the coastal aquifer often serves as the only source for fresh groundwater. The coastal aquifers are in dire ct contact with seawater. Hydraulic head in the sea remains essentially const ant (except for tidal oscillations) while head along the coastal zone cha nges due to variable recharge rates. The resulting potential difference induces f reshwater to flow from coastal aquifers into the sea. Due to density differences, the lighter freshwater floats on top of a denser saltwater wedge. In island aquifers a freshwater lens is found to occur above saline water. Since freshwater and salt water are miscible fluids, a transition zone results by the process of hydrodyna mic dispersion and molecular diffusion (Bear, 1988) at the interface between the two. In the transition zone, the mixed water flows upward, due to salinity gradient, along with the freshwater towards the sea. The salinity balance is maintained by inland flow of salt water, thus forming a wedge structure (figure 1.1). The t hickness of the transition zone can vary from a few meters to hundreds of meters an d is controlled mainly by the
5 geologic heterogeneities and physical structure of the aquifer, recharge rate and freshwater extraction through pumping (Bobba, 1993) The Ghyben-Herzberg principle (Fetter, 1988) can be used to determine t he depth to the interface for a sharp interface approximation. The characteristics of coastal aquifers can be sim ilar to those of inland aquifers e.g. a coastal aquifer can be confined, un confined or multi-layered. However, island aquifers represent a unique hydrolo gical system where the freshwater aquifer has a lens shape that resides on top of dense saltwater. Figure 1.2, modified after Custodio, 2005, shows th e different types of coastal aquifers and the influence of hydrostratigraphic un its on the freshwater-saltwater interactions. Saltwater intrusion: a global concern One of the major threats to coastal aquifers is sa ltwater intrusion. Under conditions of equilibrium, the freshwater-saltwater interface (i.e. the position of the transition zone) remains stationary. However, excess pumping of freshwater from a coastal aquifer or SLR can cause the interfa ce to migrate landward. This phenomenon of fresh groundwater being displaced by saline groundwater is termed saltwater intrusion (Bear, 1988, Post and Abarca, 2009). Custodio, 2005 discussed the following causes, besides overpumping that can lead to increase in groundwater salinity: encroachment of modern saltwater mixing with unflushed old marine water
6 saltwater spray on windy coastal strips intense evapo-concentration of surface and phreati c water in dry climates intense evaporation of outflowing groundwater in di scharge areas and wetlands dissolution of evaporite salts from geologic forma tions pollution of saline water derived from mine drainag e, leakage from industrial processes saline water imported from other areas Contamination of groundwater by saline intrusion is a common threat to most coastal countries. Cases of saltwater intrusio n were reported as early as 1889 when Badon Ghyben noted the occurrence of sali ne water in Amsterdam while Herzberg (1901) found similar cases in German islands along the North Sea. Since then, several attempts have been made t o account for the nature of freshwater/saltwater mixing zones. Many recent studies conducted in different part s of the world show that saltwater intrusion has become a problem of great c oncern. The sections below provide a brief overview of the regions that have a lready experienced or are prone to saltwater intrusion. The United States The problem of saltwater intrusion in coastal aqui fers had been identified in 1854 on Long Island, New York (Back and Freeze, 1983 ). Thereafter, the extent of groundwater salinization has been reporte d from many coastal states, with Florida (carbonate aquifers), California (allu vial aquifers ), Texas and New
7 York being the most affected by saltwater intrusion (To dd, 1974). National salinity surveys undertaken in 1977 by the U.S. Env ironmental Protection Agency reported salinity increase in groundwater in 43 sta tes due to saltwater intrusion, including upwelling of deep basinal brines. A secon d survey in 1981 by the U.S. Water Resource Council indicated occurrences of sal inity changes both due to natural saline water and saltwater intrusion. Summa rizing the problem of groundwater quality degradation in the U.S., Atkins on et al. (1986) found that the problem was widespread, occurring at varying degree s from severe to negligible. Groundwater in 39 states including most states alon g both the east and west coasts were considered to have been affected by sal twater intrusion. Several measures to prevent and manage encroachment of salt water have been taken over the years. Konikow and Reilly, (1999) discuss the mode of saltwater intrusion and preventive measures undertaken in the Biscayne aquifer of south Florida and in Orange County, California. Recent studies by Barlow and Reichard (2010) show the mode of saline intrusion due to exc ess groundwater usage in southeastern Florida and central and southern Calif ornia. Vulnerability of US coastal aquifers to saltwater intrusion has been a longterm problem which will require additional protocol s for management of water resources if sea level rises. Europe The European coastline is geologically diverse com prising of i) coastal flatland sediment deposits of river basins of Thame s (UK), Garonne (France), Danube (Romania/Ukraine), Po (Italy) Rhone (France) Rhine (The Netherlands)
8 and Ebre (Spain), ii) crystalline and metamorphic r ocks of Scandinavia, Scotland, NW France and NW Iberia, iii) carbonate rocks in SW England, N France, along the European Mediterranean coasts and islands as we ll as iv) coral-reefs and volcanic islands (Custodio, 2010). Most of the coas tal regions of Europe, especially the deltaic, carbonate massifs and small islands, are threatened by saline intrusion either by upconing or by lateral m igration due to excess abstraction of groundwater (Custodio, 2010). Cases of saltwater intrusion have been documented in Greece (Lambrakis and Kallergis, 2001), Belgium (Vandenbohede and Lebbe, 2002), The Netherlands (Ou de Essink, 1996; 2001a,b) and Spain in the Rio Verde aquifer (Calvac he and Bosch, 1995). Research on the freshwater-saltwater behavior in co astal aquifers of Europe and elsewhere are discussed in the Salt Water Intrusion Meeting conference publications (De Breuck, 1991). Africa The African coasts are characterized by varied lan dforms such as rocky shores, sandy beaches, deltas, coastal wetlands, co ral reefs, lagoons and mangrove forests. Groundwater resources occur as al luvial, sandstone, carbonate and basaltic aquifers. Steyl and Dennis (2010) described the current stat e of saltwater intrusion in different parts of the continent. Countries like Egypt, Libya, and Tunisia in the north meet their freshwater demands by over abstrac tion of groundwater leading to severe cases of saltwater intrusion. For instanc e, seawater has encroached by 60 km inland within the Nile delta (alluvial aquife r). Similar examples of saltwater
9 intrusion are reported from the basaltic aquifer of Djibouti, alluvial aquifer of Kenya and Somalia on the east coast, coastal aquife r of Mozambique in the south and the entire coast of Nigeria in the west. Recurring events of droughts often increase the dem and of groundwater usage that has already led to increased salinity in groun dwater. SLR along the coasts of Africa would severely affect human conditions of li ving in these regions. Asia Although a comprehensive account of saltwater intr usion in Asia is lacking (Post and Abarca, 2009), cases of saltwater intrusi on are not uncommon. Instances of saline intrusion due to human impact h ave been reported from China i) in the Laizhou bay region where the freshw ater/saltwater interface has moved landward by 140 m (Ou et al., 1993), ii) in t he Beihai area near the southeast coast (Zhou et al., 1999) and iii) in the estu arine region of the Yangzte river due to reduced river discharge (Chen et al., 2001). Peninsular India is also affected by saline intrus ion e.g. along the west coast, in Mangalore (Rajesh and Murthy, 2004), in c oastal limestone of Saurashtra (Pujari and Soni, 2009) and coastal area s of Goa along the Arabian Sea (Kumar et al., 2007). On the eastern coast of India, increased salinity i n groundwater is recorded from the Godavari delta (Bobba, 2002), Chennai coastal r egion of the southeast (Shanmugan et al., 2007) and coastal aquifer of Dig ha, West Bengal (Choudhury et al., 2001). Studies in Bangladesh also indicate the evidence of groundwater quality degradation through saline intrusion (Nobi and Dasgupta, 1997). With
10 only 10 % area being 1 m above mean sea level, majo r portions of the country are prone to climatic effects and saline intrusion (e.g. Ali, 1999). More examples of saltwater intrusion can be cited from southeast Asia e.g. from Jakarta, Indonesia (Delinom, 2008) as well as from the south west e.g. from coastal aquifers of Oman (Shamms and Jacks, 2007). Since some of the worlds most populated cities an d agricultural areas are located in coastal regions of Asian countries, the effect of SLR and accompanying saltwater intrusion would pose serious issues. South America Bordered by the Atlantic Ocean in the east and by the Pacific in the west, coastal aquifers of South America have also been re ported to suffer from overexploitation of groundwater thus leading to sal inization. In an overview of current knowledge of South American water resources Bocanegra et al. (2010) refer to 15 coastal aquifers and small island aquif ers. Saltwater intrusion due to SLR in these coastal aquifers could be very likely. Australia Coastal aquifers in Australia have been experienci ng saltwater intrusion due to excess abstraction of groundwater. Although, some of the aquifers such as in Queensland are managed for the risks of saltw ater intrusion, there are less well maintained groundwater resources in Western an d South Australia (Werner, 2010). Climate change, increased population and sea level rise will intensify the need for proper water resource management in the co astal areas of the continent.
11 Small islands of the Pacific Water resources in small islands (area of 10km2) are considered the most vulnerable to climate change. Along with variabilit y in precipitation, temperature, and risks of inundation from storm surges, SLR in t hese low-lying islands can be a major problem. There exist nearly 1000 small isla nds in the Pacific Ocean within the tropical to subtropical zones, where loc al populations rely on groundwater as the sole supply of freshwater (White and Falkland, 2010). Often, these islands are composed of high permeability car bonates e.g. atoll islands of Kiribati, Marshall Islands, Tuvalu, Tokelau, and some parts of the Cook Island s, French Polynesia, Papua New Guinea, and Tonga (Vacher and Quinn, 1997) where freshwater remains as a lens atop saltwater. Demand of freshwater is high and overpumping is a common practice, leading to sa line intrusion ( e.g. Tarawa and Kiritimati Atolls of Kiribati). T he effect of rising sea level could reduce the freshwater resources in these islands th us causing drastic changes to living conditions of the inhabitants. Objective of present study Vulnerability to saltwater intrusion means reducti on in freshwater resources and degradation of water quality, thereby creating stressed situations for coastal inhabitants and ecosystems. For the pur pose of freshwater resource management, vulnerability assessment, mitigation an d future adaptation in coastal zones, it is necessary to have a better und erstanding of the extent and pattern of saltwater intrusion in these areas.
12 The objective of this study is to assess the exten t of salinization and availability of freshwater resources in coastal zon es of various landforms (e.g. deltas, estuaries, atolls, barrier islands) as a re sult of climate change induced SLR.
13 Figure 1.1. Ground-water flow patterns and the zo ne of dispersion in an idealized, homogeneous coastal aquifer ( Cooper, 1959 ).
14 Figure 1.2 a. A) a fully confined coastal aquifer, B) partly leaky confined coastal aquifer formed by low permeability, thinning out, l ayer causing upward discharge of freshwater, C) a leaky coastal aquifer where a t hin semipermeable layer creates a confined aquifer that shows no freshwater discharge to the sea (modified after Custodio, 2005).
15 Figure 1.2. b. A multilayered coastal aquifer with low permeability layers forming confined aquifers with different positions of mixin g zones (modified after Custodio, 2005).
16 Figure 1.2 c. Coastal deltaic aquifers. An unconfi ned aquifer (upper figure) with a river that allows freshwater discharge to the sea ( similar situation in Lower Llobregat river valley and delta, Barcelona, Spain) Holocene seawater intruded within the deep aquifer but was washed away by fres hwater discharge from the river. In the lower figure (similar to Lower Ebre r iver valley and delta, Southern Catalonia, Spain) the river intersects the upper un confined aquifer, thus no freshwater head is present to allow freshwater disc harge. Holocene seawater within the lower deep aquifer slowly flows upstream (modified after Custodio, 2005)..
17 Figure 1.2. d. Island aquifers. In volcanic island s (upper figure) the presence of saltwater could be limited to narrow coastal strips coastal plains in recent volcanoes, alluvium or slope deposits. In a low het erogeneous island formed by carbonate and marl (middle figure) a central freshw ater lens is present while saltwater wedges intrude the more permeable layers. In lower figure, a freshwater lens floats on top of saline water prese nt within highly permeable lower units (modified after Custodio, 2005).
18 Figure 1.2. e. Freshwater-saltwater distribution i n a near-shore dune and intermediate low land environment (e.g. as in Nethe rlands, Belgium, Mar del Plata of Argentina). The lowlands may represent eva poration ponds, externally fed freshwater lakes, artificially drained areas (a s in Netherlands) (modified after Custodio, 2005).
19 CHAPTER 2 METHOD Variable-density saltwater intrusion In coastal aquifers, less dense freshwater resides over more dense seawater allowing stable fluid flow. Mixing of the two miscible fluids takes place by diffusion and hydrodynamic dispersion of total d issolved solids creating a transition/mixing zone. The process of saltwater in trusion in freshwater aquifer is a variable-density flow problem involving mass tran sport and flow of fluids of two different densities. A sharp freshwater-saltwater i nterface may be considered where the aquifer thickness is significantly large with respect to the width of the transition zone, in which case the two fluids are a ssumed to be immiscible. However, in many coastal aquifers the width of the transition zone is significant (e.g. the Biscayne aquifer in Florida). In such cas es, the groundwater flow regime should be treated as a density-dependent flow syste m. Numerous studies have addressed the topic of varia ble density flow resulting in benchmark problems such as the Elder ( 1964) and Henry (1960) models. An extensive bibliography of previous works on variable-density flow is given in Diersch and Kolditz (2002) and Simmons et al. (2001).
20 Early works on saltwater intrusion Some of the earliest theoretical studies on saltwat er intrusion date back to the late 18th and early 19th centuries when Badon Ghyben (1889) and Herzberg (1901) independently developed an equation, also kn own as Ghyben-Herzberg principle, to calculate the depth to the interface between freshwater and saltwater under hydrostatic equilibrium. Later, Hubbert (1940 ) formulated a similar equation to locate the interface under dynamic cond itions. Since then several authors have provided solutions for saltwater intru sion problems such as determining the shape of the interface (Glover, 195 9) and the movement of the interface (Bear and Dagan, 1964; Hantush, 1968; Sha mir and Dagan, 1971). Cooper (1959) considered the dynamic nature of salt water movement and its circulation pattern from the sea to the transition zone and back. Pinder and Cooper (1970) simulated a confined coastal aquifer using coupled groundwater flow and mass transport mechanisms. An elaborate re view of the historical development of the research on freshwater-saltwater phenomenon is available in Reily and Goodman (1985) and Bobba (1993). Over the years, various methods including analytic al, numerical, field investigations, geophysical and geochemical have be en used to understand the freshwater-saltwater interface location and movemen t (Bear, 1999). In this study, numerical modeling is applied to ana lyze predictive saltwater intrusion behavior.
21 Numerical methods The solution for variable-density flow problems in volves solving for the coupled density-driven flow equation and advection dispersion/diffusion equation. However, available analytical solutions a llow several assumptions (e.g. homogeneous, isotropic aquifer, one or two-dimensio nal model) leading to simplified representations (Anderson and Woessner, 2002). Moreover, these approximations for the coupled governing equations involving simple geometries and boundary conditions can become complex, thus li miting their application (Dentz et al., 2006). In such situations, solution of the saltwater intr usion problem can be attained using numerical methods (Bobba, 1993). Num erical simulations allow easy assignment of spatially variable hydraulic par ameters to represent the different hydrostratigraphic units as well as imple mentation of suitable boundary conditions. With the increase in computational capacity, numer ical simulations of three-dimensional variable-density flow systems are possible. In recent years, different computer codes based on finite-difference (e.g. USGS SEAWAT) and finite-element (e.g. USGS SUTRA) methods have becom e available for simulation of density-driven problems. Some other m odeling techniques that have been used to solve saltwater intrusion problem s include inverse modeling (Sanz and Voss, 2006) and the Lattice Boltzman meth od (e.g. Servan-Comas, 2009).
22 It is to be remembered that the numerical modeling technique is not free from complications in defining a density-dependant hydrological system. Simmons (2005) described the challenges faced in th e current state of variabledensity modeling, emphasizing on the problems relat ed to field-scale parameter measurements, incorporation of heterogeneity as a c ontrolling factor and simplified representation in spatial and temporal d iscritization of the hydrologic system in numerical models. Modeling saltwater intrusion with SEAWAT For the present study, a computer program, SEAWAT (Guo and Langevin, 2002) which simulates three-dimensional, transient, variable-density flow and transport models is used. SEAWAT has been successfu lly used for saltwater movement studies (e.g. Dausman, and Langevin, 2005; Don et al., 2006; Langevin, 2003; Robinson et al., 2006; Zimmerman e t al., 2006; Masterson and Garabedian, 2007). The variable-density groundwate r flow equation (Guo and Langevin, 2002) in terms of equivalent head is give n by where a b g are orthogonal coordinate axes, aligned with the pr incipal directions of permeability; Kf is equivalent freshwater hydraulic conductivity; Sf is equivalent freshwater specific storage ; t is time; r is effective porosity; C is
23 solute concentration; rs is fluid density source or sink water; and qs is the volumetric flow rate of sources and sinks per unit volume of aquifer. The equivalent freshwater head is defined as where r is the density of the native aquifer water ; rf is the density of freshwater and Z is the elevation at the measurement point. Hydrogeological models for the study areas in this work are reconstructed based on published stratigraphic data and sedimenta ry facies models. Detailed description of the modeling involved is given in ea ch of the following chapters. It is taken into account that the variable-density sal twater intrusion models are simplistic and exclude micro-scale heterogeneity or sedimentary structures that may control flow and transport. In practice, incorp oration of such detailed heterogeneity in modeling studies is a complex and difficult task that demands further research.
24 CHAPTER 3 IMPACT OF SEA LEVEL RISE ON DELTAIC AQUIFERS Introduction Deltaic plains, shaped by fluvial and oceanic proce sses, are the most valuable of the coastal landforms because of their importance as centers of human settlement that depend on fertile resources f or agricultural and economic growth. However, these increasingly populous coasta l areas (e.g. ~50 million on Nile delta, Syvitski, 2007; 250 million on Ganges-B rahmaputra delta, Michael and Voss, 2009) are highly vulnerable to the effects of SLR as a result of their low topographic gradients, natural subsidence from sedi ment compaction, as well as subsidence caused by anthropogenic effects, particu larly withdrawal of subsurface fluids. Human impacts leading to subside nce of deltas are primarily caused by oil exploration (e.g. in deltas of Po, Ni ger, Magdalena, Mahakam, MacKenzie, Mississippi, Yellow River ( Syvitski, 20 07) and by groundwater extraction (e.g. in Chao Phraya river delta, Thaila nd, Shanghai on Yangtze delta; Syvitski, 2007). Other anthropogenic effects on del ta morphology include modifications of sediment flux by reservoir constru ction (e.g. in deltas of Po, Mississippi, Nile, Danube, Yangtze, Colorado, ( Syv itski, 2007; Vorosmarty et al., 2001). All these factors contribute toward vulnerab ility of deltas to an effective sea level rise or ESLR (Ericson et al., 2006). Sust ainability of deltas will be
25 further challenged by an additional increase of ESL R as a result of climate change related sea level increase. Hydrogeologically, the impact of ESLR will be mani fested by inundation of coastal areas and saltwater intrusion in deltaic aq uifers. Increasing salinization in deltas can potentially jeopardize urban development by affecting agricultural activities and contaminating potable fresh ground w ater. Deltaic plains receive much attention in terms of coastal management (e.g. Mississippi delta), sedimentation processes ( e.g. Nile delta), and contamination of groundwater (Ganges-Brahmaputra de lta, Niger delta). However, saltwater intrusion in deltaic aquifers is relatively less studied. Some examples of saline intrusion include field investig ations (e.g. Nile delta, Sheriff, 1999; Yangtze delta, Chen and Zhong, 1999; Niger de lta, Edet and Okereke, 2001; Po delta, Antonelli et al., 2007; Rhone delta de Montely et al., 2008, saltwater impact from hurricane Katrina on Mississi ppi delta, Williams, 2010) and numerical models (e.g. Nile delta, Kashef, 1989; Sh eriff and Singh, 1999; Godavari delta, Bobba; parts of Ganges-Brahmaputra delta, Hassan, 1992, Po delta, Giambastiani et al., 2007, Okavango delta of Botswana, Milzow, 2009). In recent studies, Michael and Voss ( 2009 a,b ) prese nted an estimation of hydraulic properties and flow models for the Ganges -Brahmaputra delta. Their groundwater models involved single-density flows. Regional-scale groundwater flow systems are comple x due to the heterogeneous nature of the depositional environmen t. As in the case of deltas, several other geomorphologic components (e.g. barri er island, lagoons, tidal flats
26 and inlets) are present that have distinctive hydro logical properties and thus can influence the flow and transport mechanisms. The la ck of better understanding of variable-density flow processes in heterogeneous me dia (e.g. Simmons, 2005) has limited the number of studies related to saltwa ter intrusion in deltaic aquifers. This chapter addresses the effect of relative sea level rise or RSLR, i.e. eustatic sea level rise and subsidence on river-do minated, wave-dominated and tide-dominated deltaic aquifers with the aid of var iable-density groundwater flow models. Characteristics of deltas Although characterized primarily by fluvial process deltaic environments can have varied sedimentological signatures, depend ing on several factors related to the drainage basins such as tectonic set ting, geology, sediment supply, river dimensions, discharge rate, dominant depositi onal processes and precipitation rates in catchment areas (e.g Davis, 1985). Sediments in deltas show great textural variability ranging from coarse grained sand and gravel to fine silt and clay. This variation is the result of the diverse sediment load delivered to deltaic depositional environments by m ajor rivers and their distributaries. Geomorphologically, deltas consist of two parts, i ) delta plain and ii) delta front (Reading, 1986). The delta plain consists of lowlands with the major river channel and its distributaries, both active and aba ndoned. The delta plain terminates at the shoreline and is characterized by other geologic features such as tidal inlets, tidal flats marshes and mangrove s wamps (e.g. tropical areas in
27 Ganges-Brahmaputra delta) and by salinas with halit e and gypsum deposits in arid conditions (e.g. Nile delta). The delta plain can be both fluvially or tidally influenced. The delta front, extending offshore with a seaward -dipping profile, is composed of coarse sediments at the mouth of the ri ver and finer sediments in deep water. The delta front is dynamic, prograding seaward with increasing sediment supply. This allows sediments to accumulat e in the receiving basin and to form a coarsening upward sequence. Progradation and associated sedimentary facies development in deltas are contro lled by the amount of sediment supply and the positions of river distribu taries. Further descriptions of sediment deposition processes and delta facies, anc ient and modern, are available in several books (e.g. Warner and James, 1992; Reading, 1986; Davis, 1985; Woodroffe, 2003). The most commonly referred delta classification b ased on the relative influence of fluvial and marine processes (waves an d tides) is that by Galloway (1975). According to this classification, deltas ar e river-dominated, wavedominated and tide-dominated. A river-dominated del ta is characterized by a digitate shape and lateral migration of meandering channels. Tide-dominated deltas are associated tidal creeks and channels wit h tidal ridges that occur parallel to the directions of tidal currents while wave-dominated deltas develop barrier islands and lagoons at their mouths. Most d eltas, however, show influence of more than one process affecting their formation (e.g. the Mississippi delta, Ganges-Brahmaputra-Meghna delta, Wooddroffe, 2003).
