Numerical modelling of surface water and groundwater flow and solute interactions between a river and a saline floodplain in a semi-arid region

Transcription

1 Numerical modelling of surface water and groundwater flow and solute interactions between a river and a saline floodplain in a semi-arid region by Sina Alaghmand B. Eng.; M. Eng. A thesis submitted for the degree of Doctor of Philosophy in Civil Engineering Centre for Water Management and Reuse School of Natural and Built Environments September 2014

2 This work is submitted as a Thesis Containing Published Research in accordance with UniSA Academic Regulations for Higher Degrees under research clause b ( iii

3 Table of content List of Figures...ix List of Tables...xvii Summary... xviii Declaration... xx Acknowledgments... xxi Chapter 1 Introduction 1.1 Floodplain salinization Salt interception measures Numerical modeling Motivation and aims of the research Structure of the thesis Thesis overview Contribution to knowledge List of publications..10 Chapter 2 Review of SW-GW interactions and salt mobilization Overview Paper 1: A Review of the Numerical Modelling of Salt Mobilization from Groundwater-Surface Water Interactions Introduction Surface water-groundwater interactions Physically-based numerical models Coupling of surface-subsurface domains Density-dependent flows Floodplain salinity and vegetation health Effects of river/floodplain geometry Conclusions iv

4 Chapter 3 Modelling the impacts of river stage manipulation on a riverfloodplain system Overview Paper 2: Modelling the Impacts of River Stage Manipulation on a Complex River-Floodplain System in a Semi-Arid Region Introduction Floodplain salinization in arid and semi-arid regions Physical processes Research challenges Objective Materials and methods Study site Numerical model Model set up Geometry grid Parameters Boundary conditions Initial conditions Coupled flow and transport calibration Numerical model performance evaluation Scenarios Results and discussion Calibrated model Water balance Solute mass balance Solute mass in the unsaturated zone Ecological implications Conclusion Chapter 4 Quantifying the impacts of artificial flooding on a river-floodplain interaction Overview Paper 3: Quantifying the impacts of artificial flooding and groundwater lowering on river-floodplain interaction in semi-arid saline floodplain v

5 4.1 Introduction Methods Study site Field data Numerical model set-up HydroGeoSphere Governing Equations Model domain and grid Domain properties Boundary conditions Initial conditions Scenarios Model Calibration Sensitivity and uncertainty analysis Results Discussion Combination of artificial flooding and groundwater lowering Impacts of artificial flooding Impacts of groundwater lowering Conclusion Chapter 5 Impacts of groundwater extraction on floodplain salinization risk Overview Paper 4: Impacts of groundwater extraction on salinization risk in a semiarid floodplain Introduction Study site Numerical model Model set-up Coupled flow and transport calibration Results and discussion Water balance Solute mass balance Conclusion vi

6 Chapter 6 Injection of fresh river water into a saline floodplain aquifer Overview Paper 5: Injection of fresh river water into a saline floodplain aquifer as a salt interception measure in a semi-arid environment Introduction Material and methods Field trial setup Base case model Scenarios Numerical model development HydroGeoSphere Geometry grid Model parameters Boundary and initial conditions Model calibration Results and discussion Base case model Scenarios Injection rate Injected volume Injection screen depth Injection pumps configuration Solute mass mobilization Conclusion Chapter 7 Impacts of vegetation cover on SW-GW flows and solute interactions Overview Paper 6: Impacts of Vegetation Cover on Surface-Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia Introduction Material and methods Governing equations vii

7 7.2.2 Study site Numerical model set-up Model calibration Results and discussion Conclusion Chapter 8 Conclusions 8.1 Summary of key findings Recommendations for future studies viii

8 List of Figures Figure 1.1 Structure of the thesis..7 Figure 2.1 Schematic diagram of processes leading to dryland salinity after the clearance of native vegetation for agricultural practices (Leblanc et al., 2012)...20 Figure 2.2 Conceptual model of groundwater inputs to the floodplain and potential groundwater discharge pathways within the floodplain in the Lower River Murray (Holland et al., 2009b).. 22 Figure 2.3 Geological cross section (a) and cross-section of the conceptual model (b) of Clark s Floodplain at Lower River Murray (Doble et al., 2006) Figure 2.4 The coupling approach and interfaces of the IWAN model components incorporating WASIM-ETH for the simulation of the runoff generation and water balance of the unsaturated soil zone and MODLFOW for the groundwater dynamics and interactions with the surface water (Krause et al., 2007b) Figure 3.1 Schematic of the SW-GW interaction across the Lower Murray River area before (a) and after (b) human-induced activities including weir and lock installations and irrigation practices..60 Figure 3.2 a: Location of Clark s floodplain in Australia (shown in purple), b: Perimeter of the geometry model (shown in red), c: 3D visualization of the geometry of the study site including the soil types and observation (B1, B2, B3, B4, B5, B6, 31F, 33F and 35F) and SIS production wells (32F and 34F) (Z magnification= 8). The cover image is adopted from GoogleMaps Figure 3.3 a: Configuration of the model boundary conditions, b: Configuration of the vegetation and soil layers of Clark s Floodplain along transect 1 (Z magnification= 3). Observation wells are shown as black columns...71 Figure 3.4 Time-varying river stage boundary conditions for scenarios featuring river stage rise durations of (a) one month, (b) two months, and (c) three months.. 74 Figure 3.5 Modelled and observed groundwater heads at the observation wells. River stage and modelled and observed groundwater heads are shown as blue lines, black lines and red dots, respectively. The ix

9 light blue pattern represents the periods during which the SIS production wells were in operation Figure 3.6 a: Modelled groundwater salinity distribution (November 2007, time step 650 days), b: Conductivity distribution, EM31 survey in November 2007 (Berens et al., 2009), c: Recorded groundwater salinity during the study period (Holland et al., 2013).77 Figure 3.7 Groundwater balance for the calibrated model (a) and No-SIS scenario (b) 78 Figure 3.8 Solute mass balance for Only-SIS scenario (a) and No-salt management scenario (b). 79 Figure 3.9 Solute mass extracted via the SIS production wells during the study period Figure 3.10 Groundwater salinity along transect 1 for Only-SIS and No-salt management scenarios (Z magnification: 3) at time step 2070 days (2/09/2010) Figure 3.11 Dynamics of GW heads at the observation wells on Transect 2, a: 1.5 m rise for 3 months, b: 0.5 m drop for 3 months and c: 1.5 m rise for one, two and three months at observation well B Figure 3.12 a: Change in water storage in the floodplain aquifer, b: Flux exchange between the river and the floodplain aquifer and c: Floodplain aquifer recharge from regional groundwater. All the results shown here are for the three month scenarios. The blue and yellow patterns represent groundwater lowering and river stage manipulation, respectively..84 Figure 3.13 a: Total solute mass in the floodplain aquifer and b: Change in stored solute mass in the floodplain aquifer. Both are for the three month scenarios. The blue and yellow patterns represent groundwater lowering and river stage manipulation, respectively.87 Figure 3.14 Spatial distribution of modelled solute concentration along transect 1 during the 3rd trial (time step 1480 (20/01/2009)) for the 1.5 m rise and 0.5 m drop three month scenarios. Observation wells are shown in black...88 Figure 3.15 Visualization of the solute concentration distribution in the floodplain aquifer for Only-SIS (a) and No-salt management scenarios (b) along transect x

10 Figure 3.16 Visualization of the solute concentration distribution in the floodplain aquifer for the three month long +1.5 m (a) and -0.5 m (b) scenarios at time-step 1120 days (just after the 2nd trial) along transect Figure 3.17 Visualization of distribution of solute mass mobilization in the floodplain aquifer for the three month long, +1.5 m scenario at time-step 1120 days (just after the 2nd trial) along transect Figure 4.1 Conceptual model of the river-floodplain system influenced by human-induced activities (Adopted from Alaghmand et al. (2014a)).107 Figure 4.2 Configuration of the study site. Observation wells and SIS production wells are shown in red and blue circles, respectively. The yellow line indicates the floodplain depression. The green diamonds represent river salinity loggers. Inset map shows the location of the study site in Australia Figure 4.3 Clark s Floodplain depression during the second trial in August 2005 (a: River Murray water is pumped to the depression over the constructed earthen bank, b. the floodplain depression is filled with the river water) (Holland et al., 2009) Figure 4.4 a: 3D illustration of the study site (Z magnification: 5). Observation wells and SIS production wells are shown in red and blue circles, respectively. The green line indicates the extent of the floodplain depression. b: Conceptual configuration of the surface domain including the river (shown in red), the floodplain and the highland (shown in green). c: Soil layers and conceptual vegetation cover along Transect 2. Red columns represent the observation wells Figure 4.5 Configuration of the boundary conditions of the model domain 119 Figure 4.6 a: Simulated pre-trial (June 2005) salinity distribution; b: Correlation between the simulated and observed groundwater salinity in June 2005 at the location of the observation wells Figure 4.7 Simulated and observed groundwater heads at the observation wells. River levels and simulated and observed groundwater heads are shown as blue lines, black lines and red dots, respectively. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively Figure 4.8 Results of sensitivity of groundwater head and salinity to the tested parameters xi

11 Figure 4.9 Change in water storage in the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively Figure 4.10 Total water flux from the surface domain to the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively..128 Figure 4.11 Total solute mass stored in the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively Figure 4.12 Groundwater salinity along transect 2 at time steps 450, 520 and 600 days (Z magnification: 3). 133 Figure 4.13 Total solute mass stored in the floodplain depression profile for the defined scenarios during the simulated period. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively Figure 4.14 Temporal distribution of electrical conductivity using Geonics EM31 at the study site (Holland et al., 2009) Figure 4.15 River level and simulated groundwater heads at the location of observation wells A8 and A12 for the defined scenarios. The blue pattern represents the groundwater lowering period..137 Figure 4.16 Groundwater salinity along transect 2 for the No manipulation and Only SIS scenarios (Z magnification: 3) 138 Figure 5.1 Conceptual model of groundwater inputs to the floodplain and potential groundwater discharge pathways within the floodplain in the Lower River Murray (Holland et al., 2009a) Figure 5.2 Configuration of SIS production wells (in blue) and observation wells (in red) at the Clark s Floodplain. The inset map shows the location of the Bookpurnong floodplain in Australia..156 xii

12 Figure 5.3 Configuration of boundary conditions for the river, floodplain and groundwater domains (Z magnification: 10) Figure 5.4 3D demonstration of simulated initial condition along transects B1 and B2: a. Porous media saturation, b. Solute concentration distribution. Observation wells are in black (Z magnification: 5) Figure 5.5 Simulated and observed groundwater heads at observation wells..164 Figure 5.6 Simulated solute concentration distribution (a) and EM31 survey (Berens et al., 2009) (b) in November 2007 at the study site Figure 5.7 Groundwater heads at the boundary of the models (SIS wells) for the defined scenarios Figure 5.8 Changes in water storage in the porous media for the defined scenarios (light blue pattern refers to the period that the pumps were in operation)..168 Figure 5.9 Water flux from the river to the floodplain aquifer for the defined scenarios (light blue pattern refers to the period that the pumps were in operation) Figure 5.10 Cumulative evaporation from the floodplain aquifer for the defined scenarios Figure 5.11 Groundwater head dynamics at the observation wells on transect B1 for the with-sis and the without-sis scenarios Figure 5.12 Groundwater head longitudinal profiles on 7/03/2007 (left), 16/12/2008 (middle) and 29/07/2009 (right) on transect B2 for the with-sis and the without-sis scenarios Figure 5.13 Solute mass stored in the system in each time step for the defined scenarios. Cumulative pumped water is also shown in dark blue and light blue pattern refers to the period that the pumps were in operation..171 Figure D visualization of solute mass changes in the unsaturated zone for the defined scenarios; a. amount of solute mass removed from the unsaturated zone during the with-sis scenario, b. amount of solute mass that could be stored in the system if SIS was not installed on the floodplain xiii

13 Figure 5.15 Solute dynamics at the observation wells BO1 (left), BO2 (middle) and BO3 (right) Figure 5.16 Conceptual model of the impacts of groundwater extraction on salinization risk in a semi-arid floodplain Figure 6.1 Location of Site E on Clark s Floodplain on the Lower Murray River in South Australia (Purple dotted line shows the floodplain perimeter) Figure 6.2 Configuration of injection (blue dots) and observation (red dots) wells. Rectangle in yellow represents the perimeter of the injection zone Figure 6.3 Photo of the aquifer breach on 8 November 2006 next to injection well EI4 at Site E (Berens et al., 2009a) Figure 6.4 3D visualization of the geometric grid of the study site including soil types (Z magnification = 10). Red line in the inset map shows the perimeter of the model geometry grid. Blue and brown lines in the inset map show the location of constant first type (Dirichlet) and time-varying first-type (Dirichlet) boundary conditions..192 Figure 6.5 a: Recorded river stage, daily rainfall and daily ET during the trial; b: Total injection rate and injected volume of river water.195 Figure 6.6 Simulated groundwater salinity at the beginning of the study period (pre-trial) on 1/09/2006. Values in brackets represent the observed groundwater salinity Figure 6.7 Simulated and observed groundwater heads at observation wells EO1, EO3 and EO4. Light green pattern represents the injection trial period Figure 6.8 a: Plan view of simulated groundwater salinity distribution at Site E on 1/12/2006 (time step = 92 days); b: Simulated and observed groundwater salinity at observation well EO1. Light green pattern represents the injection trial period Figure 6.9 Simulated groundwater head for the injections with 1.25 l.s -1, 2 l.s - 1 and 5 l.s -1 injection rates at Site E Figure 6.10 Simulated groundwater salinity for the injection with 1.25 l.s -1, 2 l.s -1 and 5 l.s -1 injection rates at Site E xiv