28 Groundwater flow models Variable-density models of three delta types, rive r-dominated, tidedominated and wave dominated deltas were simulated based on sedimentation patterns in Mississippi river delta (e.g Fisk et al., 1954; Penland et al., 1990), Ganges-Brahmaputra delta ( Goodbred an Khuel, 2000) and Nile delta (e.g. Elewa and Nahry, 2009; Kashef, 1989; Woodroffe, 200 3). All groundwater flow models were simulated using the computer code SEAWA T (described in chapter 2) that incorporates the flow modeling codes of MOD FLOW and those of transport in MT3D. Models were designed and execute d using the software Groundwater Vistas (GWV version 5.22). The simulati ons used Hybrid Method of Characteristics technique (HMOC) for solving the tr ansport equations. HMOC combines the methods of MOC (Method of Characterist ics) and MMOC (Mixed Method of Characteristics) to automatically adapt a solution technique depending on the characteristics of the concentration field ( Zheng and Wang, 1999). Solutions derived by HMOC have greater accuracy tha n other methods and insignificant numerical dispersion over a large ran ge of Peclet numbers (Pe= D l / a where D l is a nodal spacing and a is dispersivity). HMOC also requires fewer particles to simulate the transport mechanism, whic h saves computational time. Each model was run using a maximum number of partic les equal to 64 per cell. Model assumptions The models assume that the rate of sediment supply is adequate to match the rate of RSLR. It is to be remembered that many river delta, show arrested
29 sediment supply e.g. Yangtze river (Yang et al., 20 05). No shoreline erosion, landloss and change in river course were considered The delta plains were assumed to host unconfined, single, regional aquife rs. The SLR scenarios take into an account a subsidence rate of 0.002 m/yr, re sulting in 1.8 m additional rise of sea level. So RSLR is considered in the followin g simulations. Effects of topography, vegetation and seasonal variability of precipitation were ignored. Model 1: River-dominated delta Model description Model design A three dimensional finite difference grid consis ting of 88 rows, 49 columns and 10 layers was generated to simulate a r iver-dominated deltaic coastal aquifer. Variable cell sizes were used to r epresent the varying width of the river channel. The cell sizes vary from a minim um D x equals to 1750 m and a maximum D x equal to 3500 m while D y values vary from 1250 m to 5687 m. Figure 3.1 shows the plan view and a cross-section of the model domain with the hydrologic units and boundary conditions used. Time discretization For an accurate solution of the variable-density f low problem, a specific time stepping is required which is determined by th e Courant number. Courant number is given by C = v D t / D l, where v is velocity, D t is transport time step interval and D l is nodal grid spacing
30 C = v D t / D l should be less than or equal to one, i.e. D t < D l/v (Anderson and Woessner, 1992). For the SEAWAT models, the Courant number was specified as 1, allowing maximum flow and transport over a di stance equal to one grid spacing (i.e. if C=1, then v D t = D l ). Hydrologic units The deltaic aquifer is primarily composed of mediu m to very fine sand, silt and clay units. An extensive delta plain formed of medium sand forms the upper unit. Underlying units are composed of fine to very fine sand layers. The delta front shows horizontal variation in hydraulic condu ctivity offshore due to the deposition of silt and clay units. Thicknesses of i ndividual hydrologic units forming the coastal aquifer range from 10 to 20 m. Figure 3.1 shows the distribution of hydrologic units and the layer thic knesses. The river-dominated deltaic aquifer is thus mostly characterized by mod erate to low conductivity units. Hydraulic parameters Table 3.1 gives the hydraulic parameters (hydrauli c conductivity, porosity, specific storage etc) used in the model, primarily based on values given in Schwartz and Zhang (2002). Dispersion plays an impo rtant role in variabledensity flow modeling, influencing the extent of fr esh and saline water mixing. For the regional-scale models considered, dispersivity was assumed to be scaledependent. High values of horizontal longitudinal d ispersivity were used as discussed in Gelhar et al. (1992). The large horizo ntal longitudinal dispersivity promotes solute transport in the direction horizont al to groundwater flow direction. Recharge values of 7E-005 m/day was used The parameter values
31 listed were obtained after modifying the values to minimize the mass balance error. Boundary conditions A numerical model is defined by governing equation s and boundary conditions. In groundwater flow, the flow equations solve for the hydraulic head which is a dependent variable. Boundary conditions are implemented to attain a correct solution of the problem by specifying the h ead or its derivatives, resulting in three types of boundary conditions such as i) s pecified head boundaries, ii) specified flux boundaries and iii) mixed or head-de pendant boundaries (Reilly and Harbaugh, 2004). In a variable-density flow pro blem, concentration is also required to be specified along the solution domain. In all the following models, the sea level is repr esented by a Constant Head Boundary (CHB) condition where the head of the sea level and the chloride concentration in seawater was set to zero meters an d 35 kg/m3 respectively. The head values in CHB cells were changed to successive higher values while modeling RSLR scenarios. The water table formed a surface boundary which received water through recharge. To represent a riv er in the deltaic plain, river cells were inserted into the model grid by using RI V package of MODFLOW (McDonald and Harbough, 1988). The RIV package allo ws surface water and groundwater interactions. Flow in the river cells i s determined by a conductance term. Conductance is controlled by the hydraulic co nductivity and cross-sectional area of the river channel in a RIV cell. In this si mulation, a conductance value of 100 m2/day was used. This value was obtained after severa l readjustments untill
32 a reasonable head distribution was attained and mas s balance error was minimized. In an additional scenario, the Well pack age of MODFLOW was used to simulate groundwater withdrawal. Model simulations Base case scenario A base case model was operated with an assumed sal inity distribution over the model domain. A freshwater-saltwater bound ary was considered at a distance of 9300 m inland from the tip of the delt a lobes. Freshwater concentration was set to zero kg/m3 over the delta plain while saltwater with salinity of 35 kg/m3 was set in the remaining parts of the model. The b ase case model was run for 200 years at which point equilibr ium was established. The head and concentration values obtained from this mo del served as initial conditions for the RSLR scenarios. Relative sea level rise scenarios Scenario 1 involves climate-related SLR of 0.5 m o ver a period of 90 years. Due to a subsidence of 2 mm/yr, a net subsid ence of 0.18 m in 90 years was added to 0.5 m, resulting in a RSLR of 0.68 m. The head in CHB cells were increased in time steps of 10 years. RSLR was consi dered steady within each time step. A total of 9 time steps were simulated b y using head and concentration distributions obtained from previous time steps int o successive time steps. In scenario 2, RSLR of 1.18 m by 2100 was consider ed of which 1 m rise is due to climate change and 0.18 m is due to subsi dence. Only the head values in CHB cells were changed, keeping all other parame ters constant.
33 In scenario 3, RSLR of 1.68 m (1.5 due to climate -related SLR and 0.18 m due to subsidence) was simulated in the same proc edure as the previous scenarios without modifying any parameters except h ead values in CHB cells. Groundwater withdrawal scenario Since deltaic plains are largely inhabited by huma n populations and rarely exist as natural systems, an additional scenario of groundwater withdrawal is considered to evaluate anthropogenic impact. In ord er to understand the effect of pumping fresh groundwater from the deltaic aquifer under conditions of RSLR of 1.18 m, a set of wells were inserted using well bou ndary conditions (figure 3.1). Pumping rate in each cell representing a well is 50 0 m3/day. Results and discussions Model results of RSLR scenarios are shown in figu res 3.2 and 3.3. Each figure shows a comparison of the salinity distribut ions under the three scenarios with respect to the base case at a given depth. The U.S. EPA standard chloride concentration for potable water is 0.25 kg/m3 (250 mg/l). So in the case of a riverdominated deltaic aquifer the 0.25 kg/m3 isochlor is considered to compare the extent of available potable water. In figure 3.2, at a depth of 1 m, a slight landwa rd advancement of the 0.25 kg/m3 isochlor is observed, under the three RSLR conditi ons. The movement occurs on the sides of the river as well as along i ts course. Salinity changes in the delta lobe from the base case, especially along the edges with low salinity (814 kg/m3) water displacing the high saline water (18-22 kg/ m3).
34 In figure 3.3, at a depth of 7 m, high salinity (1 8-22 kg/m3) distribution occurs over the delta lobe. Very subtle advancement s of 0.25 kg/m3 isochlor are noted along the sides and in the river channel. How ever, salinity distribution under each RSLR scenario shows no significant diffe rence. From the above results, it is evident that RSLR w ill have insignificant effect on the availability of potable water on a ri ver-dominated delta with no human impacts. As sea level rises, the head over th e coastal aquifer increases, leading to greater discharge that impedes the salin e front from landward intrusion. This assumes that the land elevation is high enough to accommodate the rise in the water table. Increased discharge cr eates greater fresh and saltwater mixing, leading to decrease in salinity a s observed in the delta lobe areas. The limited saline intrusion is also due to the widespread presence of relatively low conductivity units such as silt, cla y and fine to very fine sand that comprise a major part of the aquifer. Both groundwa ter flow and dispersion controlled saline mixing are slow processes. Thus, for the given time period and the rates of RSLR considered negligible loss in pot able groundwater takes place in a natural river-dominated delta. Result from the groundwater withdrawal scenario is shown in figure 3.4. Withdrawal of 500m3/day over 90 years under RSLR of 1.18 m can cause definite loss of potable freshwater. In figure 3.5, at different depths, it is noted that 0.25 kg/m3 isochlor advances in the vicinity of the pumping w ell. The effect is prominent at a depth of 1 m and 7 m.
35 Model 2: Wave-dominated delta Model description Model design A wave-dominated delta model was generated using three dimensional finite difference grid consisting of 98 rows, 54 co lumns and 11 layers. The cell dimensions are D x of 1523 to 5312 m and D y equal to 1283 to 5894 m. Nonuniform grid cell sizes were used to construct t he river channel. Time discretization criterion and boundary conditi ons implemented used are identical to Model 1. River conductance of 160 m3/day was used in the river cells. Hydrologic units The wave-dominated delta plain is characterized by very fine sand, silt and clay deposits. In the northern part of the mode l domain, high conductivity alluvium (sand and gravel) units are present. Other hydrologic units are composed of coarse sands and lagoonal mud. Sandy mu d is deposited around the river mouths. Medium to coarse sand units occur at the terminal end of the delta plain. The top layers are underlain by ~50 m of fine sand, followed by 200 m of high conductivity alluvium unit. Thus the aqui fer of a wave-dominated delta is mostly composed of relatively high conductivity units as opposed to low conductivity aquifer of a river-dominated delta. Of fshore, the delta front shows lateral heterogeneity formed by units of medium to fine sand to very fine sand and silt and clay units. The model plan and cross-s ectional view with hydrologic units and boundary conditions are given in figure 3 .7.
36 Hydraulic parameters Hydraulic parameters (hydraulic conductivity, poro sity, specific storage etc) used in the model are given in table 3.2. A s cale-dependent dispersivity with high longitudinal dispersivity is considered in thi s case, similar to Model 1. Recharge value of 7E-005 m/day and evapotranspirati on of 3E-005 m/day were applied. The parameter values listed were obtained after modification to initial runs. With the current parameter values, the mass b alance error is negligible. Model simulations Base case scenario An initial model with specified salinity distribut ion was simulated for 200 years with sea level at zero meters. Chloride conce ntration in the delta plain was set to that of freshwater. The delta front region r epresented sea water salinity of 35 kg/m3. The head and concentration data generated from th is model were used as initial values for the subsequent sea level rise scenarios. Relative sea level rise scenarios The RSLR scenarios for this model were simulated f ollowing the same procedure as in Model 1, described in preceding sec tion. Groundwater withdrawal scenario A set of pumping wells were inserted (figure 3.7) using the Well package. A steady-state pumping rate of 500 m3/day was applied.
37 Results and discussions The results of model scenarios are shown in figure s 3.8 to 3.10. In each figure the 1kg/m3 isochlor, considered here to repr esent the freshwater extent, is plotted at different depths. Figure 3.8a shows the results for extent of saltwater in the coastal aquifer for the base case scenario at d ifferent depths. Salinity increases within the aquifer with increasing depth. In figure 3.8b, the model result for RSLR scenario 1 is shown. Slight landward movem ent of the 1kg/m3 isochlor at a depth of 0.75 m is observed. The fresh-saltwat er interface at further depths remains static with respect to the base case scenar io. Figures 3.9 (a) and 3.9 (b) show model results fro m RSLR scenarios 2 and 3 respectively. When RSLR equals to 1.18 m afte r 90 years, the 1 kg/m3 isochlor at a depth of 20 m shifts landward and coi ncides with that the isochlor at 40 m. Isochlors at further depths of 62.5 m, 87.5 m and 125 m show landward movement by a small distance. The 1 kg/m3 isochlor at a depth of 175 m remains stable. For the RSLR 1.68 m scenario, the isochlor at a depth of 125 m moves landward by ~5,000 m (i.e. 5 km). Figure 3.10 shows the model result from the ground water withdrawal scenario under a condition of RSLR equal to 1.18m. Significant changes occur to the location of the 1 kg/m3 isochlor, as it advances landward with increasing depth. A net landward movement of ~8000 m or 8 km i s observed. The model results indicate that saltwater intrusio n occurs at greater depths when RSLR is higher for a natural wave-domin ated deltaic aquifer. Presence of high conductivity alluvium (sand and gr avel) deposits promotes
38 saline encroachment, demonstrating that flow and tr ansport are controlled by stratigraphy. Similar numerical models of the Nile delta (Sheriff and Singh, 1999) indicate that for a 0.5 m rise in sea level, saltwater intrudes a bout 9 km inland. Consideration of a groundwater withdrawal scenario suggests that a wave-dominated deltaic aquifer is highly susceptible to saline intrusion. Model 3: Tide-dominated delta Model description Model design The tide-dominated deltaic aquifer model consists of a three dimensional finite-difference grid with 98 rows, 59 columns and 10 layers. Grid cell sizes vary from a minimum D x of 933 and maximum D x of 7000 m while D y varies between 1283 m to 5894 m. Variable cell sizes were used to design the river channel and the tidal inlets. Boundary conditions Time discretization for the computations and bound ary conditions applied are identical to previous models (Model 1 and Model 2). The conductance used in the river cells is 420 m2/day. This conductance value was used as it produce d a reasonable head distribution through river and aq uifer interactions. Tidal inlets at the shoreline of the delta plain were assumed to be extensions of the ocean and were simulated with CHB cells with head and con centration values of the ocean. For RSLR scenarios, the head values in the i nlet cells were equally increased as in the main body of the ocean.
39 Hydrologic units The surfacial layer of the delta plain is mostly c omposed of fine sand with mud, sandy and silty mud deposits around the mouth of the river. The delta front extends into the ocean and is comprised of silt and clay sediments. The aquifer consists of fine to medium sand units in the upper 100 m that overlies a thick succession of coarse sand deposit of 250 m, forming a basal unit. The tidedominated coastal aquifer is thus a heterogeneous h ydrologic system. Figure 3.11 shows the plan and cross-sectional view of the model with the distribution of hydrologic units and boundary conditions used. Hydraulic parameters Table 3.3 gives the hydraulic parameters (hydrauli c conductivity, porosity, specific storage etc) used in the model that were o btained from Schwartz and Zhang (2002) and Michael and Voss (2009). The low v alues (of the order of 10-3 to 10-4 m) of vertical hydraulic conductivity used were ba sed on the parameter estimations in the tide-dominated delta of the Gang es-Brahmaputra by Michael and Voss (2009). Recharge and evapotranspiration ra tes were both set to 0.0008 m/day. The distribution of hydraulic conductivity a ssociated with the hydrologic units and the parameter values listed in table 3.3 were obtained after several modifications, thus minimizing the mass balance err ors. Model simulations Base case scenario A base case model was created with an assumed sali nity distribution with gradually decreasing concentrations landward in ord er to produce a tidal zone. The salinity field generated based on the assumed d istribution, thus has a wide
40 fresh to saltwater zone, extending over half of the delta plain. The base case model was run for 200 years, until heads became sta ble. Head and concentration distributions were used as initial co nditions for the following models. RSLR and groundwater withdrawal scenarios These simulations were operated identically as pre vious models 1 and 2. Results and discussions Figure 3.10 shows the model results for the base c ase and RSLR scenarios. The RSLR results are compared with the b ase case at depths of 0.75 m. The 0.25 kg/m3 isochlor representing the extent of potable ground water is monitored in each case. In figure 3.10, at a depth of 0.75 m, under all th ree RSLR scenarios, the 0.25 kg/m3 isochlor has advanced in the northeastern part of the domain, indicating loss of potable water. A small intrusio n is noted along the river channel. The river channel influences the distribut ion of saline water with salinity increasing along its channels with RSLR.. At furthe r depths of the aquifer, no significant salinity changes occurred. Figure 3.11 shows the salinity distributions from the groundwater withdrawal model. Comparison of these results with the second RSLR scenario (1.18 m) indicates that the 0.25 kg/m3 isochlor shows inland movement in the vicinity of the well field, particularly at depths of 3.25 m and 7. 75 m. The model results from the RSLR scenarios of a nat ural tide-dominated delta reveal that the influence of saltwater intrus ion on the deltaic aquifer is
41 minimal. The existence of very low vertical hydraul ic conductivity units (i.e. high vertical anisotropy of Kh/Kv ) controls groundwater flow at depth which in turn limits saline intrusion from depth. Also the presen ce of silt and clay units in the delta front region retards saltwater movement. As observed in the cases of previous models (1 and 2), the tide-dominated delta will be affected by groundwater pumping resul ting in reduction of potable water. This result also agrees with Michael and Vo sss (2009) study of groundwater flow in the Bengal Basin where predevel opment scenarios show long, horizontal, regional groundwater flowpaths at depth. Human influences of groundwater withdrawal have changed flow directions inducing more vertical flows in the shallow part of the aquifer. These ver tical flows can induce upward movement of saline water. Conclusions Variable-density flow models of three natural delta ic aquifers under different RSLR scenarios indicate the extent of sal inization in each case is controlled by the stratigraphy of the depositional setting. In case of a river-dominated delta, the occurrence s of low conductivity units impede saltwater intrusion. Loss of potable w ater in the aquifer is minimal, occurring under conditions of high RSLR of 1.68 m. Salinity distribution is also influenced by the course of the river channel. In a tide-dominated deltaic aquifer, with heteroge neous hydrologic units, a similar response is observed, suggesting a role of high vertical anisotropy
42 ( Kh/Kv ) in controlling horizontal groundwater flow and ar resting saline intrusion from below. Under natural conditions, a tide-domina ted delta is not affected by saltwater intrusion or suffers from significant los s of potable groundwater. In a wave-dominated, high conductivity sand and gr avel aquifer, RSLR of 1.68 m causes landward movement of the freshwater-s altwater interface by 5 km, implying saline intrusion into the aquifer. The landward shift of the interface is greater at depth. Groundwater withdrawal scenarios for all three del ta types, examined for RSLR of 1.18 m, indicate that pumping results in lo ss of potable water. The response of saline intrusion to pumping is much gre ater in case of the wavedominated deltaic aquifer. The freshwater-saltwater interface moves by 8 km for a RSLR of 1.18 m under groundwater withdrawal condi tions. Most delta plains host human population and this necessitates groundw ater usage. The model results suggest that the effect of groundwater with drawal surpasses that of RSLR in terms of salinity increase. It is worthwhile to mention that the rates of subs idence and pumping considered in this study are far less than in actua l cases. The pumping rate in the Ganges-Brahmaputra delta is 50 litre per capita per day (Michael and Voss, 2009) and the subsidence rate in the Nile delta ran ges from 0.1 m/yr to 0.25 m/yr (Stanley, 1999). The groundwater withdrawal is also limited to one area as opposed to widely distributed, as in most deltas. A lthough the model results underestimate the extent of saline intrusion that m ay essentially occur, they
43 provide an understanding of the anthropogenic effec t on natural coastal hydrologic systems.