14 Figure 6.11 Simulated groundwater salinity distribution for scenarios A, B and C at time steps 111, 76 and 42 days, respectively (Z magnification = 3) Figure 6.12 Simulated groundwater salinity distribution for scenario D at pre and post injection time steps (Z magnification = 3) 201 Figure 6.13 Simulated groundwater heads at observation well EO7 for scenarios B and D during the injections Figure 6.14 Simulated groundwater salinity distribution for scenario E at pre and post injection time steps (Z magnification = 3).203 Figure 6.15 Simulated groundwater salinity distributions for scenarios C, F and G Figure 6.16 Simulated distribution of solute mass mobilization from the unsaturated zone for the defined scenarios Figure 7.1 Schematic diagram of processes leading to floodplain salinization after the clearance of native vegetation for agricultural practices (adopted from Leblanc et al.,(2012)).215 Figure 7.2 Configuration of production wells (in red) and observation wells (in green) at the Clark s Floodplain. The inset map shows the location of the study site in Australia (red circle) Figure 7.3 a: Configuration of the model boundary condition (model perimeter is shown in red dotted line), b: Configuration of the vegetation and soil layers of Clark s Floodplain along transect B- B' (Z magnification= 3). Observation wells are shown in red columns.222 Figure 7.4 ET (a) and evaporation only (b) during the study period for the defined scenarios Figure 7.5 Cumulative river bank recharge during the study period for the defined scenarios Figure 7.6 GW head dynamics along transect B-B' (BO4, BO5 andbo6) during the study period for the defined scenarios Figure 7.7 Total solute mass in the system at each time step for the defined scenarios Figure 7.8 Cumulative solute mass stored in the system during the study period for the defined scenarios xv

15 Figure 8.1 Structure of the thesis xvi

16 List of Tables Table 3.1 Parameter values of the model for the study site Table 3.2 Model performance evaluation metrics (Means Difference; MSE = Mean Square Error; RMSE = Root Mean Square Error; r2 = Coefficient of Determination; NSE = Nash-Sutcliffe Model Efficiency coefficient)..75 Table 4.1 Selected model parameters Table 4.2 Model performance evaluation metrics Table 4.3 Sensitivity analysis parameters and their values Table 5.1 Porous media and van Genuchten function parameter values.157 Table 5.2 Surface properties values of the numerical model. 159 Table 5.3 ET component parameters values for the study site..160 Table 5.4 Results of the calibrated model performance statistics Table 6.1 Specifications details of the injection and observation wells 188 Table 6.2 Specifications of the defined scenarios Table 6.3 Soil parameter values of the model for the study site Table 7.1 Soil parameter values of the model for the study site Table 7.2 ET parameter values of the model for the study site. 222 xvii

17 Summary The Murray River is one of Australia s longest rivers but in South Australia it has become degraded through river regulation, water extraction and adjacent highland irrigation. These have decreased the natural flood frequency and increased rates of floodplain salinization. Concerns have been raised about the quality of water extracted from the Murray River for industrial, agricultural and potable uses, including for metropolitan Adelaide s water supply. This has been highlighted as the most significant hydrological risk by the Murray Darling Basin Authority (MDBA) and therefore a comprehensive understanding of flow and solute dynamics within the river and floodplain environment is essential. Hence, this research is aimed at developing a better understanding of surface water (SW) and groundwater (GW) flow and solute interactions in a semi-arid river-floodplain system using a numerical modelling approach. Collaborating closely with CSIRO, DEWNR, SA Water and NCGRT, this research initially involved a comprehensive review of the current understanding of numerical modelling of salt mobilization arising from SW-GW interactions. The review concluded that the level of understanding of arid and semi-arid environments is still relatively basic, particularly in relation to SW-GW interactions in floodplains. Following this, Clark s Floodplain in the Lower Murray River was chosen as a study site where sufficient observation facilities were available. The HydroGeoSphere model was selected for this research because it is a 3D physically-based fully integrated surface-subsurface numerical model with variable saturation and solute transport simulation capabilities. A calibrated model was developed and a number of scenarios were designed to investigate the impacts of different drivers on the river-floodplain system processes such as groundwater-table dynamics, evapotranspiration (ET), bank storage, regional groundwater recharge and floodplain salinization. The identified drivers include floodplain vegetation cover, groundwater lowering, river stage manipulation, artificial flooding and water injection to the saline floodplain aquifer. The results show that vegetation cover type can have significant impacts on the flow and solute interaction dynamics due to the influence of ET as a dominant hydrological driver. It was also found that groundwater lowering mitigates the xviii

18 floodplain salinization risk via two mechanisms, namely extraction of the solute mass and creating a divide which stops saline water reaches the floodplain by lowering the groundwater-table. Also, it appears that groundwater extraction is able to remove some of the solute stored in the unsaturated zone. Furthermore, river stage manipulation is beneficial for floodplain health because it amplifies the freshwater lens during high-flow pulses through the mixing of fresh river water with saline groundwater. In addition, it was shown that artificial flooding can temporarily form less saline groundwater and soil profiles that improve water availability for vegetation. However, it was found that this effect is generally limited to the inundated zone. Finally, it was shown that injection of fresh river water to the saline floodplain aquifer may potentially improve soil water availability in the capillary fringe. However, application of this technique seems to be relatively costly and has associated potential problems such as aquifer clogging and well breaching. To conclude, it appears that all of these salt interception measures are limited spatially and temporally. Indeed, none of these measures are able to permanently change the natural condition of the floodplain groundwater salinity and flow regime. Hence the interventions should be considered only as short term management techniques. However, if longer term strategies are required, it may be possible to implement these salt interception measures periodically. The outcomes of this research contribute to a better understanding of how to maintain a healthier floodplain in arid and semi-arid environments using different available management strategies. This research also provides knowledge regarding the ecological implications of SW-GW flow and solute interactions in a semi-arid river-floodplain system, with a particular focus on the Lower Murray River region. xix

19 Declaration This thesis presents work carried out by myself and does not incorporate without acknowledgment any material previously submitted for a degree or diploma in any university; to the best of my knowledge it does not contain any materials previously published or written by another person except where due reference is made in the text; and all substantive contributions by others to the work presented, including jointly authored publications, are clearly acknowledged. Sina Alaghmand March 2014 xx

20 Acknowledgments First and foremost, I would like to express my heartfelt gratitude to my supervisor, Professor Simon Beecham, who gave me the opportunity to develop my research skills and future career pathway. I thank him for his guidance, wisdom, patience, motivation and support during the last four years. I would like to thank Associate Professor Ali Hassanli, who was an associate supervisor for my thesis, and who provided encouraging, positive and constructive feedback and advice. My sincere thanks go to Dr. Juliette Woods (NCGRT), Mr. Ian Jolly and Dr. Kate Holland who provided significant scientific advice during my research. This research was co-funded by the University of South Australia and the Goyder Institute and was supported by NCGRT, CSIRO, DEWNR and SA Water. I would like to thank all of these organizations for their generous continuing support. In particular, I am grateful to Dr. James McCallum, Dr. Dylan Irvine and Dr. Eddie Banks (NCGRT), Dr. Rebecca Doble (CSIRO), Mr. Volmer Berens (DEWNR) and Mr. Peter Forward (SA Water). I also thank the Australian Bureau of Meteorology for providing meteorological data. As a member of SA Water Centre for Water Management and Reuse, I have been surrounded by wonderful colleagues that have provided a rich and rewarding research environment. I thank them all. I would like to particularly thank Dr. Pooria Pasbakhsh, Dr. Hamideh Nouri and Dr. Sattar Chavoshi for their valuable friendship and support during this journey. I would like to thank the UniSA technical officers and administration staff in the Water Centre for Water Management and Reuse and in the School of Natural and Built Environments at the University of South Australia, particularly, Ms. Mary Garnham, Ms. Allison Price and Mr. Neill Sanderson. I would also like to acknowledge all the anonymous journal reviewers for their detailed and helpful comments that have greatly strengthened this thesis. xxi

21 I gratefully acknowledge the funding I received from the Mawson Fellowship that made possible my 4 months exchange research visit at Nagoya University in Nagoya, Japan. In particular, I am grateful for the opportunity to work with leading academics at the Nagoya Hydraulic Research Institute for River Basin Management, especially Prof. Tetsuro Tsujimoto and his research team. Thanks for welcoming me as a friend and for helping me pursue my research there. I would not have completed this journey if not for my wonderful family, my loving parents, Seyedeh Beigom Banikamali and Ali Akbar Alaghmand, for being greatly supportive and interested and always believing in me and for giving me immense love and providing all the possible opportunities to be what I am today. Considerable thanks also to my sisters Farah and Azadeh and their families for their love and encouragement. I never forget their endless moral support while I was studying overseas. xxii

22 1 Introduction 1.1 Floodplain salinization In arid and semi-arid environments, potential evapotranspiration (ET) is typically higher than rainfall. This leads to accumulation of solute mass in the soil profile rather the salt than being leached out (Crosbie et al., 2009). Under natural conditions, occurrence of periodic overbank floods prevents the development of soil and groundwater salinity (Jolly et al., 1996). Overbank flooding is a natural mechanism that flushes the solute mass in the unsaturated zone around the plant root system and this increases soil water availability and riparian vegetation health (Holland et al., 2009a; Jolly et al., 1993; Slavich et al., 1999). In fact, recurrent overbank floods keep the floodplain salinity in an equilibrium condition and, over the long term, there is a salt balance in the soil that enables the development of stable vegetation communities (Jolly et al., 2008). This shows that potential floodplain salinization risk, attributed to excess evapotranspiration, is naturally controlled by frequent overbank flood events. However, the latter driver has been absent in many arid and semi-arid floodplains due to human-induced activities. The natural flow regime has been considerably altered due to abstraction, river regulation and climate change (Jolly et al., 1996). These changes can lead to a decrease in the magnitude and duration of overbank floods and can increase the average interval between floods (Holland et al., 2013). On the other hand, installation of weirs can lead to a shallower water table in the floodplain which increases the groundwater recharge via evapotranspiration (Doble et al., 2006; Jolly et al., 2002). In addition, in some cases, excess recharge from irrigation areas in surrounding highlands can form a groundwater mound, which displaces groundwater towards the floodplain (Berens et al., 2009a; Doble et al., 2006). Therefore, in regions with saline groundwater, solute mass 1

23 accumulates in floodplain soils at an increased rate due to increased groundwater discharge caused by raised water table levels and point source saline discharge from adjacent irrigation areas (Jolly et al., 2008). Thus, in such situations, the solute mass accumulation process (evapotranspiration) has been enhanced and solute mass flushing mechanisms (such as periodic overbank floods) have been altered which leads to increased floodplain salinization. 1.2 Salt interception measures There have been numbers of management responses to mitigate floodplain salinization and improve vegetation health by improving flow regimes and providing water to stressed floodplains through environmental flows (Arthington and Pusey, 2003; Hughes and Rood, 2003). One of the strategies is delivering environmental flows via releases of water from large storages (Richter and Thomas, 2007). Some examples include: experimental dam releases in the Olifants River in South Africa (Brown and King, 2012; Cambray et al., 1997); dam removal programs in the United States (Bednarek, 2001; Hoenke et al., 2014); and riparian vegetation restoration programs in the south-western United States (Stromberg, 2001; Stromberg et al., 2007). While, restoration of flows towards more natural regimes is ideal, these strategy requires sufficient water being available in storages at the time when environmental flows are required which is not plausible in arid and semi-arid regions (Berens et al., 2009b; Holland et al., 2009a). In such regions the scarce available water is expected to be assigned for irrigation, potable water supplies and the generation of hydroelectric power purposes (Tockner and Stanford, 2002). Hence, salt management options are generally restricted in arid and semi-arid regions. However, other strategies have been considered even though they are often limited to particular sites. A notable example is the Lower Murray River in South Australia where various surface water/groundwater manipulation strategies have been investigated to examine their impacts on floodplain salinity and vegetation health. One of the salt interception measures is groundwater lowering via a series of production wells, and in the Lower Murray River this is termed a Salt Interception Scheme (SIS). The aim of this scheme is to mitigate floodplain salinization and to reduce the immediate impact of river salt accession 2

24 (Alaghmand et al., 2013). Other attempts have been made in the Lower Murray River to inject fresh river water to the saline floodplain (Berens et al., 2009a; Berens et al., 2009b). Furthermore, artificial flooding has been implemented to test the vegetation response during a low flow period (Holland et al., 2009a; White et al., 2009). In this context, the concept of surface water-groundwater interaction plays an important role. In fact, comprehensive knowledge on how a river and a saline floodplain interact in an arid and semi-arid environment assists in the understanding of the consequences of such management strategies. 1.3 Numerical modeling Surface-groundwater (SW-GW) interaction is a term used to describe the exchange of water and/or solute between a surface water body and a groundwater aquifer. Characterisation of near-river-aquifer systems is complicated by difficulties in estimating aquifer inflows and outflows, the complex nature of SW- GW interaction processes, and the uncertainty of aquifer properties, which can vary over orders of magnitude (Eslamian and Nekoueineghad, 2009; Sophocleous, 2010). Because of this complexity, analytical (Holland et al., 2009b; Morel- Seytoux and Daly, 1975) and numerical models (Alaghmand et al., 2013; Doble et al., 2006; Kollet and Maxwell, 2006; McCallum et al., 2010) are commonly used to model SW-GW interactions and to estimate the exchange fluxes between surface waters and groundwater. Modelling of the SW-GW interaction requires a detailed understanding of the exchange processes that occur between the surface and subsurface domains (Barnett et al., 2012). In this context, today s physicallybased models are founded on the blueprint for a physically-based mathematical model of a complete hydrological system developed by Freeze and Harlan (1969). Furthermore, when a SW-GW system is hydraulically well connected and dynamic, it should be simulated considering the response of both domains (Gilfedder et al., 2012; Swain and Wexler, 1996). A recent advance in numerical modelling, originally implemented by Brown (1995), is the development of fully integrated models that solve for both the surface and subsurface domains simultaneously using a single matrix of equations. These equations are derived from the principles of conservation of mass and momentum of water and/or solute. It appears that that the fully integrated, physically-based model is an efficient tool for simulating the SW-GW integration and for examining the 3

25 impacts of different drivers on the flow and solute dynamics. However, to do this, extensive observed data is essential to develop and calibrate such models. 1.4 Motivation and aims of the research Salinity of the Murray River, as one of Australia s longest rivers (2520 km), is a significant issue, especially in South Australia. This is due to: the State s location on the lower reaches of the Murray-Darling Basin the natural geological structure of the Murray-Darling Basin in which the Murray River acts as a drain for salt out of the landscape the influence of human development in mobilising salt to the river and its adjacent floodplain the ultimate implications of salinity in terms of water quality for all uses, including the environment and for potable water supply for metropolitan Adelaide Floodplain salinity poses one of the single biggest risks to meeting River Murray salinity management objectives in South Australia (DFW, 2011). On average, about 2 million tonnes of salt per year reaches the sea via the river. River regulation and greatly reduced flows, particularly between 1994 and 2010, resulted in little discharge of salt to the sea for a long period and much solute accumulated in the floodplains. There is significant concern that the salt will be mobilised when these floodplains are inundated as a result of natural overbank flood and/or through planned environmental watering events (DFW, 2011). The Independent Audit Group for Salinity highlighted this as the most significant risk in the Murray-Darling Basin (MDBA, 2010b). Another consequence of the floodplain salinization has been the dieback of environmentally important riparian vegetation, such as red gum (Eucalyptus camaldulensis) and black box (Eucalytpus largiflorens) (Holland et al., 2013; Smith and Kenny, 2005). Some of these floodplains have been recognized as significant ecological assets by the Murray Darling Basin Ministerial Council (MDBC, 2005) and are listed as a Wetland of International Importance under the Ramsar Convention (Holland et al., 2009a). The Murray-Darling Basin Authority s (MDBA) Basin Salinity 4

26 Management Strategy (BSMS) identified a broad need to develop better understanding of the process of floodplain salinization and the development of suitable management actions (MDBA, 2010a). The first objective of the research described in this thesis was to review the current state of understanding of SW-GW interactions with particular respect to the numerical modelling of salt mobilization. The review consisted of four main components, namely the fundamental concept of SW-GW interaction, the dominant processes and drivers involved, salt interception measures and available numerical models. The review showed that groundwater discharge by evapotranspiration is the principal process driving floodplain salinization and vegetation health. Human-induced drivers (e.g. river regulation, excess drainage recharge at the highland adjacent to the floodplains) have stimulated the floodplain salinization rate and disturbed the natural balance. Moreover, different plausible salt interception measures have been summarised. These can be categorized into two groups, namely surface water-based (river stage manipulation and artificial flooding) and groundwater-based measures (groundwater lowering and freshwater injection to the saline aquifer). Furthermore, it was shown that due to the nature of this research area, its complexity, limitations in observation facilities and financial constraints, numerical modelling has become the most popular and relatively efficient tool for researchers and decision makers. However, the shortage of observed quality and quantity data is more obvious in groundwater studies. Hence, any attempt to calibrate and verify numerical models based on observed data would be an important step forward in this research area. The calibrated model can be considered as a tool to gain a better understanding of the impacts of salt interception strategies, particularly where observed data is lacking. The second objective was to build and develop a calibrated numerical model to describe the complex interaction between a river and a saline floodplain in a semiarid environment. This was conducted using a fully integrated, physically-based numerical model featuring variable saturation and solute transport simulation capabilities. Moreover, a study site with sufficient observational facilities and well-documented recorded data was selected to calibrate the model. The calibrated model was used as a tool to pursue the third and fourth objectives. 5

27 The third objective was to investigate and quantify the impacts of four salt interception measures on the floodplain salinity and their implications on vegetation health. The salt interception measures studied in this research include the following: River stage manipulation Artificial flooding Groundwater lowering Injection of fresh river water into a saline floodplain aquifer To evaluate these measures the calibrated model in the second objective was applied. Furthermore, to comprehensively explore various management strategies, a number of scenarios were designed for each salt interception measure. It is well-understood that evapotranspiration is one of the dominant driving processes in arid and semi-arid environments. The fourth objective was to evaluate the impact of floodplain vegetation cover on SW-GW interaction in a semi-arid environment. The aim was to explore whether vegetation cover can significantly influence the dynamics of flow and solute in the context of a river and a semi-arid saline floodplain interaction. This is examined through three scenarios, namely current vegetation (mix of deep rooted and shallow rooted vegetation), coverage by only deep rooted vegetation types such as Eucalyptus trees and coverage by only shallow rooted vegetation types such as grass. 1.5 Structure of the thesis This thesis is divided into eight chapters consisting of six peer-reviewed journal papers together with an introduction and conclusion (See Fig 1.1). 6

28 Numerical modelling of surface water and groundwater flow and solute interactions between a river and a saline floodplain in a semi-arid region Chapter 1: Introduction Chapter 2: Review of SW-GW interactions and salt mobilization Floodplain salinization Paper 1 Surface water manipulation Chapter 3: Modelling the impacts of river stage manipulation on a river-floodplain system Chapter 4: Quantifying the impacts of artificial flooding on a river-floodplain interaction Salt interception measures Paper 2 Paper 3 Impact of SW-based S.I. measures on floodplain salinization Groundwater manipulation Chapter 5: Impacts of groundwater extraction on floodplain salinization risk Chapter 6: Injection of fresh river water into a saline floodplain aquifer Impact of GW-based S.I. measures on floodplain salinization Paper 4 Paper 5 Evapotranspiration Floodplain vegetation Chapter 7: Impacts of vegetation cover on SW-GW flows and solute interactions Impact of ET on floodplain salinization Chapter 8: Conclusion Paper 6 Figure 1.1 Structure of the thesis 7

29 1.5.1 Thesis overview Chapter 1 introduces the concepts of the three main components of this research including floodplain salinization, salt interception measures and numerical models. Chapter 2 (Paper 1) is an extensive literature review on the current state of understanding of SW-GW interactions, the dominant processes and drivers involved, salt interception measures and available numerical models. This work was published as a review paper in the journal Water Resources. According to the 2012 Journal Citation Reports Science Edition (Thomson Reuters, 2013), the Impact Factor for Water Resources is Chapter 3 (Paper 2) presents the numerical modelling of river stage manipulation during a low-flow period as a salt interception measure. A number of scenarios were defined and the extent and magnitude of their impacts were investigated in detail. The solute mass mobilization from the unsaturated zone has also been studied. This work was published in the journal Environmental Modelling and Software (EM & S). According to the 2012 Journal Citation Reports Science Edition (Thomson Reuters, 2013), the Impact Factor for EM & S is Chapter 4 (Paper 3) describes the application of artificial flooding on a saline semi-arid floodplain and explores its potential contributions to floodplain salinity and vegetation health. The paper based on this study is currently in review in the journal Environmental Modelling and Software. Chapter 5 (Paper 4) investigates the relative impacts of groundwater lowering on SW-GW interactions in a semi-arid saline floodplain to study the both the flow and solute dynamics. The outcomes of this study were published in the journal Natural Hazards and Earth System Sciences (NHESS). According to the 2012 Journal Citation Reports Science Edition (Thomson Reuters, 2013), NHESS has an Impact Factor of Chapter 6 (Paper 5) examines whether injection of fresh river water into the floodplain aquifer can displace saline groundwater and provide water for stressed riparian and floodplain plants and trees. The results of the numerical modelling were compared with the reported outcomes of an actual application of this salt 8

30 interception measure at the study site. The paper based on this study is accepted (in-press) in the journal Ecological Engineering. According to the 2012 Journal Citation Reports Science Edition (Thomson Reuters, 2013), Ecological Engineering has an Impact Factor of Chapter 7 (Paper 6) explores the relationship between floodplain vegetation cover and flow and solute dynamics in terms of SW-GW interactions in a semi-arid floodplain. The outcome of this work was published in the journal Environmental Processes. Chapter 8 presents a summary of the research outcomes and the major conclusions drawn from this study as well as recommendations for future research into various salt interception management strategies, particularly in arid and semi-arid environments. 1.6 Contribution to knowledge Human-induced activities including abstraction, river regulation and agricultural practices, particularly on adjacent highlands, have significantly contributed to floodplain salinization and groundwater quality declines and these processes can severely impact on river flow quality and quantity (Danielopol et al., 2003; Jolly et al., 1996; Jolly et al., 1993; Sophocleous, 2002). However, many of these activities are unavoidable due to human population growth and the limited nature of available natural resources. Moreover, climate change has had detrimental impacts on river flow regimes and this has led to long low-flow periods (e.g. from 1994 to 2010 in the Lower Murray River). So, it is essential to have a comprehensive and realistic understanding of the processes involved and of the consequences of potential salt interception management strategies. There has been almost no modelling of SW-GW interactions in arid and semi-arid floodplains with respect to water fluxes, let alone salinity or ecology (Jolly et al., 2008). There is a clear need to develop modelling capabilities for the movement of salt to, from, and within river-floodplain systems. Such models can provide spatial and temporal predictions of salt interception measures which can be used to assess ecosystem outcomes (Jolly et al., 2008). 9

31 This doctoral study has involved a comprehensive review, experimental design, field work, in-situ data acquisition, extensive data collection and analysis, collaboration with authorities, hydrogeological model development, detailed numerical model calibration and a detailed investigation of different salt interception measures and their impacts on floodplain salinization and vegetation health. The original contributions of this research include the following: This research is the first attempt to employ such a comprehensive threedimensional fully integrated, physically-based numerical model on an actual study site. Previously, there has been a limitation on this type of modelling approach due to a lack of sufficient observed data as well as the required computational capabilities. This work has shown that there are linkages between floodplain salinization and vegetation health and that the ever-increasing human-induced pressures on groundwater systems have ecological impacts (Danielopol et al., 2003). However, the majority of the GW SW research related to floodplain salinization and ecology appears to have been in temperate and tropical environments (Jolly et al., 2008). This is one of the first studies to explain the impacts of some of the most efficient and practical salt interception measures on the state of floodplain salinization and vegetation health in arid and semiarid environments. There are numerous uncertainties and technical issues associated with any salt management strategy and these need to be quantified and addressed before such measures can be implemented. The outcomes of this research will assist in predicting the behaviour of river-floodplain systems in response to salt intercept measures. 1.7 List of publications From this research, six journal papers have been produced. Alaghmand, S., Beecham, S. and Hassanli, A. (2013), A Review of the Numerical Modelling of Salt Mobilization from Groundwater-Surface Water Interactions, Water Resources, Springer, 40(3), pp

32 Alaghmand, S., Beecham, S., Jolly, I. D., Holland, K. L., Woods, J. A. and Hassanli, A. (2014), Modelling the Impacts of River Stage Manipulation on a Complex River-Floodplain System in a Semi-Arid Region, Environmental Modelling and Software, Elsevier, 59, pp Alaghmand, S., Beecham, S., Woods, J.A., Jolly, I.D., Holland K.L. and Hassanli, A., Quantifying the impacts of artificial flooding and groundwater lowering on a river-floodplain interaction in a semi-arid saline floodplain, Environmental Modelling and Software, Elsevier (in review). Alaghmand, S., Beecham, S. and Hassanli, A. (2013), Impacts of Groundwater Extraction on Salinization Risk in a Semi-Arid Floodplain, Natural Hazards and Earth System Sciences, European Geosciences Union, 13(12), pp Alaghmand, S., Beecham, S., Woods, J.A., Jolly, I.D., Holland K.L., and Hassanli, A., Injection of fresh river water into a saline floodplain aquifer as a salt interception measure in a semi-arid environment, Ecological Engineering, Elsevier (Accepted for publication). Alaghmand, S., Beecham, S. and Hassanli, A. (2014), Impacts of Vegetation Cover on Surface-Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia, Environmental Processes, Springer, 1, pp

33 References Alaghmand, S., Beecham, S., Hassanli, A., Impacts of Groundwater Extraction on Salinization Risk in a Semi-Arid Floodplain. Nat. Hazards Earth Syst. Sci. 13(12) Arthington, A.H., Pusey, B.J., Flow restoration and protection in Australian rivers. River Research and Applications 19(5-6) Barnett, B., Townley, L.R., Post, V., Evans, R.E., Hunt, R.J., Peeters, L., Richardson, S., Werner, A.D., Knapton, A., Boronkay, A., Australian groundwater modelling guidelines, In: report, W. (Ed.). National Water Commission: Canberra. Bednarek, A.T., Undamming rivers: A review of the ecological impacts of dam removal. Environmental Management 27(6) Berens, V., White, M., Souter, N., 2009a. Bookpurnong Living Murray Pilot Project: A trial of three floodplain water management techniques to improve vegetation condition. Department of Water, Land and Biodiversity Conservation: Adelaide. Berens, V., White, M.G., Souter, N.J., 2009b. Injection of fresh river water into a saline floodplain aquifer in an attempt to improve the condition of river red gum (Eucalyptus camaldulensis Dehnh.). Hydrological Processes 23(24) Brown, C., King, J., Modifying dam operating rules to deliver environmental flows: Experiences from southern Africa. International Journal of River Basin Management 10(1) Brown, D.L., An analysis of transient flow in upland watersheds: interactions between structure and process. University of California: Berkeley, p Cambray, J.A., King, J.M., Bruwer, C., Spawning behaviour and early development of the Clanwilliam yellowfish (Barbus capensis; Cyprinidae), linked to experimental dam releases in the Olifants River, South Africa. Regulated Rivers: Research and Management 13(6) Crosbie, R.S., McEwan, K.L., Jolly, I.D., Holland, K.L., Lamontagne, S., Moe, K.G., Simmons, C.T., Salinization risk in semi-arid floodplain wetlands subjected to engineered wetting and drying cycles. Hydrological Processes Danielopol, D.L., Griebler, C., Gunatilaka, A., Notenboom, J., Present state and future prospects for groundwater ecosystems. Environmental Conservation 30(2) DFW, South Australia s Report to the Basin Salinity Management Strategy. Department for Water, Government of South Australia: Adelaide. 12