44 Table .1 Hydraulic parameters used for SEAWAT mode l simulation of a river-dominated delta Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. Medium sand Fine sand Very Fine sand Sandy mud Clay Silt & Clay Kh (m/day) 40 20 10 3 3.00E-04 1.00e-4 Kv (m/day) 0.4 0.2 0.1 3.00E-02 3.00E-06 1.00e-6 n 0.4 0.4 0.4 0.4 0.4 0.4 S 0.03 0.03 0.03 0.03 0.03 0.03 Sy 0.03 0.03 0.03 0.03 0.03 0.03 a aa aLh (m) 100 100 100 100 100 100 a aa aLv (m) 0.1 0.1 0.1 0.1 0.1 0.1 a aa a T (m) 10 10 10 10 10 10
45 Table 3.2 Hydraulic parameters used for SEAWAT mode l simulation of a wave-dominated delta. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. Coarse sand Medium sand Fine sand Sandy mud Marine mud Lagoonal mud Alluvium Very fine sand, silt, clay Kh (m/day) 80 40 18 3 0.2 2 100 10 Kv (m/day) 0.08 0.04 0.018 0.003 0.0002 0.002 0.1 0.01 n 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 S 0.02 0.02 0.02 0.02 0.04 0.02 0.02 0.02 Sy 0.01 0.01 0.01 0.01 0.02 0.01 0.01 0.01 a aa aLh (m) 100 100 100 100 100 100 100 100 a aa aLv (m) 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 a aa a T (m) 10 10 10 10 10 10 10 10
46 Table 3.3 Hydraulic parameters used for SEAWAT mode l simulation of a tide-dominated delta. Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. Coarse sand Medium sand Fine sand Sand y mud Mud Silty mud Silt Cl ay Kh (m/ day ) 100 50 30 3 1.2 1.5 0.3 3.0 0E04 Kv (m/ day ) 1.00E-02 5.00E03 3.00E -03 3.00E -04 1.20E -04 1.50E -04 3.00E -05 3e08 n 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.6 S 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 Sy 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 a aa aLh (m) 100 100 100 100 100 100 100 100 a aa aLv (m) 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 a aa aT (m) 10 10 10 10 10 10 10 10
47 (a) (b) Figure 3.1 a) Plan view of top layer of model domai n of river-dominated delta showing different hydrol ogic units and boundary conditions. b) shows cross-sectional view of model with hydrologic units, layer thickness (m ) vertical distribution of hydrologic units and their thicknesses. (X and Y axes are column and row numbers Msmedium sand Fsfine sand V. Fsvery fine sand Smsandy mud C-clay ScSilt and clay Ms Sm No flow BC Riv BC CHB Well BC Sc Ms F s V. F s C 20 m 2 2 m 20 m 20 m 10 m A A A A 47
48 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34distance (m)d i s t a n c e ( m ) d i s t a n c e ( m ) d i s t a n c e ( m ) d i s t a n c e ( m )distance (m)(a) (b) (c) (d)freshwater freshwater freshwater freshwaterbase case RSLR = 0.68 m RSLR = 1.18 m RSLR =1.68 m Concentration (kg/m3) Figure 3.2 Comparison of salinity distributions in a river-dominated delta at a depth of 1m under initial and RSLR conditions. 0.25 kg/m3 isoc hlor marks the extent of potable water.
49 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 distance (m) freshwater freshwater freshwater freshwater(a)(b) (c) (d) d i s t a n c e ( m ) distance (m) distance (m) distance (m)distance (m) 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34(c) (d) (b) base case RSLR = 0.68 m RSLR = 1.18 m RSLR =1.68 m Concentration (kg/m3) Figure 3.3 Comparison of salinity distributions in a river-dominated delta at a depth of 7 m under initial and RSLR conditions. 0.25 kg/m3 isoc hlor marks the extent of potable water.
50 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000 20000 40000 60000 80000 100000 120000 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 (a) ( b) (c) ( d )depth = 1mfreshwater freshwater freshwater freshwaterdepth = 7 m depth = 17 m depth = 27 m d i s t a n c e ( m ) distance (m) distance (m) distance (m)distance (m) distance (m) distance (m) distance (m) Concentration (kg/m3) Figure 3.4 Comparison of salinity distributions in a river-dominated delta with steady-state groundwater withdrawal, shown at diffe rent depths for RSLR =1.18 m
51 Figure 3.5 a) Plan view of model 3 (wave-dominated delta) showing hydrologic units on top layer and bo undary conditions b) Cross-sectional view of model 3 showing ver tical distribution of hydrologic units and their th icknesses. X and Y axes are column and row numbers). Cscoarse sand Ms-medium sand Fs-fine sand S-silt Lm-lagoonal mud Mm-Marine mud A-alluvium Sm-sandy mud Vfssc-Very fine sand, silt, clay No flow BC CHB RIV BC Well BC A Cs Vfssc Lm Sm Ms A Fs Vfssc Ms A 50 m 200 m Ocean A A A A 53
52 Figure 3.6 a. Comparison of the extent of saltwater intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for base case. 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 distance (m) distance (m) 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 distance (m) distance (m) a) base case depth ocean
53 Figure 3.6 b Comparison of the extent of saltwater intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for RSLR =0.68 m 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 distance (m) distance (m) depth b) RSLR =0.68 m
54 distance (m) Figure 3.7 a. Comparison of the extent of the saltw ater intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for RSLR=1.18m 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 20000400006000080000100000120000140000160000 20000 40000 60000 80000 100000 120000 depth (m) a) RSLR =1.18 m distance (m)
55 distance (m) Figure 3.7 b Comparison of the extent of the saltwa ter intrusion in a wavedominated delta shown by 1kg/m3 isochlors at differ ent depths for RSLR =1.68 m 20000400006000080000100000120000140000160000 40000 60000 80000 100000 120000 b) RSLR =1.68 m depth (m) distance (m)
56 Figure 3.8 Comparison of the extent of saltwater i ntrusion in a wave-dominated delta under groundwater withdrawal conditions, sho wn by 1kg/m3 isochlors at different depths. 20000400006000080000100000120000140000160000 40000 60000 80000 100000 120000 distance (m) distance (m) depth
57 (a) (b) Figure 3.9 a) Plan view of model 3 (tide-dominated delta) showing hydrologic units on top layer and bo undary conditions b) Cross-sectional view of model 3 showing ver tical distribution of hydrologic units and their th icknesses. X and Y axes are column and row numbers). The diff erent colors represent hydraulic conductivity zones associated with the hydrologic units. Cscoarse sand Ms-medium sand Fs-fine sand S-silt M-mud C-clay Sm-sandy mud Slm silty mud Fs Slm s M Sm Cs C S Slm Fs Ms Ocean RIV BC CHB 250 m 100 m Well BC A A A A 59
58 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 b) RSLR = 0.68 m c) RSLR =1.18 m d) RSLR =1.68 m distance (m) distance (m) distance (m) distance (m) d i s t a n c e ( m ) distance (m) distance (m) 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 a) base casedistance (m) Concentration kg/m3 Figure 3.10 Comparison of salinity distributions in a tide-dominated delta at a depth of 0.75m 0.25 kg/m3 (red contour line) isochlor represents the extent of potable water. X and Y axes are in meters.
59 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 20000400006000080000100000120000140000160000 60000 80000 100000 120000 140000 160000 180000 200000 220000 d i s t a n c e ( m ) distance (m) distance (m)distance (m) distance (m) distance (m) a) depth = 0.75 m b) depth = 3.25 m c) depth = 7.75 m 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 Concentration kg/m3 Figure 3.11 Comparison of salinity distributions in a tide-dominated delta at different depths for RSLR =1.18m under conditions o f groundwater withdrawal. 0.25 kg/m3 (red contour line) isochlor represents t he extent of potable water. X and Y axes are in meters.
60 CHAPTER 4 POREWATER SALINITY CHANGES IN A PARTIALLY MIXED, MICROTIDAL ESTUARY DUE TO SEA LEVEL RISE Introduction Estuaries constitute an important part of the worl ds coastal landform distribution with a considerable portion of land-to -sea interaction. Estuaries represent dynamic and complex ecosystems that suppl y nutrients through river discharge and sediment loads, supporting productivi ty of estuarine fauna (Chapman and Wang, 2000). Several factors control t he nature of an estuary, e.g. river discharge, tidal currents, sediment load temperature and salinity variations, with the latter playing a key role. It can be anticipated that climate change could alter the salinity of estuaries. Since an estuary is not a steady-state flow regime, salinity changes occur within the wate r column as well as in the sediment interstitial waters (Chapman and Wang, 200 0). The density of the estuarine water is salinity driven, as density is l inearly related to salt concentration (Langevin, 2003). In case of rising sea level, salinity changes and associated density variations would effect the circ ulation and stratifications within the estuarine system (e.g. Gibson and Najjar, 2000)
61 A wide range of biota (e.g. mollusks, crustaceans) show limited salinity tolerance. Any alterations to the salinity gradient s within an estuary could result in unfavorable living conditions for these organism s. Understanding of the controls of natural and anthr opogenic impacts on estuaries is usually accomplished through modeling attempts. Among these, the most common approach is through hydrodynamic modeli ng that involves consideration of climate effects. Several studies o f various estuarine systems including twoand three-dimensional hydrodynamic m odels (e.g. Chesapeake Bay, Johnson et al., 1991; Hilton, 2008; Hudson Riv er estuary, Baumeiste and Manson, 2001; San Francisco estuary, Gross et al., 2007; Galveston bay, Wang, 1994; Loxahatchee river and estuary, Hu and Won, 20 07; Neusse river estuary, Wool et al., 2003) have been conducted. Other metho ds of predictive modeling applied to estuaries are statistical (e.g. Risley e t al., 1993; Gibson and Najjar, 2000; Hilton, 2008) and variable density models (e. g. submarine groundwater discharge in Biscayne Bay, Langevin, 2003). Exampl es of studies that have addressed the effects of climate change and SLR on estuaries include Thames river estuary, Dawson et al., 2005; Weser estuary, Grabbeman et al., 2001 and Chesapeake Bay, Hilton, 2008. In this chapter, variable-density flow models are used to simulate the effects of predicted SLR on the salinity distributi on in a partially mixed, microtidal (tidal range < 2 m) estuary. A simplistic generic model is considered. The results of this study can be extended to the response of ot her coastal plain estuarine systems to SLR.
62 Characteristics of estuaries An estuary can be defined as a geomorphic term (e. g. Fairbridge, 1980) or oceanographic term (e.g. Pritchard, 1952), based on its tidal range (e.g. Davis, 1964; Hayes, 1975) or its freshwater-saltwater circ ulation pattern (e.g. Dyer, 1972). Dalrymple et al, 1992 defines an estuary in geologic terms as the seaward portion of a drowned valley system which re ceives sediment from both fluvial and marine sources and which contains facie s influenced by tide, wave and fluvial processes. The estuary is considered to extend from the landward limit of tidal facies at its head to the seaward li mit of coastal facies at its mouth . Based on the mode of salinity distribution, estuar ies can be i) salt-wedge estuary where river flow is the dominating process, restricting mixing of fresh and salt water (e.g. saltwater wedge at the mouth of Mi ssissippi river), ii) partially mixed where fresh and saltwater mixing is greater d ue to tidal influences resulting in deposition of fine sediments in the up stream regions (e.g. Thames river, UK, James river, USA) and iii) well-mixed w here tidal influence dominates over river inflow with salinity variations in the h orizontal direction and vertically homogeneous (e.g. Firth of Forth, Scotland, Severn estuary, UK). However, there are instances of an estuary behaving as more than o ne type, depending on seasonal variations e.g. Vellar estuary, India, Col umbia river estuary, USA (Chapman and Wang, 2000). Vertically, a transgressive sequence is represente d by 1) estuarine fluvial facies overlain by 2) estuarine and followed by 3) estuarine marine facies (e.g.
63 Nichols and Biggs, 1985). Since the process of tran sgression plays an important role in estuarine sediment deposition, it can be im plied that estuaries are primarily controlled by changes in RSLR and it is o f importance to study the effects of future SLR on these environments. Groundwater flow models In the present study, three-dimensional groundwate r flow models were simulated using the computer code SEAWAT (Guo and L angevin). Models were designed and executed using the software Groundwate r Vistas (GWV version 5.22). Hybrid Method of Characteristics technique (HMOC) w as chosen to solve the flow and transport equations as the method has grea ter accuracy and computational advantage over other methods (Zheng a nd Wang, 1998) Each model was run using a number of particles equal to 64 in each cell. Model description Model design A three-dimensional finite difference grid was des igned to simulate flow in the estuary. The model consists of 10 layers, 100 r ows and 26 columns. Uniform cell sizes of 2000 m x 2000 m were used. Time discretization For the SEAWAT models, Courant number was specifie d to 1 allowing maximum flow and transport over a distance equal to one grid spacing.
64 Boundary conditions In order to simulate the estuary, General Head Bou ndary or GHB condition of MODFLOW (McDonald and Harbough, 1998) were implemented. The GHB is a head dependent boundary condition that allows the model to calculate freshwater flux inflowing through the GHB cells into the bay (e.g. Langevin, 2003). Saline bay cells were represented by Constant Head Boundary (CHB) cells, where head and concentration values we re specified. The model domain was terminated on all edges and at the botto m by no-flow boundary conditions. Figure 4.1 shows the model grid and bou ndary conditions. Hydrologic units The estuarine basin is characterized by a crystall ine basement, immediately overlain by a high conductivity gravel bed. Fine sand deposits occur around the channel boundary and at its mouth, while silty clay deposits fill up the main channel. Sporadic deposits of clay are present in the upstream of the estuary (figure 4.1). The sediment distribution use d in this model is based on that in the Chesapeake Bay estuary (e.g. Hobbs, 1994). Hydraulic parameters Table 4.1 shows the hydraulic parameters (hydrauli c conductivity, porosity, specific storage, dispersivity etc) used in the model. The hydraulic conductivity values are based on those of unconsoli dated sediments and crystalline rocks as given in Schwartz and Zhang (2 002). Higher porosity values were used for clay units as initial model runs with uniform porosity resulted in high head values over clay deposits. Increase in st orage eliminates the problem
65 of excess water table elevations. High dispersivity values are also used, assuming dispersivity as scale dependent (Fetter, 1 988). An effective recharge of 8E-05 m/day was used for the base case scenario. Recharge was increased in subsequent sea level rise scenarios to 1E4m/day. An equal evapotranspiration rate of 1E-04 m /day was applied over the model domain. In estuarine systems, evaporation rat es are often of high magnitude (e.g. Sumner and Belaineh, 2005). All hyd raulic parameter values listed were obtained after several modifications to reduce the mass balance errors. Mass balance errors range from 0.02 to 0.05 % for the different scenarios. Model assumptions The estuarine model presented here is of a simplis tic, generic type. SLR would cause a widening of the estuary, landward mig ration, loss of saltmarshes i.e. change in vegetation pattern. These factors ha ve not been accounted for in this model. The rate of SLR was assumed to be const ant over a period of 10 years and was increased stepwise. The effects of la nd subsidence due to groundwater abstractions in nearby areas, climatic effects such as temperature variations and tidal oscillations were not consider ed. As a result of these simplifications, the simulated saltwater intrusion from SLR is likely to be the minimum intrusion that can be expected. Model simulations Base case scenario A base case scenario (Model 1) was generated with a n assumed distribution of salinity. The assumption was based on salinity distribution in the
66 Chesapeake Bay estuary obtained from Hydrodynamic E utrophication model of Virginia Marine Sciences. Salinity varied from fres h to brackish to saline in successive layers. The salinity concentration of se awater in the bay area was set to 35 kg/m3 in the CHB cells. Head values in these cells were set to zero m. Head and concentration in the GHB cells were set to 0 m and 0 kg/m3 respectively. The model was run in steady-state for 5000 days, until a reasonable head distribution was attained. SLR scenarios were operated with the base case head and salinity distributions as initial conditio ns. Sea level rise scenarios Three SLR scenarios were considered with a net rise of 0.5 m, 1 m and 1.5 m within a time period of 90 years. Head values in the CHB cells representing SLR were increased accordingly for each case. The m odels were simulated in steady-state where in each scenario, a model was fi rst run for 10 years. In each of the following time steps of 10 years, concentrat ion and head values from the end of the preceding simulated 10-year period were used as initial conditions. In the first scenario (Model 2a), head values in t he sea boundary were changed to 0.5 m over 90 years. In the second and t hird scenarios represented by Model 2b and 2c, the net rise in sea level is 1 m and 1.5 m, respectively, again over 90 years. Results and discussions Model results are shown in figures 4.2 and 4.3. In each figure, salinity distributions of porewater or sediment interstitial water, under different SLR
67 scenarios, are compared. In figure 4.2, at a depth of 4.5 m, in the base case scenario, the porewater salinity at the mouth of th e estuary has increased from the base case. In the central part of the estuary ( between (50 km and 90 km along the y-axis), porewater salinity values range between 8-10 kg/m3. Elsewhere, salt concentration is lower ( 6kg/m3) with fresher water in the upstream region near the mouth of the river in the north. For a sea level rise scenario 1 (Model 2a), when S LR = 0.5 m, porewater salinity increases along the sides of the estuarine channel to ~ 8 kg/m3 while slight decrease in salinity is noted in the central part. For SLR =1 m (Model 2b), increase in porewater sal inities are prominent along the sides of the channel. Saline intrusion oc curs over a wider area along the eastern side of the channel, corresponding to t he relatively high conductivity sand deposits. The intrusion extends upstream up to ~ 140 km along the y-axis. A slight increase in salinity occurs in the central part of the channel. For SLR =1.5 m (Model 2c), salinity increases as i n Model 2b, with saline intrusion advancing further upstream and reducing t he availability of fresher and brackish interstitial water. The above results suggest that porewater salinity variations in an estuary occur preferentially along zones of higher hydrauli c conductivity. It has been shown by laboratory experiments and field measureme nts, that salinity changes in estuaries are controlled by sediment types (Chap man and Wang, 2000).
68 Conclusions Estuarine circulation, stratification and prevailin g ecosystem are controlled by salinity variations. Variable-density flow model s of a partially mixed estuary indicate that with a climate change related SLR, sa linity distribution in the porewater (i.e. sediment interstitial water) would change extensively, with maximum increase occurring for a SLR of 1.5 m. Salinity variations are primarily controlled by se diment type. Sand deposits are more responsive to SLR and saline water movemen t than silty clay deposits along the central part of the estuary. The upstream advancement of salinity with rise in sea level would reduce fresh and brackish water (unless river inflows incr ease) and thus spatially restrict fresh/brackish water biota in the estuary. Increase in sediment salinity could also affect benthic organisms that derive their nutrient s from bottom-feeding.
69 Table 4.1 Hydraulic parameters used for SEAWAT mode l simulation of partially mixed estuary. Gravel Fine sand Silty clay Clay Basement (crystalline) Kh (m/day) 500 17 0.08 0.0004 5 Kv (m/day) 50 0.17 0.008 4.00E-05 0.5 n 0.4 0.4 0.4 0.6 0.4 S 0.2 0.2 0.2 0.4 0.2 Sy 0.2 0.2 0.2 0.4 0.2 a aa aLh (m) 50 50 50 50 50 a aa aLv (m) 0.5 0.5 0.5 0.5 0.5 a aa aT (m) 5 5 5 5 5 Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective porosity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity.