34 Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2) Eslamian, S., Nekoueineghad, B., A review on interaction of groundwater and surface water. International Journal of Water 5(2) Freeze, R.A., Harlan, R.L., Blueprint for a physically-based, digitallysimulated hydrologic response model. Journal of Hydrology 9(3) Gilfedder, M., Rassam, D.W., Stenson, M.P., Jolly, I.D., Walker, G.R., Littleboy, M., Incorporating land-use changes and surface groundwater interactions in a simple catchment water yield model. Environmental Modelling & Software 38(0) Hoenke, K.M., Kumar, M., Batt, L., A GIS based approach for prioritizing dams for potential removal. Ecological Engineering Holland, K.L., Charles, A.H., Jolly, I.D., Overton, I.C., Gehrig, S., Simmons, C.T., 2009a. Effectiveness of artificial watering of a semi-arid saline wetland for managing riparian vegetation health. Hydrological Processes Holland, K.L., Jolly, I.D., Overton, I.C., Walker, G.R., 2009b. Analytical model of salinity risk from groundwater discharge in semi-arid, lowland floodplains. Hydrological Processes Holland, K.L., Turnadge, C.J., Nicol, J.M., Gehrig, S.L., Strawbridge, A.D., Floodplain response and recovery: comparison between natural and artificial floods, Technical Report Series No. 13/4. Goyder Institute for Water Research: Adelaide. Hughes, F.M.R., Rood, S.B., Allocation of River Flows for Restoration of Floodplain Forest Ecosystems: A Review of Approaches and Their Applicability in Europe. Environmental Management 32(1) Jolly, I.D., McEwan, K.L., Cox, J., Walker, G.R., Holland, K.L., Managing Groundwater and Surface Water for Native Terrestrial Vegetation Health in Saline Areas, CSIRO Land and Water Technical Report 23/02. CSIRO Land and Water: Canberra, Australia. Jolly, I.D., McEwan, K.L., Holland, K.L., A review of groundwater-surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1(1) Jolly, I.D., Walker, G.R., Hollingworth, I.D., Eldridge, S.R., Thorburn, P.J., McEwan, K.L., Hatton, T.J., The causes of decline in eucalypt communities and possible ameliorative approaches, In: walker, G.R., Jolly, I.D., Jarwal, S.D. (Eds.), Salt and Water Movement in the Chowilla Floodplain. CSIRO Division of Water Resources: Canberra, Australia. Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4)

35 Kollet, S.J., Maxwell, R.M., Integrated surface-groundwater flow modeling: A free-surface overland flow boundary condition in a parallel groundwater flow model. Advances in Water Resources 29(7) McCallum, J.L., Cook, P.G., Brunner, P., Berhane, D., Solute dynamics during bank storage flows and implications for chemical base flow separation. Water Resources Research 46(7). MDBA, 2010a. Basin Salinity Management Strategy Annual Implementation Report. MDBA, 2010b. Report of the Independent Audit Group for Salinity Murray-Darling Basin Authority (MDBA): Canberra, Australia. MDBC, The Living Murray Foundation Report on the significant ecological assets targeted in the first step decision, Publication Number 09/05. Murray Darling Basin Commission: Canberra. Morel-Seytoux, H.J., Daly, C.J., A discrete kernel generator for stream aquifer studies. Water Resources Research Richter, B.D., Thomas, G.A., Restoring environmental flows by modifying dam operations. Ecology and Society 12(1). Slavich, P.G., Walker, G.R., Jolly, I.D., Hatton, T.J., Dawes, W.R., Dynamics of Eucalyptus largiflorens growth and water use in response to modified watertable and flooding regimes on a saline floodplain. Agricultural Water Management 39(2-3) Smith, F.M., Kenny, S.K., Floristic vegetation and tree health mapping, River Murray Floodplain, South Australia. Department for Environment and Heritage: Adelaide. Sophocleous, M., Interactions between groundwater and surface water: The state of the science. Hydrogeology Journal 10(1) Sophocleous, M., Review: Groundwater management practices, challenges, and innovations in the High Plains aquifer, USA-lessons and recommended actions. Revue critique: Pratiques, défis et innovations dans le domaine des de la gestion des eaux souterraines de l'aquifère des Grandes Plaines (High Plains), aux Etats Unis d'amérique - Leçons et recommandations 18(3) Stromberg, J.C., Restoration of riparian vegetation in the south-western United States: Importance of flow regimes and fluvial dynamism. Journal of Arid Environments 49(1) Stromberg, J.C., Beauchamp, V.B., Dixon, M.D., Lite, S.J., Paradzick, C., Importance of low-flow and high-flow characteristics to restoration of riparian vegetation along rivers in arid south-western United States. Freshwater Biology 52(4) Swain, E.D., Wexler, E.J., A coupled surface-water and ground-water flow model (MODBRANCH) for simulation of stream-aquifer Interaction, U.S. 14

36 Geological Survey Techniques of Water-Resources Investigations Report, book 6, chapter A6. U.S. Geological Survey: Washington, p Tockner, K., Stanford, J.A., Riverine flood plains: Present state and future trends. Environmental Conservation 29(3) White, M.G., Berens, V., Souter, N.J., Bookpurnong Living Murray Pilot Project: Artificial inundation of Eucalyptus camaldulensis on a floodplain to improve vegetation condition. Science, Monitoring and Information Division, Department of Water, Land and Biodiversity Conservation. 15

37 2 Review of SW-GW interactions and salt mobilization Overview Chapter 2 addresses the first objective of this research, namely to review the current state of understanding of SW-GW interactions with particular respect to the numerical modelling of salt mobilization. This chapter describes the concept of SW-GW interactions and lists the dominant drivers involved in this process. It shows how the natural long-term SW- GW balance has been interrupted by human-induced activities that have led to floodplain salinization and decline of vegetation health. It also provides some examples of land salinization around the world. Furthermore, it reviews various numerical modelling approaches that have been used to simulate the SW-GW interactions. This shows that there is a vital need to develop modelling capabilities which are capable of handling all dominant features of the SW-GW interactions, including solute exchange between the surface and subsurface domains. Such calibrated, physically-based and fully integrated models can be useful tools for providing temporal predictions of floodplain salinity. These in turn can be used to assess potential management strategies. This is particularly important in arid and semi-arid environments as fewer studies have been conducted in these areas due to the general scarcity of sufficient recorded data and observational facilities. Finally, this chapter describes some examples of different salt interception measures that have been reported in the literature. This chapter was published as a paper in the journal Water Resources in 2013 (Paper 1). The manuscript was co-authored by my Principal Supervisor, Prof. 16

38 Simon Beecham, and advisor, Dr. Ali Hassanli. The format of the paper has been changed to be consistent with the rest of this thesis. 17

39 Paper 1: A Review of the Numerical Modelling of Salt Mobilization from Groundwater-Surface Water Interactions Published as: Alaghmand, S., Beecham, S. and Hassanli, A. (2013), A Review of the Numerical Modelling of Salt Mobilization from Groundwater-Surface Water Interactions, Water Resources, Springer, 40(3), pp Abstract: Salinization of land and water is a significant challenge in most continents and particularly in arid and semi-arid regions. The need to accurately forecast surface and groundwater interactions has promoted the use of physicallybased numerical modelling approaches in many studies. In this regard, two issues can be considered as the main research challenges. First, in contrast with surface water, there is generally less observed level and salinity data available for groundwater systems. These data are critical in the validation and verification of numerical models. The second challenge is to develop an integrated surfacegroundwater numerical model that is capable of salt mobilization modelling but which can be validated and verified against accurate observed data. This paper reviews the current state of understanding of groundwater and surface water interactions with particular respect to the numerical modelling of salt mobilization. Three-dimensional physically-based fully coupled surfacesubsurface numerical model with the capability of modelling density-dependent, saturated-unsaturated solute transport is an ideal tool for groundwater-surface water interaction studies. It is concluded that there is a clear need to develop modelling capabilities for the movement of salt to, from, and within floodplains to provide temporal predictions of floodplain salinity which can be used to assess ecosystem outcomes. 2.1 Introduction Salinization of land and water is a significant challenge in most continents and particularly in arid and semi-arid regions. In such regions, salinization is a risk because potential evapotranspiration (ET) is so much greater than rainfall, and this allows salts to accumulate rather than being leached out (Crosbie et al., 2009). By 18

40 the early 1980s, dryland salinity, or saline seeps, had affected approximately 0.8 million ha of crops on the Great Plains of North America (Miller and Gray, 2002), including Alberta (Greenlee et al., 1968), Saskatchewan (Steppuhn, 2001; Steppuhn and Wall, 1999), Manitoba, Montana (Halvorson and Black, 1974) and North Dakota (Worcester et al., 1975). An increase in salt concentrations in groundwater on floodplains in the United States has also been reported by Konikow and Person (1985) and by Goff et al. (1998) for the Arkanese River Valley, Colorado. Shallow saline groundwater systems are evident in the lowlying topography of the Netherlands (de Vries, 1995; Kloosterman et al., 1995), Spain (Schmid et al., 2006) and Tunisia and Hungary (Schofield et al., 2001). The salinization of the Aral Sea in Russia due to irrigation expansion is a classic and well documented case of irrigation and salinization (Saeijs and van Berkel, 1995; Saiko and Zonn, 2000). In Zambia, the Kafue Flats are degraded due to the construction of dams upstream and downstream of the flats; the downstream dam has permanently inundated some wetlands, whereas the upstream dam restricts flood flows to these same wetlands (Mumba and Thompson, 2005). The Okavango Delta in Botswana is under threat from developments upstream diverting water away from the wetlands (Bauer et al., 2006a). The floodplains of the lower River Murray in Australia suffer from similar problems to those highlighted above. The Murray Darling Basin (Ebert, 1971; Herczeg et al., 2001; Jolly et al., 2001) and the south-western wheat-belt of Western Australia (George, 1991; Peck, 1978) are sites for extensive research into the processes and remediation of dryland salinization. Approximately one third of the world s land area is arid and semi-arid (Jolly et al., 1998; Rogers, 1981). Salinization of rivers and floodplain is one of the most significant consequences of development of these areas for agricultural production (Allison et al., 1990; Jolly et al., 1998; Orlob and Ghorbanzadeh, 1981; Person and Konikow, 1986) (Figure 2.1). Semi-arid floodplains have a natural tendency to accumulate salts and require periodic flushing to prevent the development of soil and groundwater salinity (Jolly et al., 1996). In many parts of the world, the previous high variability in surface water flows has been reduced through river regulation (Maheshwari, 1995). River regulation impairs the natural flushing cycle of floodplains by decreasing the frequency and duration of small and 19

41 medium size floods (MDBC, 2001). In addition, river regulation can increase the natural rate of floodplain salinization, e.g. through increased rates of water table evaporation in the vicinity of weirs and dams (Jolly et al., 1996). Figure 2.1 Schematic diagram of processes leading to dryland salinity after the clearance of native vegetation for agricultural practices (Leblanc et al., 2012) When a river has an adjacent floodplain, surface and groundwater flow interactions can be highly complex due to the influence of overbank floods (Khublaryan and Zyryanov, 2006). This complexity means that field studies alone are often insufficient to unravel the nature of the underlying processes operating in a given situation (Jolly et al., 1998). Moreover, they are of only limited use in determining the probable impacts of changed river and/or floodplain management. For these reasons models, mathematical (Khublaryan and Zyryanov, 2006), analytical (Hall and Moench, 1972b; Morel-Seytoux and Daly, 1975) and numerical (Diersch, 1996; Harbaugh et al., 2000; Marino, 1975, 1981; Therrien et al., 2005), have been developed for the study of surface and groundwater flow 20

42 interaction. While many of the models account for the time varying stream boundary conditions, they are often limited in that they only consider the flow of water into and out of bank storage and do not model the effects of overbank flow which occur during floods. Furthermore, they are often concerned only with the transport of water between the stream and the aquifer and not with the transport of any solute. Notable exceptions to the latter can be found in the literature (Holland et al., 2009a; Konikow and Bredehoeft, 1974; Konikow and Person, 1985; Person and Konikow, 1986). In recent decades this research area has attracted a significant number of investigations because of the importance of water and salinity management. Due to the nature of this research area, its complexity, limitations in observation facilities and financial constraints, numerical modelling has become the most popular and relatively efficient tool for researchers and decision makers. Among the investigations conducted to simulate the interactions of groundwater and surface water, most have considered water only and have ignored salt mobilization in the system. However, salt mobilization plays an essential role in this process and indeed in many catchments, groundwater-surface water interaction modelling is not complete without consideration of salt mobilization. This paper reviews the current state of understanding of groundwater and surface water interaction with particular respect to the numerical modelling of salt mobilization. 2.2 Surface water-groundwater interactions Interactions between surface and subsurface flow domains are strongly governed by the relative surface water and groundwater heads and these can vary over the short term (Rosenberry and Winter, 1997). Over the long term, changes in interactions may occur when there are changes in the heads determined by causes such as climate change (Wurster et al., 2003), management of upland areas (Allison et al., 1990; Doble et al., 2006; Grebenyukov, 2001) and changes in the flow regimes of a river due to regulation, channelization and upstream water abstractions (Jolly et al., 1996; Kovalevskii, 2003; Overton et al., 2006; Walker and Thoms, 1993). Some arid-zone floodplains experience significant recharge during flood events (Harrington et al., 2002; Marti et al., 2000), but others do not 21