70 Figure 4.1 Plan and cross-sectional view of model w ith hydrologic units, boundary conditions and layer thicknesses. Colored zones represent different hydraulic conduct ivity values of the hydrologic units. Fs Sc C C basement GHB CHB 4m 8m 8m 8m No-flow G-gravel Fs-Fine sand C-Clay Sc-Silty clay G BAY 74
71 2000040000 40000 60000 80000 100000 120000 140000 160000 180000 2000040000 40000 60000 80000 100000 120000 140000 160000 180000 2000040000 40000 60000 80000 100000 120000 140000 160000 180000 2000040000 40000 60000 80000 100000 120000 140000 160000 180000 distance (m) d i s t a n c e ( m ) 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 Concentration (kg/m3) Figure 4.2 Salinity changes in model estuary at a depth of 4.5 m (e stuary bottom) under different conditions of sea level ris e. X and Y axes represent distance in meters.
72 CHAPTER 5 MODELING SEA LEVEL RISE INDUCED SALTWATER INTRUSION IN A MIGRATING BARRIER ISLAND Introduction Barrier islands are elongate coastal landforms, for med during the Holocene period, often found to occur in chains, pa rallel to the shore and are separated from the mainland by backbarrier lagoons. Nearly 15% of the worlds coastlines are composed of barrier islands (Davis a nd Fitzgerald, 2004). Along the east coast of the United States, 3100 km of coa stline is occupied by barrier islands, while 1600 km of the Gulf of Mexico coast consists of barriers. Barrier islands have worldwide distribution including coast s of the North Sea in Europe, Eastern Siberia, Alaska, South America, India, and Australia. Because of their proximity to the sea, barrier is lands have become valued tourist destinations and residential properties. In creased anthropogenic usage of barrier islands has increased the demand for potabl e water, often obtained from groundwater. The natural source of groundwater in t hese islands is the freshwater lens recharged by precipitation. As most coastal aquifers are threatened by salinization due to i) excess groundw ater withdrawal and ii) sealevel rise, barrier island aquifers are no exceptio ns. SLR can affect coastal barrier islands in two ways :
73 i) geomorphologically, by changing the shape of th e islands and ii) hydrogeologically, by initiating saltwater intrusio n into the island aquifer, thereby reducing freshwater resources. As the shape of the freshwater lens in an island aquifer is influenced by the geometry of the island itself, any morphological change would also alter the shape of the lens. SLR will create transgressive sequences of deposits, resulting in heterogeneous zones of hydraulic conductivity This will influence groundwater flow and mass tran sport mechanisms. It is thus of interest to investigate t he roles these coupled responses play in controlling saltwater intrusion i n a barrier island setting. The objective of this chapter is to assess the availabi lity of freshwater resource in barrier islands that undergo landward migration as sea level rises. Geology of barrier islands Barrier islands are siliciclastic deposits, formed by the actions of waves, tides and wind. Their length can vary from a few hu ndred meters to many kilometers. A barrier system has the following geom orphic components: i) mainland, ii) backbarrier lagoon, iii) inlet and inlet deltas, iv) barrier islands, v) barrier platform and vi) shoreface ( e.g. Oertel, 1 985). The dynamic interaction between these six environments influences the evolu tion and stability of each one of them. Barrier islands are also classified a s i) wave-dominated and ii) mixed-energy based on the predominant depositional process (e.g. Davis, 1985).
74 Migration Growth and stability of barrier islands are determ ined by factors such as rate of sediment supply, rate of sea-level rise, fr equency of storms and geologic setting of the mainland. Barrier islands can be i ) prograding, ii) retrograding and iii) aggrading on the basis of their mode of adjust ment to the controlling factors. These islands grow seaward or prograde under conditions of abundant sediment supply or slow rise in sea level (Davis, 1985). Pro gradation takes place through the development of beach ridges that are accreted o ver time. Barrier islands move landward or retrograde when there is a lack of sufficient sand supply relative to rising sea level. Retrograding barriers are reduced in width because of erosion on the seaward margin. The island migrates landward by the deposition of storm overwash sediments over lagoons and marshe s in the backbarrier areas (figure 5.1). Continuing landward retreat of these islands result in exposed peat layers at the beach front that represent older mars h/lagoon environments. Thus retrograding barrier islands have a characteristic stratigraphy. The freshwatersaltwater interface in retrograding islands can exi st within hydrostratigraphic units that are different from those in prograding or aggr ading islands. When sediment supply rate is adequate to compensat e for the effect of SLR, barrier islands can remain stable in their are a of formation forming aggrading barrier islands. Aggrading islands are no t common as their stability is very sensitive to the rate of sediment supply.
75 Hydrogeology of barrier islands One of the primary factors controlling the hydroge ology of barrier islands is the barrier stratigraphy. The various deposition al environments in a barrier system contribute to the evolution of the island an d their signatures are preserved in the stratigraphic record of the island Spatial variations (horizontal and vertical) of the properties (e.g. grain size, p acking, porosity) of these depositional units create heterogeneity in hydrostr atigraphic units. Thus the resulting island aquifer is heterogeneous and often anisotropic. The source of freshwater in a barrier island groundwater system i s recharge from precipitation while sinks include evapotranspiration, oceanward d ischarge of freshwater and withdrawal through pumping. Factors controlling freshwater lens morphology In siliciclastic barrier islands, it has been obse rved that island stratigraphy can control the freshwater lens configuration (e.g. Harries, 1967; Fetter, 1972; Kidd and Planert, 1985; Simmons, 1986; Collins and Easley, 1999; Anderson et al., 2000). However, the freshwater lenses in barri er islands do not conform to a particular configuration but show variability in th eir shapes, often deviating from the traditional symmetric, Dupuit-Ghyben-Herzberg l ens (Schneider and Kruse, 2003). his can be attributed to the heterogeneity of the island stratigraphy, defined by changes in hydraulic conductivities (Vac her, 1988). Groundwater flow models of progarding, retrograding and aggarding ba rrier islands show hydraulic permeability variations in different facies affect the shape of the freshwater lens
76 and the characteristics of the transition zone (Kug ler, 1998). Recent studies in Padre Island, Texas, suggest that shallow lens conf iguration is determined by the variation in sedimentary deposits (e.g Weber et al. 2008). Although isolated from the main barrier islands, back-barriers, a componen t of the barrier system, can also develop freshwater aquifers. Hodgkinson et al. (2008) described the stratigraphic controls on the characterization of a two-aquifer system in a back barrier island in southeast Queensland, Australia. Other factors influencing freshwater lens morpholo gy in barrier islands include the distribution patterns of recharge, vege tation and elevation e.g. as in Dog Island, Florida (Schneider and Kruse, 2003) and anthropogenic effects of pumping as in St. George island (Ruppel, 1999; Schn eider and Kruse, 2003). Aquifers in both these islands show a skewed asymme tric shape. Freshwater lens configuration can also be defined by the dimensions of the island. Nielsen (1999) found that barrier isla nds with width less than a kilometer are subjected to an additional height to the watertable under the condition of wave runup on the ocean side. This ext ra height generates a landward flow of salty groundwater and reduces the vertical extent of the freshwater lens i.e. the resulting freshwater lens is shallow and skewed in shape. Generic model of a barrier island Introduction Although several modeling efforts of freshwate r resources in barrier islands have been made, very few modeling studies h ave addressed the problem
77 of saltwater intrusion due to sea-level rise (e.g. Kana et al., 1984, Oude Essink, 1999 and Kooi and Groen, 2000) In a variable-density flow model, Masterson (2007) simulated the effect of sea-level rise at a rate of 2.65 mm/yr from 20052050 in a coastal aquifer (sand, gravel, silt and c lay) similar to that in Cape Cod, Massachusetts. His results suggest a net decrease i n water table elevations due to the influence of streamflow as well as a reducti on in the size of the freshwater lens. Under the condition of sea-level rise, the pr oximity of a stream greatly influenced the depth of the freshwater lens, result ing in 22% to 31% decrease in thickness close to the stream but only 2% reduction in places away from the stream. Numerical study of saltwater intrusion in a sand and gravel aquifer of Shelter Island, New York (Rozell, 2007), due to sea -level rise indicates that IPCC (2007) predicted scenario of 15% increased precipit ation with 0.18 m SLR by 2099 is most favorable with a seaward movement of t he freshwater-saltwater interface by an average of 23 m. Conversely, a decr ease in precipitation by 2% and a rise in sea level by 0.6 m would cause a land ward movement of the interface by an average of 16 m. Previous work by Schneider and Kruse (2003) takes into account the effect of island erosion on the configuration of th e freshwater lens. However, barrier island migration has not been included in g roundwater modeling studies to date. The present study involves simulations of the effect of SLR on the freshwater lens of a spatially and temporally dynam ic barrier island. The conceptual model is based on the geological an d hydrogeological information available from previous studies of Hatt eras Island on the Atlantic
78 coast of North Carolina, USA (Heath, 1988; Anderson 1998, 2000). Some idea of the predicted migration rate of this island can be attained from the modeling study for island migration in the adjacent Outer Ba nks region (Moore, 2007). Descriptions of Hatteras Island in the following se ctions are based on the works of Heath, (1988) and Anderson (1998). The objective of this modeling study is to assess the future of freshwater resources in barrier islands similar to the coastal aquifer in Hatteras Island and is not directed towards saltwater intrusion problems o f a specific site. The reason for using geologic and hydrogeolgic data from Hatte ras Island is simply to create a generic model close to reality. A separate case s tudy of Hatteras Island would require hydrological field data acquisition to unde rstand the present state of the freshwater lens morphology, withdrawal techniques a nd present salinity distribution patterns. Such models will also need t o be calibrated against field observations. Geologic setting Hatteras Island, a 12 km long barrier island, cons titutes the easternmost island of the Outer Banks of North Carolina (figure 5.2). The island exhibits a north-south trend until the Buxton area, where it f ollows an east-west trend creating an L-shaped island. On the bay side of the island is the Pamlico Sound estuary. The island widens to approximately 3 km. H atteras Island hosts the only existing maritime forest on the North Carolina coas t in the Buxton Woods area providing shelter to various flora and fauna. Hatt eras Island also serves as a residential and recreational site. In response to i ncreasing tourism, the island
79 now supports motels, campgrounds and thousands of r ental homes for vacationers. Evolution of the Outer Banks, including Hatter as Island, is well explained by the theory of formation of perched barriers on an existing headland (Riggs et al., 1995). Headlands are defined as geologic subst rates consisting of indurated or semi-indurated sediments of Pleistocene or pre-P leistocene age. According to Riggs et al., 1995, Holocene sediments, in both tra nsgressive and regressive sequences, were deposited on these features, thus r esulting in barrier sands perched on headlands. The sand deficient northern p art of Cape Hatteras represents a transgressive sequence where the islan d is migrating west, exposing peat and mud deposits. The area west of Ca pe Hatteras consists of regressive shoreface, where sediment supply is abun dant and beach ridges are formed. Recent studies by Anderson, (1998) indicate that part of the Pleistocene framework was eroded at the transgressive shoreface before further progradation occurred. It is thus inferred that a heterogeneous shoreface exists that can influence local permeability and affect the size of the freshwater-saltwater mixing zone. Hatteras Island hydrogeology Heterogeneity in island stratigraphy controls the h ydrogeology of the aquifer system in Hatteras Island. The Buxton Woods aquifer, the only groundwater resource for Hatteras Island, consists of an Upper Buxton Woods surficial aquifer and a Lower Buxton Woods confined aquifer. These two reservoirs are separated by a semi-confining unit. The hydrostratigraphic units
80 forming the Buxton Woods are i) the upper 12 m cons isting of medium to coarse grained highly permeable unit which corresponds to the Upper Buxton Woods aquifer; ii) the semiconfining unit, 3 m in thickne ss acts as a semi-permeable bed, formed of fine sands and silts; iii) the Lowe r Buxton Woods aquifer, extending to a maximum depth of 24.5 m, formed of medium to coarse grained sands with shell fragments and iv) the lowermost un it, composed of 13 m of silty to clayey sand which is described to be slightly pe rmeable. The subsurface heterogeneities are considered responsible for dela yed responses in the yields for pumping tests from the upper and lower aquifers (Anderson, 2000). Permeabilities, horizontal and vertical, for both t he Upper and Lower Buxton Woods units, as estimated by aquifer testing are 2.41E-11 m2 and 2.1 E1-12 m2 respectively. Sensitivity analysis for permeabilit y indicate that the water-table elevations are very sensitive to permea bility changes in the upper aquifer but less sensitive to permeability changes in the lower aquifer. The salinity distribution is negligibly affected (Ander son, 1998). Both water-table elevation and salinity distribution are insensitive to changes in the permeability of the semi-confining unit present in between the uppe r and lower units. Calibrated horizontal and vertical permeabilities are 1.0E-13 m2 and 1.0E-14 m2 respectively. The presence of a buried wetland infl uences the water table elevation in the central part of the island. Freshwater-saltwater interface In Hatteras Island the contact between freshwater and saltwater is represented by a mixing zone/ zone of diffusion. Th e isochlors are described to
81 be wavy or interfingering caused by preferential fl ow and mixing of fresh and salt water within the heterogeneous sedimentary units of the aquifer. It is also suggested that since the Cape Hatteras part of the island is relatively younger (350 years old according to Fisher), the sediments deposited in this area were saturated with saltwater which is still being flush ed out by freshwater movement. The depth to saltwater is 25 times the height of the water table above sea level i.e. the effective Ghyben-Herzberg ratio is 1:25. T his was estimated by the USGS (1980) using 250 mg/l chloride content to indicate the freshwater-saltwater contact. Factors affecting freshwater resource in Hatteras I sland Recharge Heath estimated two recharge rates for the Cape Ha tteras area, based on previous studies by the USGS. The swales and ridges area receive a low recharge of 0.01 m/yr (7% of annual precipitation) while the open areas receive 0.5 m/yr (35% of annual precipitation). Hatteras Island shows a distinctive vegetation patt ern that controls evapotranspiration rates and therefore recharge to the water table. The thick forests of northern Hatteras Island require large v olumes of water during spring and summer. Interception is high in the forested ar ea as well. The southern part contains thinning forests and increasing number of shrubs that merge into bush and grass covered land and ultimately into dunes. T he plants in this area have lower water requirements, leading to an increase in recharge rates in the
82 southern part. An asymmetric water table is generat ed with higher elevation corresponding to the southern light vegetation cove r (Anderson, 2000). Groundwater withdrawal By 1988, a total of 40 wells were installed with an average depth of 12 m to provide potable water to Hatteras, Frisco and Bu xton areas by the Cape Hatteras Water Association (CHWA). As reported by H eath (1988) the rapid growth in population created further water supply n eeds and an estimate of 4.5 mgd of water was anticipated for the service area b y the year 2000. In 1997, the CHWA merged with the Dare County Water System which now provides water to the Hatteras Island area. Dare County established a Cape Hatteras Water Plant in 2000 which has a capacity of 2.0 MGD and is expa ndable to 3.0 MGD with increasing water demand. No significant change in salinity was observed in w ells pumping in the Frisco area from 1968 to 1988 (Heath, 1988). Furthe r information on the salinity distribution in Hatteras Island is not available. Andersons numerical modeling of a proposed well-fi eld by Heath (1990) shows saltwater intrusion in the northern part of t he island near Palmico Sound resulting from pumping from upper aquifer and lower aquifer separately at an average maximum pumping rate of 30 gpm under condit ions of low recharge. Surface drainage Some surface drainage systems operate in the islan d in order to eliminate excess surface waters and control mosquito infestat ions. Peters Ditch drains water in the northern part and culverts from Jennet te Sedge transport water to
83 the Palmico Sound. The size of the freshwater lens has been reduced by surface drainage (Heath, 1988). Tidal and storm effects Since barrier islands are composed of unconsolidat ed sands and shell fragments that act as low permeability units relati ve to high permeability carbonates in atoll islands, the influence of low a mplitude tides (e.g. 15 cm on Sound side of Hatteras Island) on barrier islands a re negligible (Anderson, 1998; Anderson and Lauer, 2008). However, inundation by s eawater during storm events introduces saline water instantaneously into the surficial aquifer which takes from weeks to months to recover (Anderson and Lauer. 2008). Groundwater flow models All groundwater flow models were simulated using t he computer code SEAWAT (described in chapter 2) that incorporates t he flow modeling codes of MODFLOW and those of transport in MT3D. Models were designed and executed using the software Groundwater Vistas (GWV version 5.22). The simulations used Hybrid Method of Characteristics t echnique (HMOC) for solving the transport equations. HMOC combines the methods of MOC (Method of Characteristics) and MMOC (Mixed Method of Characte ristics) to automatically adapt a solution technique depending on the charact eristics of the concentration field (Zheng and Wang, 1999) Solutions derived by HMOC have greater accuracy than other methods and insignificant numer ical dispersions over a large range of Peclet number (Pe= D l / a where D l is a nodal spacing and a is
84 dispersivity). HMOC also requires less number of p articles to simulate the transport mechanism which saves computational time. Each model was run using a maximum number of particles equal to 64 per cell. Model description Model design A 2D finite difference grid was designed to represe nt a cross-sectional profile of the island. The model consists of a sin gle row with 186 columns and a total of 2046 active cells. Cell dimensions were sp ecified by D x min = 25 m and D xmax =50 m, while D y min and D ymax were both set to 100 m. Along the land-sea contact, smaller cells 25 m wide were used to accou nt for sharp changes in concentration gradient at the boundary. The model w as simulated with 11 layers since a minimum of 10 layers is required for a vari able density model to allow sufficient vertical resolution (Guo and Langevin, 2 002 ). Hydrologic units The Upper Buxton Woods Aquifer is represented by t he top five layers, each with a thickness of 3 m. The semi-confining un it dips toward the ocean side. The Lower Buxton Woods Aquifer consists of layers 8 9 and 10 and is 9.5 m thick. The last layer (layer 11) forms the lower co nfining unit and extends to a depth of 40.5 m. Figure 5.3b shows the distribution of the hydrologic units. Time discretization For an accurate solution of the variable density f low problem, a specific time stepping is required which is determined by th e Courant number. Theoretically, the Courant number ,C = vDt / D l,
85 where v is velocity, D t is transport time step interval and D l is nodal grid spacing, should be less than or equal to one, i.e. D t < D l/v (Anderson and Woessner, 1992). For the SEAWAT models, the Courant number wa s specified as 1, allowing maximum flow and transport over a distance equal to one grid spacing ( If C=1, then v D t = D l ). Boundary conditions A numerical model is defined by governing equation s and boundary conditions. In groundwater flow, the flow equations solve for the hydraulic head which is a dependent variable. Boundary conditions are implemented to attain a correct solution of the problem by specifying the h ead or its derivatives, resulting in three types of boundary conditions such as i) s pecified head boundaries, ii) specified flux boundaries and iii) mixed or head-de pendant boundaries (Reilly and Harbaugh, 2004). In a variable-density flow pro blem, concentration is also required to be specified along the solution domain. In all the following models, the sea level is repr esented by a Constant Head Boundary (CHB) condition where the head of the sea level and the chloride concentration in seawater was set to zero meters an d 35 kg/m3 respectively. The watertable forms a specified flux boundary which re ceives water through recharge for precipitation. Figure 5.3 shows a sche matic of the model grid, hydrogeology and boundary conditions implemented.