43 (Akeroyd et al., 1998; Jolly et al., 1998). In fact, the presence of soil layers with a low infiltration rate is a common feature of many floodplain environments limiting vertical recharge during flood events (Burt et al., 2002; Jolly et al., 1994). On the other hand, the presence of low conductivity soil layers could facilitate the lateral transmission of flood pulses across the floodplain by generating confining conditions in the alluvial aquifer (Jolly et al., 1998). As one of the first attempts to examine the interaction between floodwater and aquifer flow during flooding, Jolly et al. (1994) analysed stream, soil and groundwater observed data before, during and after a flood event in They used the theory developed by Pinder et al. (1969) to estimate the diffusivity of the Monoman Formation as the main geological formation in the Chowilla floodplain, South Australia. Based on this study it was revealed that diffuse vertical recharge of floodwater does not occur to any significant extent. The floodplain is an important source of saline groundwater which is added to the river following floods. Jolly et al. (1994) proposed a hypothesis for Lower River Murray floodplains (Figure 2.2) that where the Coonambidal Clay is thin or absent are zones of localized recharge which was supported by a further study by Jolly et al. (1998). Woessner (2000) concluded that groundwater and stream channel interactions in fluvial plains can be classified as gaining, losing, parallel-flow and flow through. Also, exchange of surface water and groundwater may occur at the channel-bed scale as shallow water circulation into the underlying sediments facilitates the groundwater recharge and discharge. Figure 2.2 Conceptual model of groundwater inputs to the floodplain and potential groundwater discharge pathways within the floodplain in the Lower River Murray (Holland et al., 2009b) 22

44 Evolution of nitrogen dynamics during low flow periods were studied using a 1D hydro-biogeochemical model developed by Sauvage et al. (2003). The model estimated fluxes between free-flowing water and epilithic biofilms during low flow periods. However, the model could not be used to identify the areas of interaction between river and aquifer nor could it integrate the complex two-way coupling between aquifer and river during higher water periods. Furthermore, Weng et al. (2003) applied the MARTHE hydrodynamic model developed by the Bureau de Recherches Géologiques et Minières, BRGM, France to estimate the buffer function of the floodplain on the Garonne River. The model showed satisfactory estimations for replicating water depths in groundwater, both for high and low water flows in the river. However, the model required complex inputs such as topographical and hydraulic field measurements in the calibration process, and it did not simulate the direct two-way coupling between surface and subsurface water as well as did the Sauvage et al. (2003) model. Schilling and Zhang (2006b) confirmed that the contribution of nitrate from incised channel riparian zones to stream nitrate loads is considerable. They hypothesised that this will probably be greatest during the recession limb of the hydrograph as nearstream groundwater with highest nitrate concentrations discharges into the channel. To identify the temporal and spatial scales of surface and groundwater exchange in the floodplain, Lamontagne (2005) used a piezometric surface network and environmental tracers in Hattah-Kulkyne in Victoria. Lamontagne showed that the River Murray in the study area was a losing river except in short periods following flood events, although excess salt stores in floodplains could be gradually removed with carefully managed programmes of successively larger floods (Lamontagne et al., 2005). Doble et al. (2006) modelled the spatial distribution of net groundwater flux by developing a relationship to use within a finite difference groundwater flow model (MODFLOW) to develop a conceptual understanding of the hydrological processes between regional groundwater inflows and vegetation water use in the Clarks floodplain in the lower River Murray, Australia (Figure 2.3). They demonstrated that localized elevation was the most critical parameter in determining the portion of inflow to a floodplain expressed as seepage. The long-term pattern of net groundwater discharge was 23

45 found to be dependent on vegetation distribution, elevation, soil type and river geometry. (a) (b) Figure 2.3 Geological cross section (a) and cross-section of the conceptual model (b) of Clark s Floodplain at Lower River Murray (Doble et al., 2006) Models that are capable of simulating surface and groundwater interactions are commonly based on the conductance concept. These types of models link the subsurface and surface domains via an exchange flux. Because experimental evidence of such a distinct interface is often lacking in field systems, there was a need for a more general coupled modelling approach. Such a model was presented by Kollet et al. (2006) and this incorporated a new two-dimensional overland flow simulator into the parallel three-dimensional variably saturated subsurface flow code, ParFlow. This new overland flow simulator took the form of an upper boundary condition and was, thus, fully integrated without relying on the conductance concept. Another important advantage of this approach was the 24

46 efficient parallelism incorporated into ParFlow, which was exploited by the overland flow simulator. Furthermore, in order to assess the effects of different strategies for water resources management of the floodplain and wetlands an investigation has been carried out in the lowland landscape of the lower Havel River in Germany (Krause et al., 2007a). Krause et al. (2007a) utilized a coupling approach known as the IWAN model which incorporated WASIM-ETH for simulation of runoff generation and water balance and the MODFLOW model for groundwater dynamics and interaction with surface water (Figure 2.4). A comparison of lateral and vertical water balance components showed the dominance of lateral flow processes and the importance of the interaction between surface water and groundwater for the overall water balance. In another study, Krause et al. (2007b) used the same coupling model to investigate the water balance and groundwater dynamics of the lower Havel floodplain. They discussed how the dominant mechanism for the water balance and groundwater dynamics of the floodplain is temporally and spatially variable exchanges between the surface water and groundwater through the floodplain. To study water and solute exchange between a large river and its aquifer, a vertically (z-direction) averaged 2D model known as 2SWEM (Surface- Subsurface Water Exchange Model) was used by Peyrard et al. (2008). Horizontal 2D Saint Venant equations (depth averaging the Navier-Stokes equations) for river flow and a 2D Dupuit equation for aquifer flow were coupled. The Fully implicit approach (Fairbanks et al., 2001) also known as a Simultaneous solution (Gunduz and Aral, 2003; van der Kwaak and Loague, 2001) was utilized in a dynamic coupling procedure between two domains using continuity of fluxes and water level elevations. After establishment of the model by comparing the simulated results with observed hydraulic heads in a 36 km 2 section of the Garonne River (south-west France), the solute transport component was considered based on classical advection-dispersion equations (Bear, 1972). Peyrard et al. (2008) demonstrated the ability of the model to handle solute exchanges between the floodplain aquifer and the river. They modelled the dynamics of a tracer injected into the river and followed its route as infiltration through the riverbed to capture the extent of the hyporheic zone. Also, the model was able to simulate the dynamics of a solute dispersed in the floodplain (e.g. 25

47 pesticides) and to predict the subsequent concentrations along the river. Nevertheless, this study was limited to a site where a river fully penetrated a shallow aquifer and the model was not applied in more complex cases. Therefore the results can only be considered as averaged quantifications for non-complex 2D investigations. Figure 2.4 The coupling approach and interfaces of the IWAN model components incorporating WASIM-ETH for the simulation of the runoff generation and water balance of the unsaturated soil zone and MODLFOW for the groundwater dynamics and interactions with the surface water (Krause et al., 2007b) Berens et al. (2009) reported the injection of fresh water into a saline floodplain aquifer to create a fresh water lens in the Clarks floodplain, South Australia. The potential for this technique appears limited in this hydrogeological environment because of the semi-confined nature of the River Murray floodplain aquifers. They demonstrated that river fresh water injection would not have replaced surface water flooding as surface water inundation is required for long-term persistence. Filling a wetland with fresh water over several months successfully improved the health of the fringing Eucalyptus community (Holland et al., 2009a). The extent of improvement in tree health was proportional to the extent of bank recharge at different locations around the wetland. 26

48 In order to examine the hypothesis that the surface water connection between the water bodies and the river can limit salinization, Crosbie et al. (2009) used a combination of hydrodynamic, natural tracer and geophysical methods. They showed that while the water bodies (wetlands) in floodplains were groundwater recharge features under flooding conditions, they became groundwater discharge features following a manipulated drying cycle, thereby increasing salinization risk. In fact, it was surface water that prevented the wetlands from being salinized. Without connection to the river they lost all their mechanisms for salt export and became groundwater discharge features. Another study in the Clarks floodplain in the River Murray showed that surface and groundwater interaction in this regulated semi-arid wetland increases groundwater salinity as it intercepts groundwater discharge, concentrates it by evaporation and recharges more saline water to the aquifer (Banks et al., 2009). Holland et al. (2009b) developed an assessment tool to evaluate the potential salinity risk to floodplain vegetation and rivers. A simple, one-dimensional, crosssectional analytical model of a semi-arid lowland (along a 85 km reach of the lower River Murray) was applied and compared to the MODFLOW numerical model. Both models predicted groundwater discharge as seepage at the break of slope, evapotranspiration through the floodplain and base-flow to the river. In addition, it was shown that salinization associated with groundwater discharge by evapotranspiration is the principle process driving floodplain vegetation health. Lenahan and Bristow (2010) investigated the historical groundwater solute dynamics of the lower Burdekin floodplain located in North Queensland, Australia to determine the influence of agricultural irrigation on groundwater quality. Long-term observations in the lower Burdekin floodplain showed a steady decline in groundwater quality which could be attributed to changes in the hydrologic cycle that resulted in the mobilization of sub-surface solute stores, especially in arid and semi-arid areas(beven, 2002b; Ebel and Loague, 2006; Loague et al., 2006; Loague and van der Kwaak, 2004). This included displacement of unsaturated zone solutes and mobilization of saturated zone solutes (Lenahan and Bristow, 2010). 27

49 2.3 Physically-based numerical models The need to accurately forecast surface and groundwater interactions has promoted the use of physically-based numerical modelling approaches in many studies (Beven, 2001, 2002a, b, 2006; Beven and Binley, 1992; Ebel and Loague, 2006; Loague et al., 2006; Loague and van der Kwaak, 2004). Distributed physically-based simulation models of coupled surface and subsurface processes in catchments have only been developed relatively recently and many are based on the early work of Freeze and Harlan (1969) [e.g. InHM (van der Kwaak, 1999); HydroGeosphere (Therrien et al., 2005); PIHM (Qu and Duffy, 2007b); ParFlow (Kollet and Maxwell, 2006); MIKE SHE (Abbott et al., 1986), CATHY (Camporese et al., 2010), MODHMS (Hydrogeologic Inc., 2003), GSFLOW (Markstrom et al., 2008), MMS (Leavesley et al., 1996), FHM (Ross et al., 1997) and SWATMOD (Sophocleous and Perkins, 2000)]. But, this approach was considered in the last two decades with availability of powerful desktop computers. The majority of the recent models to date with this approach relate components of the surface and subsurface regimes applying externallycoupled/time-lagged schemes or iteratively-coupled approaches with respect to the compatibility of land surface interface fluxes or pressure heads (Sudicky et al., 2008). The fully integrated approach, originally implemented by Brown (1995) is relatively new; however, a number of physically-based, distributed models are now incorporating the numerical solution technique including the models developed at the University of Waterloo such as the Integrated Hydrology Model (van der Kwaak, 1999) and the HydroGeoSphere (Therrien et al., 2005). Regardless of the introduction of these numerical models, most of the applications have been limited to experimental plots or small sub-catchments (Jones et al., 2008; Loague et al., 2005; Loague and van der kwaak, 2002; van der Kwaak and Loague, 2001). There are some studies in the literature that describe large scale coupling models. For example, Jones et al. (2006) and Sulis et al. (2011) used InHM and CATHY in 75 km 2 and 690 km 2 catchments in Canada. HydroGeoSphere (Therrien et al., 2005) has been widely used in different aspects of groundwater-surface water interaction studies. Cey et al. (2006), Brookfield et al. (2009), Raymond et al. (2011) and Bonton et al. (2011) used HydroGeoSphere to model lateral mass transfer, heat transfer and temperature evolution. Similarly, 28

50 HydroGeoSphere was employed to investigate the effects of viscosity, capillarity and grid spacing on thermal variable-density flow (Graf and Boufadel, 2011). HydroGeosphere was also utilized by Li et al. (2008) and Calderhead et al. (2011) to test the capability of the fully integrated surface-subsurface numerical model to simulate in 3D two catchments in Canada and Mexico, that were 286 km 2 and 2100 km 2 in area, respectively. They showed that a physically-based modelling approach that treats the surface and subsurface flow systems as an integrated continuum can capture the dynamic response of a large heterogeneous catchment. Bunn et al. (2011) validated HydroGeoSphere simulations of the dynamic nature of the vertical gradients that form within the zone of tension saturation above the water table during pumping and recovery. Weatherill et al. (2008), Graf et al. (2008) and Rosenbom et al. (2009) demonstrated the capability of HydroGeoSphere to model water and solute transport in 3D variably-saturated fractured environments. The impacts of climate change on groundwater reserves were studied using HydroGeoSphere with advanced climate change scenarios for the Geer Basin (465 km 2 ) in Belgium (Goderniaux et al., 2009). Furthermore, different types of transition behaviour from connected to disconnected surface water bodies and temporal and spatial factors were investigated using HydroGeoSphere (Brunner et al., 2009). The model was also applied to test the hydraulic mixing-cell (HMC) method to quantify the groundwater component of surface flow along the river (Partington et al., 2011). Additionally, the solute dynamics during bank storage flows for chemical base flow separation during and after single and multiple floods was studied using HydroGeoSphere (McCallum et al., 2010). 2.4 Coupling of surface-subsurface domains One of the principal current focuses in groundwater-surface water interaction studies is the method of coupling between river and aquifer modelling equations (Kollet and Maxwell, 2006; Said et al., 2005). Indeed, despite the fact that aquifer and surface water are hydraulically interconnected, they are often modelled as two separate systems and are analyzed independently (Schmid et al., 2006). This is likely to be particularly unsuitable for catchment scale modelling (Halford and Mayer, 2000) because groundwater movement has a much larger timescale than that of free surface water movement or because of the difficulties in measuring 29