86 Hydraulic properties The hydraulic parameters, hydraulic conductivity, porosity, specific storage, dispersivity used for model generation are mostly based on values used by Anderson (2000) in his SUTRA model. Two differen t recharge rates are applied for the initial conditions. A rate of 0.000 3 m/day over vegetated areas and 0.001 m/day over dune sands were used. Table 5.1 g ives the hydraulic parameters used to generate model 1. Model assumptions and limitations The 2D model used for this study is a general repre sentation that does not include surface drainage, well fields and has a sim plified spatial distribution of recharge pattern. The influence of drainage and tid al effects were examined by Anderson (1998), and the numerical results indicate that their effects are minimal on the salinity distribution in Buxton Woods Aquife r. Sea-level changes are assumed to be steady over each interval of 10 years The model grid needs to be large enough to accommod ate landward migration in all simulations. GWV requires that the same grid specifications to be maintained for all models in order to incorporate h ead and concentration data from previous models into subsequent ones. This req uirement has limited this study to a 2D model instead of 3D that is computati onally demanding. A 3D model was initially designed with the same grid spa cing including drainage systems and well field. This model was unable to co nverge using a 3 GB RAM computer when boundary conditions were changed. As Hatteras Island is used
87 as an example of a barrier island, not including pu mping emphasizes the effects of sea level rise on the lens configuration. Model simulations Base case scenario Initial salinity conditions were set based on Heat hs report. The top layers included the freshwater lens with zero kg/m3 salt concentration to represent the Upper Buxton Woods aquifer. A head value of 0 meter s was set on the ocean and sound/lagoon side while a chloride concentratio n of 35 kg/m3 was applied to represent seawater concentrations. This model, desi gnated as Model 1 was for 7300 days (20 years) until the system reached equil ibrium. Head and salinity distributions generated in Model 1 served as the in itial conditions for the following models. Sea level rise scenarios In the following simulations, boundary conditions w ere changed as sea level is predicted to rise and the barrier island m igrates. Each scenario considers a rise in sea level by 0.5, 1.0 and 1.5 m by the ye ar 2100. Each model is run for 90 years assuming a period from 2010-2100. Every mo del is run with a time step of 10 years which means CHB head values are changed every 10 years and the island position is shifted landward according to pr edicted migration rates (Moore, 2007). Model simulations of barrier islands of the Outer Banks of North Carolina indicate change in migration rates by 15-20 m/yr fo r rise of sea-level by 1.4-1.9 m above mean sea-level by 2100 (Moore et al, 2007). T he migration is simulated by adding CHB cells representing seawater on the ocean side and removing CHB
88 cells on sound side, revealing land surface. Sea-le vel rise is represented by increasing the head value in the CHB cells correspo nding to seawater. The sealevel is assumed to remain steady over each interva l of 10 years. Recharge rates have been increased by 10 % for the models for the sea-level rise scenarios, assuming an increase in precipitation in humid regi ons in the future. Recharge distribution was readjusted to new positions on san d covered areas and reduced vegetated zones. A single transient model to simula te the entire 90 years in one run was not possible as the island shifts position in each 10 year run. Scenario I: 0.5 meters rise by 2100 The first scenario (Model 2a ) simulates a gradually migrating barrier island wi th a migration rate of 5 m/yr. At the end of 90 years i. e. by 2100, the net migration of the island is by 450 m and sea-level rise is 0.5 m. Scenario II: 1 meters rise by 2100 This scenario (Model 2b ) represents a net migration of the island by 900 m at a rate of 10 m/yr and a net rise in sea level by 1 m. Scenario III: 1.5 meters rise by 2100 In this scenario, (Model 2c ) the island had shifted landward by 1350 m at the rate of 15 m/yr while sea level rose to 1.5 m in 90 years. Additional scenarios Two additional scenarios were generated to underst and the effect of climatic change such as reduced recharge and anthro pogenic effect such as groundwater withdrawal by pumping. In practice, ma ny islands aquifers, including Hatteras Island, St George Island and She lter Island, undergo
89 groundwater withdrawal at greater rates than in the simulated model. Hatteras Island has already encountered saltwater intrusion in the populated lagoon side and presently has a reverse osmosis desalination fa cility. Reduced recharge This model (Model 3a ) is similar to Model 2b (1 m rise in sea-level) ex cept the recharge rate has been decreased to 0.000003 m/ day and 0.0001 m/day i.e. by 10% from base case. All other parameters were ke pt same as in Model 2b. Withdrawal In Model 3b a well boundary condition was set to withdraw wat er from the top layer of Upper Buxton aquifer at a rate of 100 m3/day. This condition was applied to Model 2b. All other parameters remained unchanged. Results and discussion The freshwater lens obtained from Model 1 is charac terized by a slightly asymmetric shape with irregular freshwater-saltwate r transition zone (figure 5.4) as described in Heaths report (1988). The 1 kg/m3 isochlor is considered to represent the limit of freshwater. The model resul ts indicate that the freshwater lens extends to a depth of approximately 18 m, near ly coinciding with the bottom of the semi-confining bed. The transition zone, def ined here by the 1 kg/m3 and 10 kg/m3 isochlors, is wider along the sound side than on t he ocean side. Model results of the sea-level rise scenarios are g iven in figures 5.5 to 5.7. Under the conditions of sea-level rise by 0.5 meter s in 90 years, 10% increased recharge and a migration rate of 5 m/year, a deeper freshwater lens forms (figure
90 5.5). The resulting depth of the freshwater lens is ~25 m. The transition zone is wider on the sound side which could be possible due to the retarded flow within the low conductivity, semi-confining unit over whic h the island has migrated. Generally, the presence of a low conductivity unit produces a narrow transition zone due to retarded saltwater influx. The width an d depth to the transition zone is controlled by the location of the low conductivi ty unit, especially in retrograding barrier islands (Kugler, 1998). In this case, the migration of the island position s itself over the semiconfining layer that is now at a shallow depth. Ini tially, the lagoon side with the underlying shallow semi-confining unit was saturate d with saline water. Due to the low conductivity of this unit, the flushing of saline water by recharge is a slow process, allowing greater mixing of saltwater with percolating freshwater and creating a wider transition zone on the lagoon/soun d side. Salinity distributions at different timesteps indicate that the movement of t he 1 kg/m3 isochlor gradually slows down from an initial fast landward movement i .e. lateral saltwater intrusion lessens over time. Figure 5.6 shows the results from Model 2b. As the island migrates at a rate of 10 m/yr with a sea-level rise of 1 m in 90 years, the freshwater lens development continues, reaching a depth of ~22 m. T he movement of the freshwater-saltwater boundary slows down from 60 ye ars onwards and nearly coincides with the shoreline, indicating that islan d migration rate still predominates over sea-level rise induced transgress ion rate.
91 In case of Model 2c after 90 years, the freshwater lens has reached a depth of ~21 m (figure 5.7). Transition zone on sou nd side remains wider while that on the ocean side becomes narrow. Slight taper ing of the freshwater lens is noted along the sound side within the upper aquifer At depth, the transition zone becomes thicker. The 1 kg/m3 isochlor moves landward by 150 m with respect to the shoreline at the end of 90 years. SLR induces an increase in heads over the island (F igure 5.8a). The heads include the combined effect of SLR, increased recharge and island migration. Maximum head of 4 m is noted for Model 2 c, followed by 3.5 m in Model 2b and ~ 2 m in Model 2a. The heads produced in Models 2b and 2c are relatively higher than in Model 2a. This differenc e can be explained by the fact that in Model 2b and 2c, the island has shifted to a new position where the low conductivity semi-confining unit delays the infiltr ation process for the given time interval and creates a high head over the island. T he overall conductivity of the upper aquifer decreases in the new position due to the presence of the shallow semi-confining unit. As mentioned earlier (under Ha tteras Island hydrogeology), head values are very sensitive to permeability chan ges within the upper aquifer. To understand the effect of increased recharge and island migration on head distribution, the model head values are compar ed (figure 5.8b). This is examined for a sea-level rise scenario of 1 m rise in 90 years under different conditions as stated. Head values are highest for t he case where recharge is increased by 10 % and the island migrates and chang es the least when the recharge remains the same as in base case and the i sland remains static. An
92 intermediate increase is attained when the recharge remains constant but the island migrates. It is clear from this comparison t hat island migration alone could cause a change in head distribution depending on th e subsurface stratigraphy over which the island moves. Figure 5.9 shows the result from Model 3a which con siders a case of reduced recharge (decreased by 10% from base case) in a migrating barrier island with 1 m rise in sea level in 90 years (i.e. Model 2b). The maximum extent of the freshwater lens is ~ 12 m which means that a reduction in depth by 6 m from base case and 13 m from sea level scenario Mod el 2b has occurred. This result indicates that a decrease in precipitation a nd thereby recharge is least favorable for maintaining freshwater resource in a migrating barrier island. Pumping results in saline intrusion by upconing lea ding to a distorted freshwater lens shape with gradually decreasing dep th e.g. to ~10 m in 60 years (Figure 5.10). Although the depth is restored to 20 m by the next 30 years, the position of the freshwater lens changes with a redu ced depth on the ocean side. Continued pumping in this position would result in further upconing of saline water diminishing the lens size. A comparison of head distribution (figure 5.11) und er conditions of increased recharge, decreased recharge and pumping withdrawal shows that reduction in recharge is the most influential facto r in decreasing the size of the freshwater lens. Effect of wihdrawal is also eviden t by the formation of a trough in the head profile.
93 The results obtained in this study are in agreement with the recent variable-density groundwater flow models of Shelter Island (Rozell, 2007) where an increased recharge with least sea-level rise see med the most favorable scenario with increased freshwater lens volume whil e a reduced recharge caused landward movement of the fresh-salt interfac e, leading to a reduced lens size. Conclusions The numerical simulation results indicate that if a migrating barrier island survives a 1.5 m rise in sea level by 2100, its fre shwater resource will still be available provided there is also an increase in rec harge. This suggests that in the case of predicted climate change scenarios, increas ed precipitation in tropical regions could possibly aid in freshwater lens reten tion in sandy islands. However, in islands receiving low recharge, a reduction will drastically decrease the size of the freshwater lens as seen in the model results. For a case of 1.5 m rise in sea level in 90 years, it is found that the rate of landward movement of the 1 kg/m3 due to sea level rise acts as a dominating process over the rate of migration (15 m/year). Thi s creates a rapid landward movement of the isochlor due to lateral saltwater i ntrusion. In other scenarios of 0.5 m and 1 m SLR in 90 years, a reverse effect is seen with island migration rate dominating over isochlor movement rate. In the se cases, the 1 kg/m3 isochlor regains its position close to the shorelin e.
94 Anthropogenic effects such as groundwater withdrawa l could adversely affect freshwater resources in barrier islands. The freshwater lens would diminish in size as a result of continued withdrawal of grou ndwater and rapid vertical saltwater intrusion through upconing. Saltwater int rusion as a result of pumping could occur within time scales of years while that due to sea-level rise occurs in timscales of decades (e.g. Oude Essink, 2001). Thus saltwater intrusion in migrating barrier islands is likely to occur due to SLR but to a greater extent due to human impacts and adverse climatic changes.
95 Table 5.1 Hydraulic parameters used for SEAWAT mode l simulation Layer thickness is given by b, Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective por osity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. Upper Aquifer Semiconfining unit Lower Aquifer Confining Unit Buried wetland b, (m) 15 3 to 15 9.5 23 3 Kh (m/day) 25 0.0845 20.3 0.0845 0.204 Kv (m/day) 18 0.000845 17.7 0.000845 0.178 n 0.15 0.2 0.2 0.02 0.2 S 0.01 0 0.01 0 0 Sy 0.1 0 0.1 0 0 a aa aLh (m) 0.8 0.8 0.8 0.8 0.8 a aa aLv (m) 0.008 0.008 0.008 0.008 0.008 a aa aT (m) 0.001 0.001 0.001 0.001 0.001
96 Figure 5.1. Schematic showing landward migration/re trogration of a barrier island by the process of overwash deposition (adapted from Rosati, 2009) Figure 5.2 Location map of Hatteras Island. Geolog ical and hydrogeological data available from previous studies have been used for modeling purpose in the present study.
97 Figure 5.3a Schematic of Hatteras Island (cross-se ction) with hydrologic units as used in model simulations (from Anderson, 2000)
98 0 150 distance (m) Figure 5.3b. Cross-section of 2D model showing mode l geometry, boundary condition and hydrogeologic units designed for pres ent study based on Andersons model (Figure 5.3a). Constant Head/Conc. B.C. buried wetland Upper Buxton Woods Aquifer Semi-confining layer Lower Buxton Woods Aquifer Confining Layer Pamlico Sound (north) Atlantic Ocean (south) 7500
99 Figure 5.4 Salinity distributions from model simula tions. a) Base case scenario (Model1) where freshwater lens extends upto a depth of ~18 m. The lens shape is slightly asymmetric, indicating stratigraphic co ntrol over lens formation. The transition zone (defined by 1kg/m3 and 10kg/m3 isochlors) is wider along the Sound side. The land-sea boundary position is shown by the asterisk. (All figures are vertically exaggerated).
100 Figure 5.5 Salinity distributions from Model 2a (SL R =0.5m in 90 years, island migration rate =5m/yr) shown at different timesteps
101 Figure 5.6 Salinity distributions from Model 2b (SL R =1.0 m in 90 years, island migration rate =10 m/yr) shown at different timeste ps.
102 Figure 5.7 Salinity distributions from Model 2c (SL R =1.5 m in 90 years, island migration rate =15 m/yr) shown at different timeste ps.
103 Figure 5.8a Changes in water table elevations as th e barrier island migrates with rising sea-level. The head distributions generated are due to increased recharge and rise in sea level along with new migrated posit ions of the island over the underlying, low conductivity semi-confining unit.
104 Figure 5.8b Comparison of water table elevation cha nges under conditions of increased recharge, constant recharge as the island migrates and hypothetical scenario with constant recharge but no migration. A ll models consider 1m rise in sea-level in 90 years. The scenario with increased recharge in a migrating island (i.e. Model 2b) produces the maximum head values in the central part of the island. When recharge is kept the same as in initia l conditions, the head value drops indicating the effect of 10% increased rech arge. The effect of island migration dominates on the head distribution, as se en in the case of a static island with a much lower head value.
105 Figure 5.9 Salinity distribution obtained from Mode l 3a. a) Base case scenario shown for comparison purpose. b) Result from Model 4a, where recharge was reduced by 10% from the base case while sea-level r ise of 1m in 90 years and island migration rate of 10m/yr were considered (i. e. similar to Model 2b). The depth of the freshwater lens decreases to ~12 m by 6m from base case (18m) and by 13 m from sea-level scenario Model 2b (25 m) This shows that recharge plays a very important role in maintaining the fres hwater resource.
106 Figure 5.10 Salinity distributions (kg/m3) obtained from Model 3b shown at different timesteps. The model is similar to sea le vel rise scenario, Model 3b, with a pumping well extracting groundwater from the uppe r aquifer at 100 m3/day causing saltwater intrusion
107 Figure 5.11 Comparison of changes in head distrib ution over the barrier island under conditions of increased recharge (by 10% from base case), reduced recharge (by 10% from base case) and withdrawal thr ough pumping. The most influencing factor is recharge as also in figure 5. 8b.
108 CHAPTER 6 VARIABLE-DENSITY FLOW MODELS OF SALTWATER INTRUSION IN ATOLL ISLAND AQUIFER Introduction Atoll islands are small, carbonate reef islands, f ormed atop subsiding volcanic rims with a completely or partly enclosed lagoon (figure 6.1). Most atoll islands are found in the Pacific and Indian Oceans (Woodroffe, 2003). Many of these atoll islands host human populations e.g. nea rly 1000 islands of the Pacific are populated (White and Falkland, 2010). On atoll islands, shallow freshwater lenses, recharged by precipitation, are the sole so urce of groundwater. Most atoll islands depend on groundwater resources, especially during drought periods for domestic supplies of freshwater. Changes in climati c conditions, e.g. increase in temperature, changes in precipitation, rise in sea level, could lead to a drastic loss in freshwater resources and adversely affect l iving conditions on these oceanic islands. To cope with this stress, over-pum ping would become a probable, but short-term solution. Under all circum stances, the freshwater lens morphology of these carbonate islands will be alter ed. IPCC reports that a 10% decrease in average rainfall in the Pacific would r esult in 20% reduction in the size of freshwater lenses in Tarawa Atoll, Kiribati Numerical models of climatic effects on atolls in the Federal States of Micrones ia show that the freshwater resource may be depleted due to 6 month long drough ts that would require
109 approximately 1.5 years to recover from (e.g. Baile y et al, 2009). Examples of freshwater availability in different small oceanic islands, including atolls and volcanic, is given in Table 6.1 This chapter focuses on predictive numerical model s of saltwater intrusion in an atoll island with rising sea level, using the Laura area of Majuro Atoll, Marshall Islands in the Pacific Ocean as an example The results from this study provide useful insights for management of freshwate r aquifer in carbonate islands with geology similar to that of Laura, Maju ro Atoll. Hydrogeology of atoll islands The hydrogeology of several atoll aquifers has bee n well studied (e.g. Vacher and Quinn, 1997). From the geological framew ork of atoll islands, it is evident that sea-level changes have greatly control led the formation and growth of these islands. Holocene carbonate sediments form the rims of the atolls and overlie older Pleistocene limestone. This results i n a distinctive aquifer system called the dual aquifer (figure 6.2) where the Ho locene carbonate materials constitute a lower permeability aquifer and the und erlying Pleistocene carbonates form a high-permeability aquifer (Vacher, 1997). Th e contact between these two lithotypes, referred to as the Thurber Discontinuit y, occurs at 15 m to 25 m below sea level (Bailey, 2009). The presence of a high pe rmeability lower unit (higher by one to two orders of magnitude from overlying un its) generates a low head profile over the island (e.g. Vacher, 1988). In gen eral, freshwater lens is limited to a depth of 10-20 m, depending on the width of the i sland (Figure 6.3).
110 The nature of the freshwater lens in an atoll is a lso controlled by the presence of the reef flat plate on the ocean side o f the atoll. In Deke Island, Pingelap Atoll of the Federal States of Micronesia, the reef flat present at sea level extends for some distance under the island an d acts as a confining unit, thereby restricting the depth of the freshwater len s (Ayers and Vacher, 1986) on the ocean side. On the lagoon side, an unconfined a quifer exits that receives direct freshwater through recharge. The lens is dee per on the lagoon side than on the ocean side. The difference in hydraulic cond uctivity across the island creates an asymmetric lens that differs from the Du puit-Ghyben-Herzberg lens shape (e.g. Vacher, 1988). Unlike in siliciclastic barrier islands, the role of tidal forcing is significant in atoll hydrogeology. Some research in recent years h ave addressed the role of tides in coastal aquifers (e.g, Li et al., 2002, 20 08.; Cartwright et al., 2004; Jeng et al., 2005), and saltwater intrusion (e.g Ataie-A shtiani et al., 1999). Underwood (1990) described the effects of tides on freshwater lenses and transition zones in atoll islands in modeling studies. His results show that tidal oscillations facilitate mixing of fresh and saltwater within the transition zone by vertical flow and that tidal mixing is controlled more by vertical longitu dinal dispersivity than by the transverse component. Tidal efficiencies were found to increase with depth, indicating tidal wave propagation from below within the high permeability units, followed by upward movement. Underwood (1992) also simulated the relation between freshwater lens thickness and the island wi dth under different conditions of recharge (figure 6.4).
111 In a similar study, Bailey et al., (2009) conducte d numerical investigations of atoll hydrogeology based on Ayer and Vachers (1 986) conceptual model. Their results demonstrate that freshwater lens thic kness increases with an increase of island width, depth to the Thurber Disc ontinuity, the presence of the reef flat plate and with a decrease in hydraulic co nductivity of the Holocene Upper Aquifer. The authors also examined the effect of climatic change such as drought conditions due to El-Nino events and found that the freshwater lens shape would undergo reduction in size during drough t. To summarize, the overall dimensions of the freshw ater lens and the associated transition zone is controlled by the fol lowing factors as discussed by Underwood, (1990) based on Wentworths findings (19 47): These are i) the degree of aquifer permeability that permits infiltr ation by recharge but arrests freshwater-saltwater mixing at discharge areas and in the transition zones, ii) sufficient recharge to maintain the freshwater lens iii) tidal amplitude, iv) preferential flowpaths that facilitates mixing with in the transition zone and v) the presence of a confining unit that hinders discharge and allows storage resulting in increased lens size. Generic model of atoll island Introduction Characteristics of freshwater resources in atoll i slands have been studied using both analytical (e.g. Fetter, 1972) and numer ical methods (e.g. Llyod et al., 1980; Falkland 1983, Herman and Wheatcraft, 1984). A summary of modeling
112 studies of atoll islands by several authors is disc ussed in Bailey et al. (2009). Few studies have addressed the issue of saltwater i ntrusion in atoll islands due to sea level rise. Bobba (1998) investigated the e ffect of sea level rise by 0.05, 0.075 and 0.1 m in Agatti, Laccadive Islands in the Arabian Sea with the aid of a two-dimensional SUTRA models. The results indicate a reduction in freshwater lens thickness under all three SLR scenarios and in crease in salinity in the coastal zones due to saltwater intrusion. In the present study, a generic model of an atoll island is considered to examine the availability of freshwater under differ ent SLR scenarios with steadystate pumping. Geologic and hydrogeologic informati on required for the design of the model is based on information available for Lau ra, Majuro Atoll, Marshall Islands in west-central Pacific Ocean. Geological setting The geologic framework of Marshall Islands as desc ribed by Peterson, (1998) is rephrased here. Majuro Atoll lies in the south-eastern part of the chain of islands forming the Republic of Marshall Islands The islands reside on a basaltic platform formed by the Pacific Plate. A th ick sequence (1.3-1.4 km) of carbonate deposits consisting of unconsolidated lag oonal sediments, loose reef debris, well-lithified coralline sediments and reef plate have been encountered while deep drilling in parts of the Marshall Island s, such as in Bikini and Enewetak. The evolutionary phases of the islands in clude an initial phase of reef formation after subsidence of volcanic edifices dur ing the Cretaceous period, followed by significant effects of sea-level change s during the Quaternary times.
113 Fluctuations in sea level caused repeated events of emergence and submergence of the reef platform leading to erosion of older carbonate deposits and deposition of newer materials. This process has created a number of unconformities. The Thurber Discontinuity represent s the most hydrogeologically significant unconformity, separating moderately per meable Holocene sediments from the older highly permeable Pleistocene deposit s. Hydrogeology of Laura, Majuro Atoll The freshwater lens of the Laura area serves as th e major freshwater source for Majuro Atoll. The hydrologic units compr ising the aquifer are i) an Upper Sediment Lithofacies composed of uncemented g rainstone, formaminiferal sand and coralline algae, ii) a Lower Sediment Lith ofacies composed of both lithified and unlithified grainstone and wackestone and iii) a Lower Limestone Lithofacies (figure 6.5). The upper limestone unit (reef flat plat) is composed of algal boundstones and grainstones (Anthony et al., 1989). The shape of the lens is asymmetric with greater t hickness towards the lagoon side. The presence of a reef flat plate on t he ocean side controls the lens thickness and impedes the areal extent of freshwate r. According to previous studies, the thickness of freshwater lenses in Laur a extended upto 20 m, which is in agreement with Underwoods model results. The le ns configuration is controlled by factors such as seasonal variability of precipitation, hydrostratigraphy, groundwater withdrawal and storm events. Anthony et al, 1989, studied the hydrochemical aspects of the atol l aquifer and noted that the
114 flow and transport mechanisms are controlled largel y by asymmetric distribution of lithofacies. Presley (2005) found that the freshwater lens thi ckness in Laura fluctuated on a seasonal basis, with thinning occurring during the dry season and thickening during the wet season. This is due to the increase in groundwater demands during dry seasons. Continued withdrawal has also r esulted in upconing and saltwater intrusion around pumping wells. Reduction in lens size as well as high chloride concentrations in monitoring wells, especi ally near the center of the lens were noted during the drought period (figure 6.6.). The island has been subjected to repeated drought events and emergency situations have been declared in recent times, as local sources report. It is unders tood from the results of previous monitoring report, that the freshwater resource in Majuro atoll is very sensitive to climate change impacts. Groundwater flow models The dual aquifer system of this area has been prev iously studied as well as numerically simulated by Griggs and Peterson (19 93). A two dimensional density-dependent SUTRA model was used to simulate the availability of freshwater with increased pumping. In the present study, three-dimensional groundwate r flow models were simulated using the computer code SEAWAT (Guo and L angevin, 2002). Models were designed and executed using the software Groun dwater Vistas (GWV version 5.22).
115 Hybrid Method of Characteristics technique (HMOC) w as chosen to solve the flow and transport equations as the method has grea ter accuracy and computational advantage over other methods (Zheng a nd Wang, 1999) Each model was run using a number of particles equal to 64 in each cell. Model description Model design Laura island area is represented by a three-dimen sional finite-difference grid. The model consists of 10 layers, 63 rows and 111 columns. Uniform cell dimensions of 25 m by 25 m were used for this model ing Time discretization For the SEAWAT models, Courant number was specified to 1 allowing maximum flow and transport over a distance equal to one grid spacing. Hydrologic units The upper four layers (total thickness = 12 m) com prise the Upper Sediment Lithofacies. Layers 5 to 8 (total thicknes s=16 m) correspond to Lower Sediment Lithofacies and layers 9 and 10 form the L ower Limestone Lithofacies. The upper limestone unit forming the reef flat plat e is 3 m thick and extends upto 320 m landward from the ocean side. Boundary conditions In all of the models, sea level is represented by a Constant Head Boundary (CHB) condition where the head of the sea level and the chloride concentration in seawater were set to zero meters a nd 35 kg/m3 respectively. Noflow boundary conditions were employed around the e dges of the model domain.
116 The watertable forms a specified flux boundary whic h receives water through recharge. Figure 6.7 shows a schematic of the model grid, hydrogeology and boundary conditions. Hydraulic parameters Hydraulic parameters (hydraulic conductivity, poro sity, specific storage, dispersivity, recharge etc) used for model generati on are similar to values used in Petersons SUTRA model. Initially, the exact values for conductivity and dispersivity were used as in Petersons study. Howe ver, these values did not generate an aquifer as deep as is expected in an at oll island of ~1000 m width. Hence the values were changed to produce a reasonab ly thick (~10 m) freshwater lens for the initial conditions. Table 6 .2 gives the hydraulic parameters used for the models. A recharge rate of 0.008 m/day was used for the base case over the island area and 8E-05m/day over the reef f lat plate. The recharge over the island was changed to 0.002 m/day for the sea-l evel rise scenarios while that over the reef was kept unchanged. A higher recharge rate was used for the base case to generate a deep lens and to limit the compu tational run time. Tidal effect simulation Since the model simulations are steady-state over a time interval of 10 years, the transient effect of tidal variations cou ld not be directly employed as tidal boundary conditions. A tidal model would requ ire implementation of many stress periods each day to allow head changes at th e ocean side to simulate tidal fluctuations. Such a simulation would be computatio nally expensive. An alternate method for simulation of tidal effects is to increa se the transverse dispersivity
117 value in the models. Underwood (1990) demonstrated that in non-tidal models that do not consider actual tidal boundary conditio ns, the transition zone expansion occurs transverse to flow direction. Thus an increase in transverse dispersivity value compensates for synthetic genera tion of tidally influenced fresh and saltwater mixing. Model simulations Base case scenarios A presumed salinity distribution was used with fre shwater of zero kg/m3 concentration representing the atoll aquifer within the top layers. The salinity concentration of seawater was set to 35 kg/m3 in the CHB cells representing ocean and lagoon water. Head values in these cells were set to zero m. A recharge rate of 0.008 m/day was applied and the mo del was run until steadystate for 7300 days or 20 years. This model served as a predevelopment case. In the next model, Model 1 six pumping wells were placed using the well package in the previous model. The distribution of wells and pumping rates (table 6.3) were based on data available from previous the USGS report (Presley, 2005). The model was simulated for 0 years with re charge of 0.002 m/day, using head and concentration values obtained from the pre development case model. In Model 1, the system is no longer in equilibrium. Co ncentration and head distributions from Model 1 were used as initial con ditions for the subsequent sea level rise models. Sea level rise scenarios Three SLR scenarios were assumed with a net rise of 0.5 m, 1 m and
118 1.5 m within 90 years. Head values in the CHB cells representing SLR were increased accordingly for each case. The models wer e simulated in steady-state where in each scenario, a model was first run for 1 0 years. The following model, run for the next 10 years, used concentration and h ead values from the first simulation as its initial conditions. Thus for ever y scenario, a succession of nine models was generated to obtain the final results. Constant withdrawal in wells was maintained at steady-state. Model 2a represents the first scenario with a rise in sea-level of 0.5 m in 90 years. In the second and third scenarios represente d by Model 2b and 2c, the net rise in sea level is 1 m and 1.5 m respectively. Re charges of 0.002 m/day over the island part and 8E-05 m/day over the reef flat were used for all the models. Results and discussion Figure 6.8 shows the result from Model 1, the base case scenario that includes withdrawal at different places. A dual aqu ifer system is generated with a fresher lens within the upper Holocene lithofacies while a saline aquifer underlies in the lower Pleistocene lithofacies. The reef flat controls the extent of the freshwater lens on the ocean side, where a wide tra nsition zone is created and freshwater is absent. The transition zone on the la goon side is narrow. The freshwater lens is deeper on the lagoon side, resul ting in an asymmetric lens configuration. The effect of groundwater withdrawal from the upper 3 m is evident by the presence of salinity increase around the pum ping wells. The freshwater lens extends upto a depth of ~7.5 m. (Not included in figure)
119 The model results of SLR scenarios are given in fig ures 6.9 to 6.10. In each figure the salinity distribution is shown f or the three models (2a, 2b and 2c) at the same depth. Figure 6.9 shows that at a d epth of 1.5 m. For a SLR of 0.5 m (as in Model 2a), saltwater intrudes from the ocean side resulting in advancement of high salinity isochlors (10 kg/m3, 20 kg/m3) toward the freshwater lens. Although less in extent, saline en croachment occurs along the lagoon side, especially due to withdrawal effect. P umping from well site 1 induces saltwater flow towards the well, causing sa ltwater intrusion in that direction. When SLR is 1 m (Model 2b), saline intrusion conti nues further inland, decreasing freshwater availability along the lagoon side. The result from Model 2c shows overall reduction in the areal extent of t he freshwater lens with 1.5 m rise of sea level. In this case, however, the salin ity over the reef flat area of the island along the ocean side decreases from the prev ious two scenarios. As sea level rises, it causes rise in the head values over the island. Head gradient controls the direction of flow. In this case i.e. w hen SLR is 1.5 m, increased head forces greater flow and discharge of freshwater alo ng the island edges, leading to greater mixing of fresh and saltwater. The groun dwater flow is parallel to transition zone along the island edges. Since trans ition zone thickness is sensitive to transverse dispersivity, a high value of transverse dispersivity used in the simulation facilitates widening of the transiti on zone along the ocean side. During mixing of fresh and salt water, less saline water displaces more saline water and drives the high salinity isochlors oceanw ard.
120 In cases of 0.5 SLR and 1 m SLR, this response is not noted. It is possible that at these rates, the head elevations and related dis charge are not sufficient to force the saline front oceanward or induce greater mixing. Figure 6.10 shows the salinity distributions obtai ned at a depth of 4.5 m. For the first scenario i.e. Model 2a, the freshwate r lens extent has decreased and is only present at the central part of the aquifer while saline water occupies major part of the aquifer. Proximity of pumping wells to the lagoon side allows saltwater intrusion to advance inland. Upconing of saline wat er is also evident at well sites. For second scenario (Model 2c), the freshwater len s in the center of the island is even more diminished than in the first ca se, leaving only pockets of freshwater. This indicates a further increase of sa linity due to sea level rise effects. Saltwater intrusion occurs on the lagoon s ide due to pumping effects at both the well sites that are close to the boundary. Results from Model 2c show the disappearance of th e freshwater lens with insignificant number of freshwater pockets remainin g around the central part of the island. From the above results it can be inferred that sea level rise induces saline intrusion from below the freshwater aquifer and hen ce a reduction in the lens thickness in the vertical direction. The effect of pumping, as studied earlier by Peterson, 1990, causes further saltwater advancemen t. Figure 6.11 shows the changes in water table eleva tions across the island under the three sea level rise scenarios. Head valu es increase by ~0.2 m for each case of 0.5 m rise in sea level, measured at 7 00 m on the x -axis.
121 Conclusions The three dimensional model simulations of an atol l island aquifer under different sea level rise scenarios indicate that fr eshwater resources will be adversely affected by saltwater intrusion, particul arly along the coastal areas. In all three cases (SLR =0.5 m, 1.0 m and 1.5 m by 210 0), the areal extent of the freshwater lens is reduced. If sea level rises by 0 .5 m, a freshwater nucleus exists at a depth of 4.5 m, but under conditions of higher SLR the nucleus is diminished to sporadic freshwater pockets. Saline intrusion is further accelerated due to freshwater withdrawal through pumping. The models in this study did not consider the effect of reduced recharge along with SLR. However, atoll islands are likely to be subjected to less recharge in the future, an effect that would further diminish the availability of fresh groundwa ter.
122 Table 6.1 List of some Pacific Island Countries with populated atolls and volcanic islands. Islands with freshwater lens are shown in italics ( adapted from White and Falkland, 2010).
123 Table 6.2 Hydraulic parameters used for SEAWAT mode l simulation Upper Sediment Lithofacies Lower Sediment Lithofacies Lower Limestone Unit Upper Limestone Unit b, (m) 15 3 to 15 9.5 23 Kh (m/day) 70 700 7000 0.1 Kv (m/day) 7 70 700 0.01 n 0.2 0.2 0.3 0 S 0.15 0.15 0.18 0 Sy 0.15 0.15 0.18 0 a aa aLh (m) 3 3 3 3 a aa aLv (m) 0.01 0.01 0.01 0.01 a aa aT (m) 0.2 0.2 0.2 0.2 Layer thickness is given by b, Kh is horizontal hydraulic conductivity, Kv is vertical hydraulic conductivity, n is effective por osity, S is storage co-efficient, Sy is specific yield, aLh is horizontal longitudinal dispersivity, aLv is vertical longitudinal dispersivity, aT is transverse dispersivity. Table 6.3 Pumping rate used in model simulations Well No. Pumping rate (m3/day) 1 175 2 135 3 130 4 65 5 160 6 130 Pumping data is from Presley, 2005.
124 Figure 6.1 Atoll islands with different sizes with enclosed or semi-enclosed lagoons. A) Pingelap Atoll, Federal states of Micro nesia B) Diego Garcia Atoll, Indian Ocean and C) Majuro Atoll, central-western P acific Ocean. The direction of prevailing winds influence the size of the islan ds, the size larger along the leeward side than the windward side where wind forc es cause destruction of the atoll margins (from Bailey et al., 2008).
125 Figure 6.2 Typical hydrogeology of an atoll island showing a shallow freshwater lens with a transition/mixing zone within Holocene formations separated by the Thurber discontinuity from the underlying Pleistoce ne deposits. The arrows at the bottom indicate the direction of tidal propagation (adapted from Bailey, 2009).
126 Figure 6.3 Relationship between atoll island width and maximum thickness of freshwater lens base on their mode of occurrences w ithin the atolls (from Bailey, 2009).
127 Figure 6.4 Model results of potable water (2.6% sal inity) depth at island centers with respect to island width for different atoll is lands under different recharge (R) conditions (from Underwood, 1992).
128 Figure 6.5 Hydrogeologic units of Laura area, Majur o Atoll (from Anthony et al., 1989.
129 Figure 6.6 a) Model plan view showing the position of well sites. Well site 1 consists of wells 1,2 and 3 and well site 2 has wel l numbers 4,5 and 6. Pumping rates are given in Table 6.3. b) Cross-sectional vi ew of the model along AA showing the hydrologic units and boundary condition s. a. b. well site 2 well site 1 A A A A No-flow boundary condition Constant head & concentration boundary conditions Upper Sediment Lithofacies Lower Sediment Lithofacies Lower Limestone Lithofacies Reef Flat Plate
130 Figure 6.7 Salinity distribution map of Model 1 sho wn at different depths. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens boundary. At a depth of 1.5 m (upper figure) and de pth of 4.5 m (lower figure)
131 Figure 6.8 Salinity distribution maps of Models 2a (upper figure), 2b (middle figure) and 2c (lower figure) shown at a depth of 1 .5 m. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens b oundary. For a rise of 0.5 m sea level, saltwater intrusion occurs from the ocea n side, increasing salinity over the reef flat plate area of the island. Pumping ind uces saline intrusion along the lagoon side. Overall lens extent has reduced. For s ea level rise of 1m, saltwater intrusion increases due to pumping close to the lag oon side, creating an irregular shape of the lens. Lens extent reduces from the fir st scenario. For a sea level rise of 1.5 m, saltwater intrusion continues along the lagoon side, while fresh and salt water mixing widens the transition zone along the ocean side. Details given in text.
132 Figure 6.9 Salinity distribution maps of Models 2a (upper figure), 2b (middle figure) and 2c (lower figure) shown at a depth of 4 .5 m. Concentration is in kg/m3, with 1kg/m3 isochlor representing freshwater lens b oundary. For a rise of 0.5 m sea level, only a freshwater nucleus exists at the island centre with saline water elsewhere. At high sea level of 1m, the freshwater nucleus reduces to pockets and diminishes with higher sea level rise of 1.5m. Saltwater intrusion due to pumping is evident under all three sea level rise c onditions.
133 distance (m) Figure 6.10 Changes in water table elevations under different rates of sea level rise. X axis represents distance in meters across t he model and y-axis is water table elevation above MSL (m)
134 CHAPTER 7 SUMMARY OF RESULTS The response of coastal groundwater systems to cli mate-change related sea level rise (SLR) is significantly controlled by the stratigraphy of the depositional environments. Variable-density flow mo dels of major coastal landforms (e.g. deltas, estuary, and small islands) indicate the extent of salinization varies in different natural coastal sy stems. Model results of deltaic aquifers reveal that a wave-dominated delta is most susceptible to SLR compared to a river and tide-dominated deltas. Unde r natural conditions, with a relative sea level rise of 1.68 m, the freshwater-s altwater interface advances landward by 5 km in 90 years in a wave-dominated de lta. Wave-dominated deltas are characterized by high conductivity grave l and sand deposits that allow saltwater intrusion in the aquifer. Groundwater res ources in natural riverand tidedominated deltas are negligibly affected by r elative sea level rise. This is primarily due to the presence of widespread distrib ution of low conductivity sediments such as silt and clay in riverand tidedominated deltas that impedes saltwater influx. In a partially mixed, microtidal estuary, sea leve l rise would cause severe salinity increases in the sediment interstitial wat er for a sea level rise of 1.5 m in 90 years. Salinity changes are primarily controlled by the distribution of
135 sediments, particularly sand deposits which have co mparatively high hydraulic conductivity than the silty clay units deposited in the central portion of the estuarine channel. Saltwater intrusion models of small islands indica te that carbonate atoll islands are highly sensitive to SLR scenarios with depletion of freshwater lens. Under the three sea level rise scenarios (0.5 m, 1 m and 1.5 m) considered in this study, the volume of the freshwater lens reduc es due to saltwater intrusion along the coastal areas. The presence of a high per meability lower Pleistocene carbonate unit (higher by one to two orders of magn itude from overlying units) causes increased saltwater intrusion. Effects of ti dal forcing, simulated with high values of transverse dispersivity, are also promine nt in atoll islands that induce vertical flow and fresh and saltwater mixing, resul ting in a widened freshwatersaltwater transition zone. For sandy barrier islands, sea level rise would ca use the island to migrate landward, resulting in morphological changes of the freshwater lens. Numerical simulations indicate that the freshwater lens in a natural migrating barrier island could be retained with increased recharge, even wit h a sea level rise of 1.5 m, provided that the island survives from inundation. For a case of 1.5 m rise in sea level in 90 years, it is noted that the rate of lan dward movement of the saltwater interface due to sea level rise acts as a dominatin g process over the rate of migration (15m/year). This creates a rapid landward movement of the freshwatersaltwater interface due to lateral saltwater intrus ion. In other scenarios of 0.5 m and 1 m sea level rise in 90 years, a reverse effec t is seen with island migration
136 rate dominating over interface movement rate. In th ese cases, the interface regains its position close to the shoreline. Althou gh, sea level rise does not cause saline intrusion in the island, increased evapotran spiration caused by change in vegetation pattern would lead to decrease in rechar ge resulting in reducation of freshwater lens size. Anthropogenic impacts of groundwater withdrawal in addition to SLR, in all the coastal landforms aid in acceleration of saltwa ter intrusion. The impact is severe in small islands, especially in atolls, with depletion of freshwater lens. Saltwater advancement is also greater in wave-domin ated delta compared to the river and tide-dominated deltas, where the freshwat er-saltwater interface moves landward by 8 km for a relative sea level rise of 1 .18 m. Groundwater abstraction from coastal aquifers causes vertical saltwater int rusion through upconing. The numerical simulation results suggest that anthropog enic impacts on coastal aquifers can easily exceed the effects of saline in trusion resulting from sea level rise.