51 and modelling their interactions (Liang et al., 2007). In recent years, the threat of flooding has received greater attention as a result of the impact of global warming and sea level rise. In prolonged wet weather over a large area, rising water tables can have a significant impact on flooding (ICE, 2001). Furthermore, it has been found that the water motion in the near-surface soil layer at depths of 0 to 5m can have a similar timescale to that of the surface water flow (Liang et al., 2007). Nevertheless, the need to consider surface water and groundwater as a single system has become increasingly important, in terms of both quantity and quality, especially near wetlands, lakes, rivers and coastal regions (Winter, 1998). It has been shown that fully coupled models for river and aquifer flows are necessary to obtain a better understanding of the hydrological pathways in hydrosystems (Gunduz and Aral, 2005; Panday and Huyakorn, 2004). Fairbanks et al. (2001) demonstrated that the fully implicit approach (Gunduz and Aral, 2003; van der Kwaak and Loague, 2001), in which both systems of equations are solved in a single global matrix, is the most numerically stable method to couple surface and subsurface models. In fact, solving the surface, subsurface, and interface fluxes simultaneously is both theoretically and aesthetically more satisfying due to the elimination of the artificial boundary condition between the surface-subsurface interface that exists when applying externally- or iteratively-coupled approaches (Sudicky et al., 2008). Historically, analytical solutions for describing groundwater-surface flow interaction, such as those proposed by Hall and Moench (1972a), Glover (1974) and Gill (1985) have ignored riverbed leakage (Perkins and Koussis, 1996). In fact, Cooper and Rorabaugh (1963) and Pinder and Sauer (1971) were the first researchers to consider the importance of groundwater-surface water coupling (in their case in the form of a flood wave travelling down a long river reach). Furthermore, a coupled model was considered in the rainfall-runoff process simulations by Freeze (1972) and Morita and Yen (2002) who recognised the importance of two domains of water exchange in the early stage of rainfall events. Many of the coupled groundwater-surface water numerical simulations represent interactions between surface water bodies and groundwater using the popular SUTRA (Bobba et al., 2000; Gardner and Wilson, 2006; Jolly et al., 1998; McKenzie et al., 2007; Simmons and Narayan, 1998; Weatherill et al., 2004) or 30

52 MODFLOW (Harbaugh et al., 2000; McDonald and Harbaugh, 1988; Niswonger and Prudic, 2005; Prudic, 1989; Prudic et al., 2004) models to represent the groundwater flow component. Coupling of short-term transient open channel and groundwater flows has been accomplished within the MODBRANCH (Swain and Wexler, 1996; Wilcox et al., 2007) and DAFLOW-MODFLOW models (Jobson and Harbaugh, 1999). The surface and subsurface domains are coupled at the common interface, where infiltration takes place. The simple models that have been applied by Singh and Bhallamudi (1996) and Fiedler and Ramirez (2000) approximate the subsurface contribution by using an algebraic infiltration formula, without recourse to any differential equation. These approaches are limited due to the spatial variability of the free surface head that may not be represented properly and errors are expected when the interaction is complex (Freeze, 1972). To overcome this limitation, the subsurface flow should also be simulated using partial differential equations. Coupled numerical models can be categorised based on the spatial dimensions of the two domains. Most of the models have used one-dimensional surface flow with one-, two- or three-dimensional subsurface flow (Morita and Yen, 2002; Sparks, 2005). The one-dimensional surface flow refers to open channel flow in the longitudinal direction, while one-dimensional subsurface flow refers to the saturated/unsaturated flow in the vertical direction (Liang et al., 2007). Gunduz and Aral (2005) tested a solution by coupling a one-dimensional open channel flow model based on the dynamic wave form of the Saint Venant equations with a two-dimensional vertically averaged saturated groundwater flow model. The proposed method was based on the idea of solving a single global matrix rather than solving separate matrices for each flow domain while improving the solution iteratively. They concluded that this approach provides an efficient solution for the coupled flow problem formulated for both systems. Liang et al. (2007) used a similar approach to build a two-dimensional numerical model for predicting flood flows. They used the TVD-MacCormack scheme to solve the shallow water equations and the standard MacCormack scheme to solve the transient Boussinesq equations for unconfined groundwater flows. The dynamic linking of the surfacesubsurface models captured the interactions between the surface water flow and the neighbouring groundwater flow in the horizontal plane (often the two domains 31

53 are coupled vertically). Liang et al. (2007) showed that the coupled surfacesubsurface model gives a high degree of flexibility in representing building structures in a flood flow simulation. The solution proposed by VanderKwaak and Loague (2001) and Gunduz and Aral (2005) can be particularly efficient for studying two-way interactions between rivers and aquifers, but examples of such applications are still rare (Langevin et al., 2005). Many coupled models of variable complexity have been developed to simulate the interactions between surface and subsurface systems (Langevin et al., 2005; Morita and Yen, 2002; Panday and Huyakorn, 2004; van der Kwaak and Loague, 2001; Weng et al., 2003). For most of the models mentioned above, the spatial extent of the interactions between groundwater and surface water is assumed to be constant, or spatially and temporally transient changes in processes are considered in a rudimentary manner (Krause and Bronstert, 2007). In addition, the full three-dimensional (3D) numerical solution encounters limitations in mesoscale applications because of the high computation time required (Krause and Bronstert, 2007). Gunduz and Aral (2003) also concluded that the coupled models tend to be increasingly complex and their solution can suffer from numerical complications. It appears that the complexity of the model should thus depend on the objectives of each investigation. 2.5 Density-dependent flows Understanding the complex interaction between groundwater and surface water is essential for the effective management of water resources (Sophocleous and Perkins, 2000). Most of the studies on groundwater-surface water interactions have focused on hydraulic (Kollet and Maxwell, 2006; Meire et al., 2010; Partington et al., 2011; Schubert, 2002; Winter, 1999) or biogeochemical aspects (Baskaran et al., 2009; Dousson, 1997; Huang et al., 2011; Massoudieh et al., 2010; Sauvage et al., 2003; Weng et al., 2003) of the groundwater-surface water interaction. The influence of density differences between a fresh surface water body and saline groundwater also plays an important role. Density-driven flow processes are important in a number of natural systems (Holzbecher, 1998; Simmons, 2005), where the density varies as a function of suspended solid content, temperature and pressure of a fluid (Massmann et al., 32

54 2006; Weatherill et al., 2004). Hydrogeological investigations that have emphasised the importance of density-driven flow include upcoming below wells (Diersch, 1984; Reilly, 1987), seawater intrusion in coastal regions (Cheng, 2001; Huyakorn, 1987; Khublaryan et al., 2008; Volker, 1982) and waste disposal in deep salt formations (Kolditz, 1998; Oldenburg, 1995) or saline disposal basins and salt lakes (Simmons, 1997, 2002; Simmons and Narayan, 1998; Wooding, 1997) and dense contaminant plume migration (Liu, 1996; Oostrom, 1992; Schincariol, 1997). In hydraulically connected groundwater-surface water systems, the flow is assumed to be controlled by the same mechanisms as those that occur in interaquifer leakage (Rushton, 1979). In other words, the flow is considered to be a direct function of the head difference between the surface water body (river, lake or wetland) and aquifer and of the hydraulic conductivity of the semi-permeable river sediments (Massmann et al., 2006). Sophocleus (2002) mentioned that this simple approach of considering the flow between a river and aquifer is often too simplistic and that more appropriate models exist. Whilst the simplest approach might be to convert measured heads into equivalent fresh water heads, this is problematic particularly where there is vertical density stratification and vertical flows are of interest. Indeed, Lusczynski (1961) recognized that equivalent fresh water heads cannot be used to determine the vertical hydraulic gradient in an aquifer with water of non-uniform density. Thus, the use of equivalent fresh water head analyses together with a standard density-invariant version of Darcy s Law is expected to lead to errors in estimation and prediction of surface watergroundwater interaction processes and the subsequent estimates of salt discharge to the surface water body. Surface and groundwater flow interaction in arid and semi-arid areas often involves a large contrast in salinity between the fresh stream/floodwater and groundwater (Jolly et al., 1998). The standard approach to modelling salt transport to streams is to utilise a groundwater flow model to predict the water fluxes to the stream and to multiply these by the observed or inferred salt concentrations in the aquifer (Ghassemi et al., 1989). However, studies reported by Herbert et al. (1988) and Oostrom et al. (1992) suggest that large salinity contrasts can result in density-dependent flow. SUTRA, a variable numerical density flow and solute 33

55 transport model (Voss, 1984), has been used for numerical simulation of densitydependent flows (Bobba et al., 2000; Gardner and Wilson, 2006; McKenzie et al., 2007; Misut and Voss, 2004; Simmons and Narayan, 1998). Jolly et al. (1998) utilized SUTRA by consideration of time varying boundary conditions of surface water and groundwater (stream-aquifer) interactions in the Chowilla floodplain in the lower River Murray. The aims were to better understand the processes operating in the transport of saline groundwater to floodplains and to determine the impact of various flooding scenarios on the movement of salt to the floodplain. Furthermore, Charlesworth et al. (1994) simulated flooding and a proposed salt interception scheme at Chowilla using a two dimensional SUTRA model. The three common benchmark tests for variable-density codes that are applied are the Henry problem of seawater intrusion comparing numerical results with the known semi-analytical solutions (Oswald and Kinzelbach, 2004; Segol, 1994); the Hydrocoin problem (Level 1, Case 5) of a simplified salt dome(the International HYDROCOIN Project, 1991); and the Elder problem of fingering in a non-stable situation (Elder, 1967). These three benchmark tests are all 2D, and they are either not sensitive (Henry problem), or the solution is not exactly known (Hydrocoin and Elder problem) so that only inter-comparison between different codes is possible. In this regard, Oswald and Kinzelbach (2004) suggested a set of measurements to represent a 3D benchmark test case, which can be considered to examine the reliability of time-dependent variable-density flow and transport numerical models. Weatherill et al. (2004) examined SUTRA on the three variations of the density-dependent Horton-Rogers-Lapwood problem (Horton and Rogers Jr, 1945; Lapwood, 1948). They termed these three test cases as infinite horizontal box, finite horizontal box and infinite inclined box. They suggested the possibility of extension of these 2D benchmark tests to 3D situations. Most modelling studies of flow density effects in fractured media have been limited to orthogonal fracture networks, consisting only of vertical and horizontal fractures (Shikaze et al., 1998). Graf and Therrien (2005) modelled 3D variabledensity flow and transport in fractured porous media using FRAC3DVS (a saturated-unsaturated numerical groundwater flow and solute transport model) 34

56 (Therrien et al., 2005; Therrien and Sudicky, 1996). Moreover, Graf and Degener (2011) gave guidance on appropriate spatial-temporal grids for a variable-density flow problem that included vertical fracture zones. Massmann et al. (2006) studied the role of variable density flow on the discharge behaviour of saline groundwater into fresh surface water bodies. They pointed out that the density gradients could significantly modify the behaviour of the interaction between the saline groundwater and the fresher surface water body. They also concluded that it was plausible for density driven flow to suppress discharge rates, while density gradients may act in the opposite direction to advection in the case where the surface water body is a discharging feature. However, they emphasized that as the approach was purely theoretical, it would be valuable to consider subsequent laboratory experiments and 2D and 3D numerical modelling. 2.6 Floodplain salinity and vegetation health Bornman et al. (2004) showed that the distribution and health of vegetation in floodplain depends on the depth to the water table and the salinity of the groundwater. In fact, anthropogenic changes to flooding regimes in highly variable arid catchments have a critical effect on floodplain vegetation (Alexander and Dunton, 2006; Capon, 2005; Mensforth, 1997). Flood events are necessary to periodically flush the accumulated salts that build up around the rootzone because of evapotranspiration (Jolly et al., 1993; Slavich et al., 1999). Flood events also recharge the floodplain aquifer (Jolly et al., 1993). Short-lived periods of fresh water inundation at specific times of the year alleviate stress in saline and dry soils and promote seed germination (Alexander and Dunton, 2002; Callaway et al., 1990; Forbes and Dunton, 2006; Noe and Zedler, 2001; Zedler and West, 2008). But, catastrophic flooding on the other hand can lead to sedimentation and increased soil salinity (Zedler et al., 1986; Zedler and West, 2008). Soil salinization in floodplains was modelled by Overton et al. (2006) using a spatial and temporal model, WINDS-GIS in the Lower River Murray floodplain, Australia. They modelled three management scenarios including raising a weir, lowering groundwater and enhancement of flow regimes. It was shown that a combination of lowering groundwater and enhancement of flow regimes was more effective at reducing the soil salinization rate and for flushing salt from the 35

57 root zone. Surface water pulses (in the form of natural or artificial flooding) are a well-recognized control on ecological functions in arid and semi-arid areas. Fresh water pulses can be the primary means by which salt stored in both the water column and the underlying sediments are flushed from wetlands (Jolly et al., 2008). Holland et al. (2009a) conceptualized and described the response of aquifers and vegetation in the Chowilla floodplain in the River Murray to artificial watering in March They attempted to show that vegetation response was controlled by the floodplain s hydraulic conductivity. Moreover, artificial watering was shown to be effective and that regular artificial watering was required during low-flow periods. Bornman and Adams (2010) showed that the high salinity concentrations of the soil and groundwater needed to be lowered for natural rehabilitation by introducing fresh water to the floodplain. The limited impact of natural floods and the long gap between high peak floods indicated that the natural flushing of salts was not enough. In this case, the only feasible measure was to replace the saline groundwater with fresh/less saline water by pumping out the groundwater. This formed a hydraulic gradient resulting in the groundwater being recharged with less saline water. Jolly et al. (2008) emphasized the need for modelling of groundwater-surface water interactions in arid and semi-arid wetlands with respect to water fluxes, let alone salinity or ecology. In addition, they identified a clear need to extend the numerical modelling of salt movement to, from, and within wetlands to provide temporal predictions of wetland salinity. Plausible floodplain salt management measures such as artificial flooding, saline groundwater extraction and fresh water injection to floodplain aquifers also need to be modelled. In this regard, the availability of observed water quality and quantity data is essential to calibrate and validate the numerical models so that they can be used as effective prediction tools. 2.7 Effects of river/floodplain geometry Some studies have focused on the role of the water body s geometry (wetland, aquifer and etc.) on groundwater-surface water interactions (Frei et al., 2010; Nield et al., 1994; Smith and Townley, 2002; Townley and Davidson, 1988; Townley and Trefry, 2000). The effects of micro-topography on infiltration and runoff generation processes were first investigated by Dunne et al. (1991). They pointed out that hill slope runoff was controlled by an intricate interplay between 36

58 rainfall intensity, surface flow depth, vegetation coverage and the specific microtopography of the slope. Depressions in floodplains can attenuate and delay surface flows (Kværner and Kløve, 2008), because these first need to be filled until a specific threshold is exceeded and surface flow towards the stream channel (as back-water) can be initiated (Antoine et al., 2009; Fiedler and Ramirez, 2000). Several investigations have highlighted the effects of floodplain depression on runoff dynamics. Dunne et al. (1991) used a conceptual approach to simulate overland flow and infiltration processes for uniform sinusoidal micro-topography. They demonstrated the significance of floodplain depressions in the process. Esteves et al. (2000) and Fiedler and Ramirez (2000) showed the effects of depressions on flow directions, flow velocities and flow depths using finite difference solutions to the two-dimensional depth-averaged dynamic wave equations. Each of the abovementioned modelling studies were limited to surface water and infiltration and did not consider the interaction between surface and subsurface flows, which is an important process in floodplains (Devito and Hill, 1997; Gibson et al., 2000). However, Qu and Duffy (2007a) applied a finite element coupled surface-subsurface flow model to demonstrate how small-scale topography can control local surface saturation. However, the spatial resolution was too coarse to account for floodplain depression on the sub-metre scale. Frei et al. (2010) evaluated the complex hydrologic dynamics of a virtual river floodplain with depression (micro-topography) using the HydroGeoSphere numerical model. They concluded that surface water storage was a better predictor for the occurrence of surface runoff than groundwater levels. Channelization of rivers in order to provide flood protection or hydropower installation is one of the other issues that has influenced flow regimes and the interactions of rivers and their adjacent groundwater aquifers by decreasing the longitudinal connectivity (Peter, 2010). Hence, river restoration, and particularly enlarging the river bed, has been considered as one of the most efficient methods to enhance river ecology. Successful river restoration requires careful consideration of the impact on groundwater in the river corridor. These operations can result in gradual changes of the groundwater composition (Hoehn and Scholtis, 2011). Understanding changes in groundwater quality from bank filtration, which increases water exchange by unclogging the river bed, is critical 37

59 for efficient restoration (Lautz and Fanelli, 2008; Zobrist, 2010). In this regard, Hoehn and Scholtis (2011) analysed water samples from the River Thur floodplain in North Switzerland. They demonstrated the risk of deteriorating groundwater quality due to increases in bank filtration water and decreases in subsurface residence times after river bed enlargement. 2.8 Conclusions The following conclusions can be drawn from this review: 1. Modelling surface and subsurface flows in two separate domains does not generally provide an accurate representation of their interaction. Considering the availability today of powerful processors and more efficient solutions (including fully implicit), modelling of surface and subsurface flows should be coupled. 2. Modifications to the river or floodplain geometry, soil properties or flow management may affect the groundwater-surface water interactions and must be included in any modelling approaches. 3. To represent realistic groundwater-surface water interactions, densitydependent flow processes must be taken into account, especially where differences in salinity are significant. 4. The interaction of groundwater and surface water has been studied in detail. Currently, more attention is required to extend the numerical modelling from 1D or 2D to 3D to better understand and represent this natural phenomenon. 5. The effects of floodplain salt management measures such as artificial flooding, saline groundwater extraction and fresh water injection to floodplain aquifers need to be included in any modelling approaches. 6. A 3D fully integrated, physically-based fully coupled surfacesubsurface numerical model with the capability of modelling densitydependent, saturated-unsaturated solute transport is an ideal tool for groundwater-surface water interaction studies. 38

60 7. There is a clear need to develop modelling capabilities for the movement of salt to, from, and within floodplain to provide temporal predictions of floodplain salinity which can be used to assess ecosystem outcomes. The detailed interactions of a river and its floodplain aquifer are required in order to develop a reasonable understanding of salt accumulation and mobilization in the floodplain. To this end, a validated numerical model is an important tool but one which principally relies on accurate observation data. The shortage of observed quality and quantity data is more obvious in groundwater studies. Hence, any attempt to collect field data and verify numerical models based on observed data would be an important step forward in this research area. 39

61 References Abbott, M.B., Bathurst, J.C., Cunge, J.A., O Connell, P.E., Rasmussen, J., An introduction to the European Hydrological System Syst`eme Hydrologique Europ een, SHE, 2: Structure of a physically-based distributed modeling system. Journal of Hydrology 87(1-2) Akeroyd, M.D., Tyerman, S.D., Walker, G.R., Jolly, I.D., Impact of flooding on the water use of semi-arid riparian eucalypts. Journal of Hydrology 206(1-2) Alexander, H.D., Dunton, K.H., Freshwater inundation effects on emergent vegetation of a hypersaline salt marsh. Estuaries 25(6 B) Alexander, H.D., Dunton, K.H., Treated wastewater effluent as an alternative freshwater source in a hypersaline salt marsh: Impacts on salinity, inorganic nitrogen, and emergent vegetation. Journal of Coastal Research 22(2) Allison, G.B., Cook, P.G., Barnett, S.R., Walker, G.R., Jolly, I.D., Hughes, M.W., Land clearance and river salinisation in the western Murray Basin, Australia. Journal of Hydrology 119(1-4) Antoine, M., Javaux, M., Bielders, C., What indicators can capture runoffrelevant connectivity properties of the micro-topography at the plot scale? Advances in Water Resources 32(8) Banks, E.W., Simmons, C.T., Jolly, I.D., Doble, R.C., McEwan, K.L., Herczeg, A.L., Interactions between a saline lagoon and a semi-confined aquifer on a salinized floodplain of the lower River Murray, southeastern Australia. Hydrological Processes Baskaran, S., Ransley, T., Brodie, R.S., Baker, P., Investigating groundwater-river interactions using environmental tracers. Australian Journal of Earth Sciences 56(1) Bauer, P., Gumbrich, t.t., Kinzelbach, W., 2006a. A regional coupled surface water/groundwater model of the Okavango Delta, Botswana. Water Resources Research 42 W Bauer, P., Held, R., Zimmermann, S., Linn, F., Kinzelbach, W., 2006b. Coupled flow and salinity transport modelling in semi-arid environments: the Shashe River Valley, Botswana. Journal of Hydrology Bear, J., Dynamics of Fluids in Porous Media. Elsevier, New York. Berens, V., White, M.G., Souter, N.J., Injection of fresh river water into a saline floodplain aquifer in an attempt to improve the condition of river red gum (Eucalyptus camaldulensis Dehnh.). Hydrological Processes Beven, K., On explanatory depth and predictive power. Hydrological Processes 15(15)

62 Beven, K., 2002a. Towards a coherent philosophy for modelling the environment. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 458(2026) Beven, K., 2002b. Towards an alternative blueprint for a physically based digitally simulated hydrologic response modelling system. Hydrological Processes 16(2) Beven, K., A manifesto for the equifinality thesis. Journal of Hydrology 320(1-2) Beven, K., Binley, A., The future of distributed models: model calibration and uncertainty prediction. Hydrological Processes 6(3) Bobba, A.G., Singh, V.P., Bengtsson, L., Application of environmental models to different hydrological systems. Ecological Modelling 125(1) Bonton, A., Rouleau, A., Bouchard, C., Rodriguez, M.J., Nitrate transport modeling to evaluate source water protection scenarios for a municipal well in an agricultural area. Agricultural Systems 104(5) Bornman, T.G., Adams, J.B., Response of a hypersaline salt marsh to a large flood and rainfall event along the west coast of southern Africa. Estuarine, Coastal and Shelf Science 87(3) Bornman, T.G., Adams, J.B., Bate, G.C., The influence of floodplain geohydrology on the distribution of Sarcocornia pillansii in the Olifants Estuary on the West Coast, South Africa. Journal of Arid Environments 56(4) Brookfield, A.E., Sudicky, E.A., Park, Y.J., Conant Jr, B., Thermal transport modelling in a fully integrated surface/subsurface framework. Hydrological Processes 23(15) Brown, D.L., An analysis of transient flow in upland watersheds: interactions between structure and process. University of California: Berkeley, p Brunner, P., Simmons, C.T., Cook, P.G., Spatial and temporal aspects of the transition from connection to disconnection between rivers, lakes and groundwater. Journal of Hydrology 376(1-2) Bunn, M.I., Rudolph, D.L., Endres, A.L., Jones, J.P., Field observation of the response to pumping and recovery in the water table region of an unconfined aquifer. Journal of Hydrology 403(3-4) Burt, T., Pinay, G., Matheson, F., Haycock, N., Butturini, A., Clement, J., Danielescu, S., Dowrick, D., Hefting, M., Hillbricht-Ilkowska, A., Maitre, V., Water table fluctuations in the riparian zone: comparative results from a pan-european experiment. Journal of Hydrology Calderhead, A.I., Therrien, R., Rivera, A., Martel, R., Garfias, J., Simulating pumping-induced regional land subsidence with the use of InSAR and field data in the Toluca Valley, Mexico. Advances in Water Resources 34(1)

63 Callaway, R.M., Jones, S., Ferren Jr, W.R., Parikh, A., Ecology of a mediterranean-climate estuarine wetland at Carpinteria, California; plant distributions and soil salinity in the upper marsh. Canadian Journal of Botany 68(5) Camporese, M., Paniconi, C., Putti, M., Orlandini, S., Surface-subsurface flow modeling with path-based runoff routing, boundary condition-based coupling, and assimilation of multisource observation data. Water Resources Research 46(2). Capon, S.J., Flood variability and spatial variation in plant community composition and structure on a large arid floodplain. Journal of Arid Environments 60(2) Cey, E., Rudolph, D., Therrien, R., Simulation of groundwater recharge dynamics in partially saturated fractured soils incorporating spatially variable fracture apertures. Water Resources Research 42(9). Charlesworth, A.T., Narayan, K.A., Simmons, C.T., Modelling salt accession within the Chowilla Ananbranch and possible mitigation schemes, Divisional Report 94/7. CSIRO, Division of Water Resources. Cheng, J.M., Chen, C.X., Three-dimensional modeling of density dependent salt water intrusion in multilayered coastal aquifers in the Jahe River Basin, Shandong Province, China. Ground Water 39(1) Cooper, H.H.J., Rorabaugh, M.I., Ground-water movements and bank storage due to flood stages in surface streams, Water Supply Paper 1536-J. US Geological survey. Crosbie, R.S., McEwan, K.L., Jolly, I.D., Holland, K.L., Lamontagne, S., Moe, K.G., Simmons, C.T., Salinization risk in semi-arid floodplain wetlands subjected to engineered wetting and drying cycles. Hydrological Processes de Vries, J.J., Seasonal expansion and contraction of stream networks in shallow saline groundwater systems. Journal of Hydrology Devito, K.J., Hill, A.R., Sulphate dynamics in relation to groundwatersurface water interactions in headwater wetlands of the southern Canadian Shield. Hydrological Processes 11(5) Diersch, H.J.G., Interactive, graphics-based finite-element simulation system FEFLOW for modeling groundwater flow, contaminant mass and heat transport processes, In: Research, I.f.W.R.P.a.S. (Ed.): The Netherlands. Diersch, H.J.G., Prochnow, D., Thiele, M., Finite-element analysis of dispersion-affected saltwater upcoming below a pumping well. Applied Mathematical Models Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2)