137 CHAPTER 8 TIME-FREQUENCY ANALYSIS OF GROUND PENETRATING RADAR SIGNALS Introduction G round-penetrating radar (GPR) is widely used for in terpretation of sedimentary deposits, where beds often occur in lay ered sequences and are often too thin to be individually resolved. Guha et al. (2005) showed that GPR traces over laminated sequences shift toward higher frequencies and spectral analysis can be used to detect thin beds. In that s tudy, very nely laminated sequences (well below the tuning thickness, 1/250th of dominant wavelength) were considered and the radar frequency response wa s obtained using the Fourier transform. Estimating frequency response th rough the Fourier transform, however, does not provide information regarding the variation of frequency with time. On the other hand, joint time-frequency analy sis, or JTFA, is a processing method that captures energy localization of a signa l with time and allows representation of variations in spectral content of a signal in both the time and frequency domains. Direct calculation of thickness is only possible in the time domain when the bed is resolvable, i.e., when the bed thickness is greater than the tuning thickness, or ~1/4 of the dominant wavelength. Howe ver, studies of seismic resolution have shown that thin beds (thickness < t uning thickness) can be
138 revealed through frequency-domain analysis. The present study expands on our earlier work to e xamine whether timefrequency analysis of GPR data can locate thin beds within discrete intervals in radar records. We use the S-transform (Stockwell et al., 1996) for the transformation of nonstationary, temporal GPR signa ls to their frequency components. The S-transform can be considered a wav elet transform in which the mother wavelet is multiplied by a phase factor. The window function that translates over time is Gaussian. An advantage of t his JTFA is that the window function has a scalable length that depends on the frequency content of the signal; the window is longer for low frequencies an d shorter for high frequencies. Synthetic model examples A nite-di erence time domain (FDTD) method was used to simula te GPR wave propagation through thin layers in one dimensi on. The 1D model has zero o set between the transmitter and receiver, resulting in normal incidence of the emitted plane wave. A trilobed pulse, similar to a Ricker wavelet, has been used. In order to capture the minimum layer thickness of 1 mm, a cell size of 0.25 mm and time step of 5 10-4 ns has been maintained. Five models have been designed with the rst four models representing depositional sequences in which the individual beds within the sequence have a mean thickness of 30 cm, 15 cm, and 3.75 cm and 0.4 6 cm, respectively (Figure 1). These bed separations are the mean for Gaussian distributions of layer thicknesses in each model. In all models, the indiv idual beds are separated by a 1-mm layer which represents a contact zone. In orde r to maintain an average
139 velocity of 0.12 m/ns (as measured for unsaturated sands in eld data), the individual beds have relative dielectric permittivi ty r = 6 while the contact layers have r = 35. Both beds and the contact layers have relati ve magnetic permeability, r =1 and conductivity, = 1 mS/m. Real parts of permittivity and conductivity are considered. The fifth model consists of two sequences of beds with the first having a mean thickness of 30 cm and a second underlying set with mean thickness of 3.75 cm. The bed thicknesses used in this sequence are different realizations from those used in the first and third models. The purpose of this model is to examine how the proposed JTFA meth od captures a transition zone from a thick bed sequence to thin beds. Models were run with a pulse center frequency of 200 MHz. The dominant wavelength ( d) is 60 cm for the frequency and velocity used. The bed thicknesses of the four models thus correspond to d/2 (30 cm), d/4 (15 cm), d/16 (3.75 cm), and d/128 (~0.46 cm). The purpose of selecting this range of models is to analyze the frequency characteristics of beds with thickness greater than equal to, and less than the tuning thickness. As observed by Guha et al., a spe ctral shift towards higher frequency is expected from layered sequences (mm-sc ale) below tuning thickness. Time-frequency representations of the fi ve model traces are shown in Figure 2. For an average bed thickness of d/2 (30 cm), the S-transform of the simulated trace shows little variation of frequency distribution (model 1). Input frequency ( 200 MHz) is maintained throughout the sequence as i s expected for a thick bed (i.e., bed thickness > tuning thickness ). When the bed thickness is d/4 (15 cm), a peak frequency of 400 MHz is consistently attained, as this is
140 the tuning frequency corresponding to the tuning th ickness (model 2). Trace analyses for bed thickness equal to d/16 (3.75 cm) and d/128 show rises in peak frequency from the input frequency at times co rresponding to higher amplitude returns (e.g., within 5060 ns in model 3 and 3040 ns in model 4). The higher amplitudes are presumably the result of constructive interference of the reflecting energy. These model results clearly indicate that different frequency responses are generated by sequences with beds with thicknesses less than, equal to, or greater than the tuning thi ckness. TFA of the trace in model 5 shows a distinct trans ition in frequency from 200 to 300 MHz at 35 ns. This transition correspond s with the thick bed and thin bed contact, indicating that the S-transform method has good temporal resolution of the frequency variations. Localization of high-e nergy responses, as generated from the thin-bed models, in both time and frequenc y could be different with different realizations of the Gaussian layer thickn ess distributions. It is evident from the JTFA results that sequences of beds of thi ckness tuning thickness are distinguishable from sequences of thick beds by the ir high peakfrequency response. This is similar to single-bed frequency a nalysis (e.g., as noted by Barnes et al., 2004) for which high average frequen cies are related to beds thinner than the tuning thickness. It is, however, not possible from our model results to extract information on bed thicknesses b elow the tuning thickness, (i.e., to distinguish between cmand mm scale beds for a 200-MHz center frequency pulse).
141 Example 1: Thin beds in coastal deposits The study area is on Santa Rosa Island on the coast of the panhandle of Florida. Santa Rosa is a long (73 km) barrier island that wa s intensely affected by hurricanes Ivan (in 2004) and Dennis (in 2005). GPR data described and analyzed in the following sections are from a site close to the Gulf of Mexico on the washover platform. GPR data collected with a Se nsors and Software Noggin Smartcart by Wang and Horwitz (2007) used 250-MHz a ntennas to identify the base of the hurricane washover deposit. A common-of f set mode was used with antenna separation of 0.3 m and step size of 0.05 m A CMP profi le was collected at a frequency of 200 MHz with the PulseE KKO 100 system of Sensors and Software. A GPS unit, with a wide-area augmenta tion system (WAAS), was used to acquire profile positions. Topographic surv eying used an electronic total survey station (Sokkia, SET500). Th e radar record was ground-truthed with cores and trenches at places of interest. Because t he study targeted washover deposits, no ground truthing was done for character ization of the pre-Ivan sediments that underlie the Iw facies. Reflection profiles were dewowed, gained (constant ), and corrected for topographic effects. CMP velocity analysis yields a velocity of 0.12 m/ns for the unsaturated zone (above water table) and 0.06 m/ns for the saturated zone (below water table). A post-Ivan washover deposit ( Iw) is revealed in the 250MHz profile (Figure 3a). The base of Iw is marked b y a freshly buried grass layer, detected by trenching, extending from 5.5 m to 7 m along the transect (Figure 3b). The layered sequences at this location, within 1.2 m of the ground surface,
142 consist of beds with thickness of 2 cm and are composed of pure homogeneous quartz sand. The GPR profile shows both dipping and horizontal reflectors. GPR images the bounding surfaces whose thicknesses can be measured in the time domain. However, it does not resolve the cm-scale l ayered sequences between these bounding surfaces observed in the trench data We refer to these groups of thin beds bounded by reflective surfaces as package s. Two GPR traces are extracted from the profile for JTFA, trace 1 at 2 m and trace 2 at 5.8 m (Figure 3a). For the JTFA, the traces are dewowed but not g ained to preserve the original reflection amplitude and signal attenuatio n trend. Figure 3d shows the trace analysis from the GPR profile. Trace 1, at 2 m, contains reflections from Iw beds up to 20 ns followed by reflections off pre-Iv an strata. Noted that the peak frequency corresponding to the downgoing pulse has increased from 290 MHz at the more homogeneous site to 300 MHz. This could be due to reflections off cm-scale bed sequences starting immediately below t he ground surface, or differences in antenna-ground coupling between the two sites. The S-transform of the trace shows a gradual increase in peak frequ ency after 10 ns, reaching a maximum value of 400 MHz at 20 ns. From ~ 5 ns to 3 0 ns, a gradual downward trend is noticed in the peak-frequency distribution reaching a value of 300 MHz. A spectral splitting is observed at 30 ns, at the i nterpreted water table reflection. At 35 ns, a rise in frequency occurs up to 400 MHz. This corresponds to a reflective horizon in the radar profile which has n ot been identified in the absence of ground truthing. Interpretation of the frequency response of trace 1 is based on model results. The high frequencies observed within the 1530 ns time window
143 are a result of interference from reflections off t hin beds within the reflective surfaces in the GPR data. The drop in frequency at 30 ns to ~250 MHz, corresponding to the water table, could be due to t he dominant contribution of this boundary reflection. The second rise in peak f requency at 35 ns, corresponding to the prominent radar reflector, cou ld be due to the presence of a package consisting of layered sequences of thin bed s (thickness tuning thickness in the saturated zone). Trace 2, at 5.8 m was selected to correlate with the available trench data. Here 300 MHz is maintain ed until 10 ns. A rise to 400 MHz is observed between 10 and 13 ns. Within the ti me window of 1325 ns, the peak frequency remains at ~400 MHz. The contact bet ween Iw facies and preIvan sediments occurs at 18 ns. At ~26 ns, a spectr al splitting, similar to the frequency response of trace 1, is observed. The wat er table here is also interpreted to occur at 30 ns. A rise in frequency is observed at ~32 ns, corresponding to the highly reflective package. The frequency response within the time window of 1018 ns corresponds very well t o the zone of horizontal packages with cm-scale beds. Thus, JTFA of both GPR traces appears to capture the location of thin beds that are unresolv ed in the time domain. Example 2: Contacts bounding layered packages in vo lcanic deposits Cerro Negro, a basaltic cinder cone near Leon, Nic aragua, has erupted numerous times since 1850, most recently in 1992 an d 1995. Figure 4a shows a GPR transect that runs across the blanket of tephra deposits from repeated eruptions of the volcano. Data were collected with Sensors and Software
144 PulseEKKO 100 system with 100-MHz antennas over the central portion and 200-MHz antennas over the far ends of a line that c rosses a blanket of tephra deposits resulting from repeated eruptions of the v olcano. This profile lies approximately 1 km downwind of the vent, where the accumulated tephra from eruptions since 1850 reaches a thickness of ~20 m. The basal reflector that extends to two-wave travel times of 300 ns is inter preted as the contact between the 1850 tephra blanket and underlying lava flows a nd older tephra. The strongest reflecting horizons are the contacts that mark intervals between eruptive events. In some cases these contacts are p aleosols; in other cases they may represent principally a change in grain size an d porosity. The contacts can in practice only be examined in the field by trench ing in the distal parts of the tephra blanket where deposits are thinner. Thus GPR permits imaging of otherwise inaccessible near vent portions of the tephra fallout. The S-transform of a trace (Fig ure 4b) shows a distinction in the peak frequencies of returns coming from within individual deposits and returns from the contacts that mark the boundaries between eruption events. Returns from within the deposits of an eruption sho w peak frequencies greater than ~160 MHz (from a 100-MHz antenna source). We i nterpret this to indicate that these tephra deposits are internally stratifie d. This is plausible, given that individual eruption events were often episodic, and grain size distributions should vary between episodes. (Since trenching is only fea sible in the distal thin portions of the tephra blanket, such internal layering may b e less pronounced and more difficult to resolve than where ground truthing is possible.) In contrast, returns
145 from contacts between major eruptions show lower pe ak frequencies, ~100 MHz, in particular the reflection from the base of the 1 992 deposits and the reflection marking the base of the major 1850 erupt ion. These contacts presumably separate units with very different poros ities. Thus contacts preserved between major eruptions are spectrally distinct fro m the returns within eruption packages. It is interesting that the base 1992 retu rn stands out from adjacent returns in the GPR trace in peak frequency, not in amplitude. Thus JTFA may prove useful in extrapolating tephra stratigraphy b etween GPR lines. Figure 4c shows relative energy at 125 MHz, obtained by spect ral decomposition of the central part of the GPR profile. The distinct low e nergy at 125 MHz from the 1850 reflector suggests this is a thick bed single con tact with peak frequency below 125 MHz. The zones with higher energy at 125 MHz are interpreted as higher frequencies generated from thinly stratified zones within the tephra deposits. Conclusions Joint time-frequency analysis of traces from numer ical simulations and GPR data indicates that this signal-processing meth od has potential for locating zones of thin and thick beds within GPR reflection profiles. Thus JTFA adds information useful for stratigraphic interpretation of radar data. The S-transform used here for the time-frequency representation sat isfactorily resolves thick bedthin bed transitions in both time and frequency dom ains. Analysis of model and real data shows that layered sequences of beds thin ner than the tuning thickness have an overall rise in frequency. This spectral si gnature is not limited to mm-
146 scale beds but extends to cm-scale beds as well for antenna frequencies commonly used in geological GPR surveys. While elev ated frequencies in the GPR response may signal the presence of thin layers no simple methods exist for extracting more quantitative information on the thickness of such thin layers.
147 Figure 8.1. Schematics of the fi rst four models to illustrate the bed thickness distribution within each sequence. Note sequences a re longer than shown in schematics.
148 Figure 8.2. JTFA of 1D FDTD model traces of layered sequences w ith input frequency of ~200 MHz. (The first 10 ns record the outgoing pulse.) Beds in all sequences are separated by a 1-mm layer representin g a contact between successive beds. Upper panel of each figure shows t he model result over layered sequences. Lower panels represent corresponding S-t ransform of the model results. Results described in text.
149 Figure 8.3. (a) GPR reflection profile over hurricane washover deposit in Santa Rosa Island, Florida (modified from Wang and Horwit z, 2007). Black box indicates trench position. Red line (Iv) demarcates the Ivan washover deposits (Iw) from the underlying pre-Ivan sediments. Blue l ines indicate the trace positions that are extracted for analysis. (b) Phot o showing packages of dipping beds in homogeneous quartz sand. Close inspection r eveals internal layers composed of thin beds (~2 cm). (c) Amplitude spectr a of the downgoing GPR pulse shows a dominant frequency of 290 MHz. All Stransform results of GPR traces are compared with respect to this dominant f requency. (d) Traces extracted from GPR profile and their S-transforms. Dashed red lines mark the contact of Ivan deposits (Iw) and the pre-Ivan depo sits. Dashed blue lines indicate the water-table reflector. Results of JTFA discussed in text.
150 Figure 8.4. (a) GPR profile over tephra deposit at Cerro Negro volcano, Nicaragua. Data at center of profile were collected with 100-MHz frequency. Data at both ends were collected with 200-MHz frequency. The deposit represents a sequence of eruptive events. The base 1850 reflecto r is believed to be underlain by lava flows and older tephra. (b) Top panel shows 100-MHz trace from reflection profile. Bottom panel shows its S-transf orm. Contacts marking the base of major deposits, such as the 1992 and 1850 events are characterized by lower frequencies than returns from within depositional u nits. Thin beds within the tephra deposit correspond to higher frequencies. (c ) Map of relative energy at 125 MHz on GPR profile (4A). Zones with internal la yers have stronger returns at 125 MHz (red) than does the base 1850 reflector (we aker blue response).
151 LIST OF REFERENCES Ali, A., 1999. Climate change impact and adaptation assessment in Bangladesh, Climate Research, vol. 12: 109-116. Anderson Jr. W. P., 1999. The hydrology of Hatteras Island, North Carolina. PhD Diss. North Carolina State University, Rayleigh. Anderson, M.P. and W.W. Woessner, 1992. Applied gro undwater modeling: Simulations of flow and advective transport, New York, Academic Press. Anderson Jr., W.P., D. G. Evans and S.W. Snyder, 20 00. The effects of Holocene barrier island evolution on water-table e lveations, Hatteras Island, North carolina, USA. Hydrogeology Journal, vol. 8: 390-404. Anderson, Jr, W. P. and R. M. Lauer, The role of ov erwash in the evolution of mixing zone morphology within barrier islands, Hyd rogeology Journal, vol. 16: 1483-1495. Anthony, S.S., F. L. Peterson, F. T., Peterson and S.C. Hamlin, 1989, Geohydrology of the Laura fresh-water lens, Majuro atoll: A hydrochemical approach, Geological Society of Amer ica Bulletin, vol. 101: 1066-1075. Antonellini, M., P. Mollema, B. Giambastiani, K. B ishop, L. Caruso, A. Minchio, L., Pellegrini, M. Sabia, E. Ulazzi, and G. Gabb ianelli, 2008. Saltwater intrusion in the coastal aquifer of the southern P o Plain, Italy, Hydrogeology Journal, vol. 16: 1541-1556. Arnell, N. W., 2004. Climate change and global wate r recourses: SRES emissions and socio-economic scenarios, Global env ironmental change, vol:14, 31-52. Arnell, N.W., 1999a. Climate change and global wate r resources, Global Environmental Change, vol. 9: S31S49. Ataie-Ashtiani, B., R. E. Volker, and D.A., Locking ton, 1999. Tidal effects on sea water intrusion in unconfined aquifers, Journal o f Hydrology, vol. 216:1731.
152 Ayers, J. F. and Vacher, H. L., 1986. Hydrogeology of an atoll island: a conceptual model from detailed study of a Micrones ian example, Ground Water, vol. 24:185-198. Bailey, R. T., J. W. Jenson, and A. E., Olsen, 2009 Numerical modeling of atoll island Hydrogeology, vol. 47: 184-196. Barlow, P.M. and Reichard, E. G., 2010, Saltwater i ntrusion in coastal regions of North America, Hydrogeology Journal, vol. 18, 247 -260. Bear, J., 1988. Dynamics of fluids in porous medium Dover Publications, New York. Bear, J. and Dagan, G., 1968. Solving the problem o f local interface upconing in a coastal aquifer by the method of small perturbat ions, Journal of Hydraulic Research, vol. 1 Bocanegra, E., C. De Silva, G. Custodio, E. M. Man zano and S. Montenegro, State of knowledge of coastal aquifer management i n South America, Hydrogeology Journal, vol. 18: 261-267. Bobba, A. G ., 1993. Mathematical models for saltwater intrusion i n coastal aquifers, Water Resources Management, vol.7: 3-37. Bobba, A.G., 1998. Application of a numerical model to predict freshwater depth in islands due to climate change: Agatti island, In dia, Journal of Environmental Hydrology, vol. 6: 6-13. Bobba, A. G., 2002. Numerical modeling of saltwater intrusion due to human activities and sea-level change in the Godavari De lta, India, Hydrogeological Sciences Journal, vol.47: S67-S80 Cabanes, C, A. Cazenave, C. Le provost, 2001. Sea l evel rise during past 40 years determined from satellite and in situ obser vations, Science, vol. 294: 840-842. Calvache, M. L. and A. Pulido-Bosch, 1995. Effects of geology and human activity on the dynamics of saltwater intrusion in three coastal aquifers in southern Spain,Environmental Geology, vol. 30: 215-223. Cartwright, N., L. Li., and P. Nielsen, 2004. Respo nse of the salt-freshwater interface in a coastal aquifer to a wave-induced g roundwater pulse: field observations and modelling, Advances in Wate r Resources, vol. 27: 297-303.
153 Chapman, P.M., and F. Wang, 2000. Assessing sedimen t contamination in estuaries, Environmental Toxicology and Chemistry, vol. 20: 322 X. Cheng, Y. Zongb,1999. Major impacts of sea-level rise on agriculture in the Yangtze delta area around Shanghai Applied Geograp hy vol.19: 6984 Chen, X., Zong, Y., Zhang, E., Xu, J and Li, S., 20 01, Human impacts on the Changjiang (Yangtze) River basin, China with speci al reference to the impacts of the dry season water discharges into t he sea, Geomorphology, vol.41: 111-123. Choudhury, K., D.K. Saha and P.Chakraborty, 2001. G eophysical study for saline water intrusion in a coastal alluvial terrain, Jo urnal of applied geophysics, vol.46:189-200. Church, J.A. and N.J. White, 2006. A 20th century a cceleration in global sealevel rise.Geophysical Research Letters, vol. 33: L01602, DOI: 10.1029/2005GL024826. Church, J.A., White, N.J., Aarup, T., Wilson, W.S., Woodworth, P.L., Domingues, C.M., Hunter, J.R. and Lambeck, K. 2008: Understa nding global sea levels: past, present and future. Sustainability Science, vol. 3: 1167. Collins III, W.H. and D. H. Easley, 1999. Fresh-wat er lens formation in an unconfined barrier-island aquifer, Journal of the American Water Resources Association, vol.35: 1-21. Cooper, H. H., 1959. A hypothesis concerning the dy namic balance of freshwater and saltwater in a coastal aquifer, Journal of Geo physical Research, vol. 64: 461-467. Custodio, E., 2005. Coastal aquifers as important n atural hydrogeologic structures, Groundwater and Human Development, AA. Balkema Publ., Leiden, The Netherlands. Custodio, E., 2010. Coastal aquifers of Europe; an overview, Hydrogeology Journal, vol.18: 269-280. Dalrymple, R. W., B. A. Zaitlin and R. Boyd, 1992. Estuarine facies models: conceptual basis and stratigraphic implications, J ournal of Sedimentary Petrology, vol. 62: 1130-1146. Dawson, R. J., J. W. Hall, P. D. Bates and R. J. Ni cholls, 2005. Quentified analysis of the probability of flooding in the Tha mes Estuar under imaginable worst-case sea level rise scenarios, Wa ter Resources Development, vol. 21: 577-591.
154 Dausman, A. M. and C. D. Langevin, 2005. Movement o f the saltwater interface in the Surficial Aquifer System in response to hy drologic stresses and watermanagement practices, Broward County, Flori da: USGS scientific investigation report 2004-5256. Davis, R. A., 1985. Coastal sedimentary environment s, Springer-Verlag, New York. Davis, R. A. and D. M. Fitzgerald, 2004. Beaches an d coasts, Blackwell Sciences Ltd. Delonom, R. M., 2008. Groundwater management issues in the Greater Jakarta area, Indonesia, Proceedings of Intl. Workshop on Integrated watershed management for sustainable water use in humid tropical region. Dentz, M., Tartakovsky, M., Abarca, E., Guandagnini A., Shanchez-Villa, X. and Carrera, J., 2006, Variable-density flow in porous media, Journal of Fluid Mechanics, vol. 561:209-235. de Montety V., O. Radakovitch, C. Vallet-Coulomb, B Blavoux, D. Hermitte, V. Valles, 2008. Origin of groundwater salinity and h ydrogeochemical processes in a confined coastal aquifer: Case of t he Rhne delta (Southern France) Applied Geochemistry, vol. 23: 2 337-2349. Diersch, H. J. G. and O. Kolditz, 2002. Variable-de nsity flow and transport in porous media: approaches and challenges, Advances in water resources, vol. 25: 899-944. Dyer, K. R., 1972. Sedimetation in estuaries, in Ba rnes, R. S. R., and Green, J., (eds.), Chemistry and biochemistry of estuaries, W iley, Chichester, 121151. Edwards, R, 2008. Sea levels: science and society, Progress in Physical Geography, vol. 32, 557-574. Elewa, H. H. and A. H. El Nahry, 2009.Hydro-environ mental status and soil management of the River Nile Delta, Egypt, Environ mental Geology, vol. 57: 759-774. Ericson, J.P., C. J. Vrsmarty, S. L. Dingman, L. G. Ward, M. Meybeck, 2006. Effective sea-level rise and deltas: Causes of cha nge and human dimension implications, Global and Planetary Chang e, vol. 50: 6382
155 Falkland, A.C., 1983. Christmas Island (Kiritimati) Water Resources Study, vol. 1. Victoria, Australia Department of Housing and Cons truction. Fairbridge, R. W., 1980. The estuary: its definitio n and geodynamic cycle, in Olausson, E., and Cato I., (ed.), Chemistry and bi ogeochemistry of estuaries, 1-36. Fetter, C. W., 1972. Position of the saline water i nterface beneath oceanic islands. Water Resources Research, vol. 8: 1307-13 15. Fetter, C.W., 1988. Applied Hydrogeology, Merrill P ubl. Co. Columbus, Ohio. Fisk, H. N., E. McFarlan, C. R. Kolb and L. J. Wilb ert, 1954. Sedimentary framework of the modern Mississippi delta, Journal of sedimentary petrology, vol. 24: 76-99. Fisher, J.J., 1968. Barrier island formation: discu ssion. Geol Soc Am Bull, vol. 79: 1421-1426. Galloway, W.E., 1975. Process framework for describ ing the morphologic and stratigraphic evolution of deltaic depositional sy stems: Deltas, models for exploration, Houston geological Society. Houston, 87-98. Giambastiani B. M. S., M. Antonellini, H.P. Oude Es sink R. J. Stuurman, 2007. Saltwater intrusion in the unconfined coastal aqui fer of Ravenna (Italy): A numerical model, Journal of Hydrology, vol. 340: 9 1 104 Gibson J. R. and R. G. Najjar, 2000. The response o f Chesapeake Bay salinity to climate-induced changes in streamflow, Limnology O ceanography, vol.54:1764-1772. Goodbred, S. L., and S. A., Kuehl, 2000. Enormous G anges-Brahmaputra sediment discharge during strengthened early Holoc ene monsoon. Geology, vol. 28: 1083-1086. Grabemann, H. J., I. Grabemann, D. Herbers, and A. Muller, 2001. Effects of a specific climate scenario on the hydrography and t ransport of conservative substances in the Weser estuary, Germany: a case s tudy, Climate Research,vol. 18: 77-87. Griggs, J. E. and F. L. Peterson, 1993. Ground-wate r flow dynamics and development strategies at the atoll scale, Ground Water, vol. 31:209-220. Gross, E. S., M. L. MacWilliams and W. Kimmerer, 20 06. Simulated periodic startification in San Francisco bay, Proceedings o f the Estuarine and Coastal Modeling Conference, ASCE.
156 Glover, R. E., 1959. The pattern of freshwater flow in a coastal aquifer, Journal of Geophysical Research, vol. 64: 457-459. Guo, W. and Langevin, C. D., 2002. Users guide to SEAWAT: A computer program for simulation of three-dimensional varia ble-density groundwater flow, USGS OpenFile Report, 01-434. Ghyben, W. B., 1889. Notes on the probable results of well drilling near Amsterdam.Tijdschrift van het Koninklijk Inst. van Ingenieur, The Hague, 21. Hantush, M. S., 1968. Unsteady movement of freshwat er in thick unconfined saline aquifers. Bull. Int. Assoc. Sci. Hydrol. v ol. 13: 40-60. Harvey N., 2006. Rates and impacts of global sea-le vel change. In: Glazer MP (eds) New frontiers in environmental research, Ch ap 1, Nova Science Publishers, New York Harvey, N. and R. Nicholls, 2008. Global sea-level rise and coastal vulnerability, Sustainability Science, vol. 3: 5-7. Harris, P. J. 1985. Water quality impacts from on-s ite waste disposal systems to coastal areas through groundwater discharge. Envir onmental Geology, vol. 26: 262-268. Hassan, M. Q., 1992. Saline water intrusion and hyd rogeological modeling of SW Bangaldesh. Report. Hayes, M. O., 1975. Morphology of sand accumulation s in estuaries, in Cronin, L. E., (ed.), estuarine Research, vol. 2: 3-22, Geolo gy and Engineering, Academic Prfess, New York. Heath, R. W., 1988. Ground water resources of the c ape hatteras area of North Carolina. Report to the Cape hatteras Water Associ ation. Henry, H. R., 1960. Saltwater intrusion into coasta l aquifer, Int. Ass. Sci. Hydrol. Publ. vol. 52: 478-487. Herman, M. E., and S.W. Wheatcraft, 1984. Groundwat er dynamic investigation of Enjebi Island, Enewetak Atoll: An interpretive computer model simulation. In Finite Elements in Water Resources : Proceedings, vol. 5, 133-142, ed. J.P. Laible, C.A. Brebbia, W Gray, G. Pinder. London, England: Springer-Verlag.
157 Herzberg, B., 1901. Die Wasserversorgung einiger No rdseebadar, J. Gasbeleuchtung und Wasserversorgung, vol. 44: 815819. Hilton, T. W., R. G. Najjar, L. Zhong and M. Li, 20 08. Is there a signal of sea-level rise in Chesapeake bay salinity?, Journal of Geophy sical Research, vol. 113: C09002, doi:10.1029/2007JC004247. Hubbart, M. K., 1940. The theory of groundwater mot ion, Journal of Geology, vol.48, 785-944. Intergovernmental Panel on Climate Change, 2007, Cl imate Change 2007, Synthesis Report. Intergovernmental Panel on Climate Change, 2008, Cl imate Change and Water, IPCC technical paper VI. Jeng, D.-S., X. Mao, P. Enot, D. A. Barry, and L. L i 2005, Spring-neap tideinduced beach water table fluctuations in a slopin g coastal aquifer, Water Resources Research, vol. 41: W07026, doi:10.1029/ 2005WR003945 Jevrejeva, S., J.C. Moore, A. Grinsted and P.L. Woo dworth, 2008: Recent global sea level acceleration started over 200 years ago? Geophysical Research Letters 35, L08715, doi: 10.1029/2008gl033611. Johnson, B. H., R. E. Heath, B. B. Hsieh, K. W. Kim and H. L. Butler, 1991. Development and verification of a three-dimensiona l numerical hydrodynamic, salinity and temperature model of Ch esapeake Bay, vol. 1 Kashef, A. I., 1983. Salt-water intrusion in the Ni le delta, Groundwater, vol. 21: 160-167. Kidd, R.E. and M. Planert, 1985. Development of a f resh water supply from the watertable aquifer on a barrier island. American Water Resources Association Technical Publication Series 85-1: 6971. Kumar, C.P., A. G. Chachadi, B. K. Purandara, S. Ku mar, and R. Juyal, 2007. Modeling of seawater intrusion in coastal area of North Goa. Kugler, N. B., 1998. Numerical simulation of freshw ater lenses in barrier islands, MS thesis, University of South Florida. Langevin, C.D., 2003. Simulation of submarine groun dwater discharge to a marine estuary: Biscayne Bay, Florida, Groundwater vol. 41: 758-771.
158 Li, H., J. J., Jiao, M. Luk, and K. Cheung, 2002. T ide-induced groundwater level fluctuation in coastal aquifers bounded by L-shape d coastlines, Water Resources Research, vol. 38: 1024, 10.1029/2001WR0 00556. Lyoyd, J.W., J.C., Miles, G.R. Chessman, and S.F. B ugg, 1980. A groundwater resources study of a Pacific Ocean atollTarawa, Gilbert Islands. Water Resources Bulletin, vol. 16, 646-653. Masterson, J. P. and S. P. Garabedian, 2007, Effect s of sea-level rise on groundwater flow in a coastal aquifer system, Gro undwater, vol. 45, 209217. Meehl, G. A., Washington, W. M., Collins, W. D., Ar blaster, J. M., Hu, A., Buja, L. E., Strand, W. G. & Teng, H. 2005 How much more gl obal warming and sea level rise? Science 307, 17691772. (doi:10.11 26/science.1106663) Michael, H. A. and C. I. Voss, 2009a. Estimation of regional-scale groundwater flow properties in the Bengal Basin of India and B angladesh, Hydrogeology Journal, vol.17: 1329-1346. Michael, H. A. and C. I. Voss, 2009b. Controls on g roundwater flow in the Bengal Basin of India and Bangladesh: regional modeling a nalysis, Hydrogeology Journal, vol. 17: 1561-1577. Milzow, C., L. Kgotlhang, 2009. Regional review: th e hydrology of the okavango delta, Botswana-processes, data and modeling, Hydr ogeology journal, vol. 17: 1297-1328. Moore, L. J., J. H. List, S. J. Williams and D. Sto pler, 2007. Modeling barrier island response to sea-level rise in the Outer Ban ks, North Carolina, Coastal Sediments, ASCE. Nicholls, R.J., 1995 Coastal megacities and climate change. Geojournal vol.37 369379. Nichols, M. M., and R. B. Biggs, 1985. Estuaries, i n Coastal sedimentary environments, Davis, R. J., (ed.), Springer-Verlag New York, 80-186. Nicholls R. J. and J. A. Lowe 2004. Benefits of mit igation of climate change for coastal areas, Global Environmental Change, vol.14 : 229-244. Nicholls, R. J., Wong, P.P., Burkett, V., Woodroffe C. D and Hay, J., Climate change and coastal vulnerability assessment: scena rios for integrated assessment, Sustainability Science, vol. 3: 89-10 2.
159 Nobi, N. and A. Dasgupta, 1997. Simulation of regio nal flow and salinity intrusion in an integrated stream-Aquifer system in coastal region: southwest region of Bangladesh, Groundwater, vol. 35: 786-79 6. Oertel, G. F., 1985. The barrier island system, Mar ine Geology, vol. 63: 1-18. Oude Essink, G. H. P., 2001, Improving fresh ground water supply-problems and solutions, Ocean and Coastal Management, vol.41, 4 29-449. Oberdorfer, J. A., and R.W. Buddemeier, 1988, Clima te change: Effects on reef island resources. In Proceedings; Sixth Internatio nal Coral Reef Symposium, vol. 3, 523-527, ed. J.H. Choat, D. Barn es, M.A. Borowtizka, J.C. Coll, P.J. Davies, P. Flood, B. G. Hatcher, D. Hopley, P.A. Hutchings, D. Kinse, G.R. Orme, M. Pichon, P.F. Shale, P. Samm arco, Wallace, C. C., C. Wilkinson, E. Wolanski and O. Bellwood. Miam i, Florida, University of Miami. Penland, S., R. Boyd, and J. R. Suter,1988. Transgr essive depositional systems of the Mississippi delta plain: a model for barrie r shoreline and shelf sand development, Journal of sedimentary petrology, vol 58: 932-949. Peterson, F.L. 1997. Hydrogeology of the Marshall I slands, in Geology and Hydrogeology of Carbonate Islands. Developments in Sedimentology, vol. 54, 611-666 ed. H. L. Vacher and T. Quinn, Amsterd am, the Netherlands: Elsevier. Pfeffer, W.T., J. J. Harper, and S. ONeal, 2008. K inematic constraints on glacier contributions to 21st-century sea-level rise, Science, vol. 321: 1340-13 43. DOI: 10.1126/science.1159099 Pilkey, O. H. and J. A. G. Cooper, 2004. Society an d sea level rise, Science, vol., 303:1781-1782. Pinder, G. F. and H. H. Cooper, 1970. A numerical t echnique for calculating the transient position of the saltwater front, Water R esource Research, vol. 6: 875-882. Post, V. and E. Abarca, 2010. Preface: saltwater an d freshwater interactions in coastal aquifers, Hydrogeology Journal, vol.18:1-4 Presley, T.K., 2005. Effects of the 1998 drought on the freshwater lens in the Laura area, Majuro Atoll, Republic of the Marshall Islands. USGS Scientific Investigation Report 2005-5098. Reston, Virginia: USGS. Prtichard, D. W., 1955. Estuarine circulation patte rns. Proc. Amer. Soc. Civil Engr. vol. 81: 717-1-717-11.
160 Pujari, P. R. and A. K. Soni, 2009. Seawater intrus ion studies near Kovaya limestone mine, Saurashta coast, India, Environ. M onit. Assess., vol.154: 93-109. Rahmstorf, S. 2007. A semi-empirical approach to pr ojecting future sea-level rise.Science vol. 315: 36870. Rajesh, R. and T. R. S. Murthy, 2004, Groundwater q uality and its change over a decade:An analysis of coastal urban environment fr om the west coast of India, Environmental Geology, vol. 45: 978-981. Reading, H. G., 1986. Sedimentary environments and facies, Blackwell Scientific Publ. Reilly, T. E. and A. S. Goodman, 1985. Quantitative analysis of saltwaterfreshwater relationships in groundwater systems a historical perspective, Journal of Hydrology, vol. 80: 125-16 0. Riggs, S. R., W. J., Cleary and S. W. Snyder, 1995. Influence of inherited geologic framework on barrier shoreface morphology and dynamics. Marine geology, vol. 126: 213-234. Robinson, C., B. Gibbes, and L., Li 2006. Driving m echanisms for groundwater flow and salt transport in a subterranean estuary, Geophysical Research Letters, vol. 33. Rosati, J. D., 2009. Barrier island migration over a consolidating substrate, PhD Diss. Louisiana State Univeristy. Rozell, D. J., 2007. Quantifying the impact of glob al warming on saltwater intrusion at Shelter Island, New York using a grou ndwater flow model, M.S. thesis, Stony Brook University Ruppel, C., G. Schultz, and S. Kruse, 2000. Anomalo us freshwater lens morphology on a strip barrier island. Groundwater, vol. 38: 872-881. Sanz, E. and C. I. Voss, 2006. Inverse modeling for seawat intrusion in coastal aquifers: Insights about parameter sensitivities, variances, correlations and estimation procedures derived from the Henry p roblem, Advances in water resources, vol.29: 439-457. Schneider, J.C. and S.E. Kruse, 2003. A comparison of controls on freshwater lens morphology of small carbonate and siliciclast ic islands: examples from barrier islands in Florida, USA, Journal of H ydrology, vol. 284: 253269
161 Servan-Camas, B. and F. T. C. Tsai, 2009. Saltwater intrusion modeling in heterogeneous confined aquifers using two-relaxati on-time lattice Boltzmann method, Advances in water resources, vol 32: 620-631. Shanmungam, P., S. Neelamani, Y. Ahn, L. Philip, an d G. Hong, 2007. Assessment of the levels of coastal marine polluti on of Chennai city, southern India, Water Resources Management, vol. 2 1: 1187-1206. Shammas, M. I. and G. Jacks, 2007. Seawater intrusi on in the Salalah plain aquifer, Oman, Environmental Geology, vol. 53: 575 -587. Sherif M.M. and V. P. Singh, 1999. Effect of climat e change on sea water intrusion in coastal aquifers Hydrological Process es, vol. 13: 1277-1287. Shindell, D., 2007. Estimating the potential for tw enty-first century sudden climate change, Philosophical Transactions of the Royal So ciety of America, vol. 365,2675-2694, doi:10.1098/rsta.2007.2088 Simmons, C. T., 2005. Variable density groundwater flow: From current challenges to future possibilities, Hydrogeology J ournal, vol. 13: 116-119. Simmons, C. T., T. R. Fenstemaker, and J. M. Sharp, 2001. Variable-density groundwater flow and solute transport in heterogen eous porous media: approaches, resolutions and future challenges, Jou rnal of Contaminant Hydrology, vol. 52: 245-275. Simmons, D. L., 1986. Geohydrology and ground-water quality on Shelter Island, Suffolk County, New York, 1983-84. USGS Water Reso urces Investigations Report 85-4165. Small C., and R. J. Nicholls, 2003. A global analys is of human settlement in coastal zones, Journal of Coastal Research vol.19: 584599. Stanley, D. J., 1990. Recent subsidence and northea st tilting of the Nile delta, Egypt, Marine Geology, vol. 94: 147-154. Steyl, G. and I. Dennis, 2010. Review of coastal ar ea aquifers on Africa, Hydrogeology Journal, vol. 18: 217-225. Syvitski, J. P. M., 2007. Deltas at risk, Sustainab ility Science. Underwood, M. R. 1990. Atoll island hydrogeology: C onceptual and numerical models.Ph.D. Diss., Department of Geology and Geop hysics, University of Hawaii at Manoa, Manoa, Hawaii.
162 Underwood, M. R., F.L. Peterson, and C.I. Voss, 199 2. Groundwater lens dynamics of atoll islands, Water Resources Resear ch, vol. 28: 28892902. Vacher, H. L., 1997. Introduction: Varieties of car bonate islands and a historical perspective, in Geology Hydrogeology of Carbonate Islands. Developments in Sedimentology, vol. 54: 1-33 ed. H L. Vacher and T. Quinn, Amsterdam, the Netherlands: Elsevier. Vacher, H. L, 1988. Dupuit-Ghyben-Herzberg analysis of strip-island lenses. Bulletin Geological Society of America, vol. 100: 580-591. Vacher H. L. and T. Quinn, 1997. Hydrogeology of Ca rbonate Islands. Developments in Sedimentology, vol. 54: Amsterdam, the Netherlands: Elsevier. van Dam, J., 1999. Impacts of climate change and cl imate variability on hydrological regimes. Cambridge University Press. Vandenbohede, A. and L. Lebbe, 2006. Occurrence of saltwater above freshwater in dynamic equilibrium in coastal groun dwater flow system near De Panne, Belgium, Hydrogeology Journa l, vol. 14: 462-472. Vorosmarty, C.J., P.J. Green, J. Salisbury, R.B. La mmers, 2000. Global water resources: vulnerability from climate change and population growth, Science vol. 289: 284288. Vorosmarty, C. J., M. Meybeck, B. Fekete, K. Sharma ., P. Green, and J. P. M. Syvitski, 2003. Anthropogenic sediment retention: major global impact from registered river impoundments, Global and Pla netary Change, vol. 39: 169-190. Wang, K.H., 1994. Characterization of circulation a nd salinity change in Galveston Bay, Journal of Enrg. Mech. vol. 120: 55 7-579. Walker, R. G. and James, N. P.,1992. Facies Models response to sea level change, Geological Association of Canada. Weber, E., R. Schultz and J. R. Garrison, 2008. Fre shwater lens development on Padre Island, Texas, 20th Saltwater Intrusion Meeting, Programs and Proceeding, 306-309. Wentworth, C.K., 1947. Factors in the behavior of g roundwater in a GhybenHerzberg system, Pacific Science, 172-184.
163 Werner, A.D., 2010. A review of seawater intrusion and its management in Australia, Hydrogeology Journal, vol. 18: 281-285. White, I. and Falkland, T, 2010. Management of fres hwater lenses on small Pacific islands, Journal of Hydrology, vol. 18: 22 7-246. Woodroffe, C.D., 2003. Coasts: form, process and ev olution. Cambridge University Press. Woodworth, P.L., N.J. White, S. Jevrejeva, S.J. Hol gate, J.A. Church, and W.R.Gehrels, 2008. Evidence for the accelerations of sea level on multidecade and century timescales. International Journ al of Climatology, vol. 29:777-789. Wool, T. A., 2003. Development of three-diemnsional hydrodynamic and water quality models to support total maximum daily load decision process for the Neuse river estuary, North Carolina, Journal o f water recourse planning and management, vol. 129: 295-306. Wu, J., Y. Xue, P.Liu, J.Wang, Q. Jiang, and H. Sh i, 1993. Sea-water intrusion in the coastal area of Laizhou Bay, China. Sea-water intrusion monitoring, Groundwater, vol. 31:740-746. Zheng, C. and P.P. Wang, 1999. MT3DMSA modular thr ee-dimensional multispecies transport model for simulation of adv ection, dispersion and chemical reactions of contaminants in ground-water systems; documentation and users guide: Jacksonville, Fla ., U.S. Army Corps of Engineers Contract Report SERDP-99-1. Zhou, X., M. Chen, and C. Liang, 2003. Optimal sche mes of groundwater exploitation for prevention of seawater intrusion in the Leizhou Peninsula in southern China, Environmental Geology, vol. 43: 978-985. Zimmermann, S., P. Bauer, R. Held, W. Kinzelbach, a nd J. H. Walthe, 2006. Salt transport on islands in the Okavango Delta: Numeri cal investigations, Advances in Water Resources, vol. 29: 11-29.
ABOUT THE AUTHOR Swagata Guha received MS in Geology from the Univer sity of Calcutta, India in 2001 and a second MS in Geology from the University of South Florida in 2004. She entered the PhD program at the University of So uth Florida in 2005 and has worked on near-surface geophysics and groundwater m odeling. Swagata has participated in and presented at several national a nd international conferences. She has contributed to peer-reviewed journals, revi ewed journal articles and is also the recipient of the GSA and GCAGS student gra nt awards.