64 Dousson, C., Poitevin, G., Ledoux, E., Detay, M., River bank filtration: modelling of the changes in water chemistry with emphasis on nitrogen species. Journal of Contaminant Hydrology Dunne, T., Zhang, W., Aubry, B.F., Effects of rainfall, vegetation, and microtopography on infiltration and runoff. Water Resources Research 27(9) Ebel, B.A., Loague, K., Physics-based hydrologic-response simulation: Seeing through the fog of equifinality. Hydrological Processes 20(13) Ebert, C.H.V., Irrigation and salt problems in Renmark, South Australia. Geographical Review Elder, J.W., Transient convection in a porous medium. J. Fluid Mech. 27(3) Esteves, M., Faucher, X., Galle, S., Vauclin, M., Overland flow and infiltration modelling for small plots during unsteady rain: Numerical results versus observed values. Journal of Hydrology 228(3-4) Fairbanks, J., Panday, S., Huyakorn, P., Comparisons of linked and fully coupled approaches to simulating conjunctive surface/subsurface flow and their interactions, In: Seo, B., Poeter, E., Zheng, C. (Eds.), MODFLOW 2001 and Other Modeling: Odysseys, Greece, pp Fiedler, F.R., Ramirez, J.A., A numerical method for simulating discontinuous shallow flow over an infiltrating surface. International Journal for Numerical Methods in Fluids 32(2) Fiedler, F.R., Ramirez, J.A., A numerical method for simulating discontinuous shallow flow over an infiltrating surface. International Journal for Numerical Methods in Fluids Forbes, M.G., Dunton, K.H., Response of a subtropical estuarine marsh to local climatic change in the southwestern gulf of Mexico. Estuaries and Coasts 29(6 B) Freeze, R.A., Role of subsurface flow I generating surface runoff: 1. Base flow contributions to channel flow. Water Resources Research 8(3) Freeze, R.A., Harlan, R.L., Blueprint for a physically-based, digitallysimulated hydrologic response model. Journal of Hydrology 9(3) Frei, S., Lischeid, G., Fleckenstein, J.H., Effects of micro-topography on surface-subsurface exchange and runoff generation in a virtual riparian wetland - A modeling study. Advances in Water Resources 33(11) Gardner, L.R., Wilson, A.M., Comparison of four numerical models for simulating seepage from salt marsh sediments. Estuarine, Coastal and Shelf Science 69(3-4) George, R.J., Management of saline seeps in the wheatbelt of Western Australia. Agricultural Water Management

65 Ghassemi, F., Jakeman, A.J., Thomas, G.A., Ground-water modelling for salinity management: An Australian case study. Ground Water Gibson, J.J., Price, J.S., Aravena, R., Fitzgerald, D.F., Maloney, D., Runoff generation in a hypermaritime bog-forest upland. Hydrological Processes 14(15) Gill, M.A., Bank storage characteristics of a finite aquifer due to sudden rise and fall of river level. Journal of Hydrology 76(1-2) Glover, R.E., Transient Groundwater Hydraulics. Goderniaux, P., Brouyère, S., Fowler, H.J., Blenkinsop, S., Therrien, R., Orban, P., Dassargues, A., Large scale surface-subsurface hydrological model to assess climate change impacts on groundwater reserves. Journal of Hydrology 373(1-2) Goff, K., Lewis, M.E., Person, M.A., Konikow, L.F., Simulated effects of irrigation on salinity in the Arkanese River valley in Colorado. Ground Water 36(1) Graf, T., Boufadel, M.C., Effect of viscosity, capillarity and grid spacing on thermal variable-density flow. Journal of Hydrology 400(1-2) Graf, T., Degener, L., Grid convergence of variable-density flow simulations in discretely-fractured porous media. Advances in Water Resources 34(6) Graf, T., Therrien, R., Variable-density groundwater flow and solute transport in porous media containing nonuniform discrete fractures. Advances in Water Resources 28(12) Graf, T., Therrien, R., A test case for the simulation of three-dimensional variable-density flow and solute transport in discretely-fractured porous media. Advances in Water Resources 31(10) Grebenyukov, P.G., Interaction between Surface and Subsurface Waters: Case Study of a Region in Kazakhstan. Water Resources 28(1) Greenlee, G.M., Pauluk, S., Bowser, W.E., Occurrences of soil salinity in drylands of southwestern Alberta. Canadian Journal of Soil Science Gunduz, O., Aral, M., Simultaneous solution of coupled surface water/groundwater flow systems, In: Brebbia, C.A. (Ed.), International Conference on River Basin Management: Gran Canaria Islands, pp Gunduz, O., Aral, M., River networks and groundwater flow: simultaneous solution of a coupled System. Journal of Hydrology 301( ). Halford, K.J., Mayer, G.C., Problems associated with estimating ground water discharge and recharge from stream-discharge records. Ground Water 38(3)

66 Hall, F.R., Moench, A.F., 1972a. Application of the convolution equation to stream-aquifer relationships. Water Resour. Res. 8(2) Hall, F.R., Moench, A.F., 1972b. Application of the convolution equation to stream aquifer relationships. Water Resources Research Halvorson, A.D., Black, A.L., Saline seep development in dryland soils of north-eastern Montana. Journal of Soil and Water Conservation 29(2) Harbaugh, A.W., Banta, E.R., Hill, M.C., McDonald., M.G., MODFLOW- 2000, the U.S. Geological Survey modular ground-water model, User guide to modularization concepts and the ground-water flow process-usgs Open File Report USGS: Reston, Virginia. Harrington, G., Cook, P., Herczeg, A., Spatial and temporal variability of ground water recharge in central Australia: a tracer approach. Ground Water Herbert, A.W., Jackson, C.P., Lever, D.A., Coupled groundwater flow and solute transport with fluid density strongly dependent upon concentration. Water Resources Research Herczeg, A.L., Dogramaci, S.S., Leaney, F.W., Origin of dissolved salts in a large, semi-arid groundwater system: Murray Basin, Australia. Marine and Freshwater Research Hoehn, E., Scholtis, A., Exchange between a river and groundwater, assessed with hydrochemical data. Hydrology and Earth System Sciences 15(3) Holland, K.L., Charles, A.H., Jolly, I.D., Overton, I.C., Gehrig, S., Simmons, C.T., 2009a. Effectiveness of artificial watering of a semi-arid saline wetland for managing riparian vegetation health. Hydrological Processes Holland, K.L., Jolly, I.D., Overton, I.C., Walker, G.R., 2009b. Analytical model of salinity risk from groundwater discharge in semi-arid, lowland floodplains. Hydrological Processes Holzbecher, E., Modeling Density-driven Flow in Porous Media. Springer, Berlin. Horton, C.W., Rogers Jr, F.T., Convection currents in a porous medium. Journal of Applied Physics 16(6) Huang, J.C., Mitsch, W.J., Johnson, D.L., Estimating biogeochemical and biotic interactions between a stream channel and a created riparian wetland: A medium-scale physical model. Ecological Engineering 37(7) Huyakorn, P.S., Anderson, P.F., Mercer, J.W., White, J.H.O., Saltwater intrusion in aquifers: development and testing of a three-dimensional finite element model. Water Resources Research Hydrogeologic Inc., MODHMS software (Version 2.0) documentation. Hydrogeologic Inc. 45

67 ICE, Learning to live with rivers. Institution of Civil Engineers (ICE), London, UK. Jobson, H.E., Harbaugh, A.W., Modifications to the diffusion analogy surface-water flow model (DAFLOW) for coupling to the modular finitedifference ground-water flow model (MODFLOW), USGS Open File Report USGS: Reston, Virginia. Jolly, I.D., McEwan, K.L., Holland, K.L., A review of groundwater surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1(1) Jolly, I.D., Narayan, K.A., Armstrong, D., Walker, G.R., The impact of flooding on modelling salt transport processes to streams. Environmental Modelling & Software 13(1) Jolly, I.D., Walker, G.R., Hollingworth, I.D., Eldridge, S.R., Thorburn, P.J., McEwan, K.L., Hatton, T.J., The causes of decline in eucalypt communities and possible ameliorative approaches, In: walker, G.R., Jolly, I.D., Jarwal, S.D. (Eds.), Salt and Water Movement in the Chowilla Floodplain. CSIRO Division of Water Resources: Canberra, Australia. Jolly, I.D., Walker, G.R., Narayan, K.A., Floodwater recharge processes in the Chowilla anabranch system, South Australia. Australian Journal of Soil Research Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4) Jolly, I.D., Williamson, D.R., Gilfedder, M., Walker, G.R., Morton, R., Robinson, G., Jones, H., Zhang, L., Dowling, T.I., Dyce, P., Nathan, R.J., Nandakumar, N., Clarke, R., McNeill, V., Historical stream salinity trends and catchment salt balances in the Murray-Darling Basin, Australia. Marine and Freshwater Research Jones, J.P., Sudicky, E.A., Brookfield, A.E., Park, Y.J., An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow. Water Resources Research 42(2). Jones, J.P., Sudicky, E.A., McLaren, R.G., Application of a fully-integrated surface-subsurface flow model at the watershed-scale: A case study. Water Resources Research 44(3). Khublaryan, M., Frolov, A., Yushmanov, I., Seawater intrusion into coastal aquifers. Water Resources 35(3) Khublaryan, M., Zyryanov, V., Modeling the interaction between water flows. Water Resources 33(5) Kloosterman, F.H., Stuurman, R.J., van der Meijden, R., Groundwater flow systems analysis on a regional and nation-wide scale in the Netherlands; the use of flow systems analysis in wetland management. Water Resources Research

68 Kolditz, O., Ratke, R., Diersch, H.-J.G., Zielke, W., Coupled groundwater flow and transport: 1 Verification of variable-density flow and transport models. Advances in Water Resources Kollet, S.J., Maxwell, R.M., Integrated surface-groundwater flow modeling: A free-surface overland flow boundary condition in a parallel groundwater flow model. Advances in Water Resources 29(7) Konikow, L.F., Bredehoeft, J.D., Modelling flow and chemical changes in an irrigated stream aquifer system. Water Resources Research Konikow, L.F., Person, M., Assessment of long-term salinity changes in an irrigated stream aquifer system. Water Resources Research Kovalevskii, V.S., Principles of Substantiating the Functional Reliability of Systems Based on the Combined Use of Surface and Subsurface Water Resources. Water Resources 30(6) Krause, S., Bronstert, A., Water balance simulations and groundwatersurface water interactions in a mesoscale lowland river catchment. Hydrological Processes Krause, S., Bronstert, A., Zehe, E., 2007a. Groundwater-surface water interactions in a North German lowland floodplain - Implications for the river discharge dynamics and riparian water balance. Journal of Hydrology 347(3-4) Krause, S., Jacobs, J., Bronstert, A., 2007b. Modelling the impacts of land-use and drainage density on the water balance of a lowland-floodplain landscape in northeast Germany. Ecological Modelling 200(3-4) Kværner, J., Kløve, B., Generation and regulation of summer runoff in a boreal flat fen. Journal of Hydrology 360(1-4) Lamontagne, S., Leaney, F.W., Herczeg, A.L., Groundwater surface water interactions in a large semi-arid floodplain: implications for salinity management. Hydrological Processes Langevin, C., Swain, E.D., Wolfert, M., Simulation of integrated surface water-groundwater flow and salinity for a coastal wetland and adjacent estuary. Journal of Hydrology Lapwood, E.R., Convection of a fluid in a porous medium. Proc. Camb. Phil. Soc Lautz, L.K., Fanelli, R.M., Seasonal biogeochemical hotspots in the streambed around restoration structures. Biogeochemistry 91(1) Leavesley, G.H., Restrepo, P.J., Markstrom, S.L., Dixon, M., Stannard, L.G., The modular modeling system (MMS): User s manual, USGS Open-File Report USGS: Reston, Virginia. Leblanc, M., Tweed, S., Van Dijk, A., Timbal, B., A review of historic and future hydrological changes in the Murray-Darling Basin. Global and Planetary Change

69 Lenahan, M.J., Bristow, K.L., Understanding sub-surface solute distributions and salinization mechanisms in a tropical coastal floodplain groundwater system. Journal of Hydrology 390(3-4) Li, Q., Unger, A.J.A., Sudicky, E.A., Kassenaar, D., Wexler, E.J., Shikaze, S., Simulating the multi-seasonal response of a large-scale watershed with a 3D physically-based hydrologic model. Journal of Hydrology 357(3-4) Liang, D., Falconer, R., Lin, B., Coupling surface and subsurface flow in a depth averaged flood wave model. Journal of Hydrology Liu, H.H., Dane, J.H., A criterion for gravitational instability in miscible dense plumes. Journal of Contaminant Hydrology 23(3) Loague, K., Heppner, C., Abrams, R., Carr, A., VanderKwaak, J., Ebel, B., Further testing of the integrated hydrology model (InHM): event-based simulations for a small rangeland catchment located near Chickasha, Oklahoma. Hydrological Processes 19(7) Loague, K., Heppner, C.S., Mirus, B.B., Ebel, B.A., Ran, Q., Carr, A.E., BeVille, S.H., VanderKwaak, J.E., Physics-based hydrologic-response simulation: Foundation for hydroecology and hydrogeomorphology. Hydrological Processes 20(5) Loague, K., van der kwaak, J.E., Simulating hydrological response for the R-5 catchment: Comparison of two models and the impact of the roads. Hydrological Processes 16(5) Loague, K., van der Kwaak, J.E., Physics-based hydrologic response simulation: platinum bridge, 1958 Edsel, or useful tool? Hydrol. Proc. 16(5) Lusczynski, N., J. Geophys. Res. 66 (12), , Head and flow of ground water of variable density. Journal of Geophysical Research 66(12) Maheshwari, B., Walker, KF., McMahon, TA., Effects of regulation on the flow regime of the river Murray, Australia. Regulated Rivers: Research and Management 10(1) Marino, M.A., Digital simulation model of aquifer response to stream stage fluctuation. Journal of Hydrology 25(1-2) Marino, M.A., Analysis of the transient movement of water and solutes in stream-aquifer systems. Journal of Hydrology 49(1-2) 1-7. Markstrom, S.L., Niswonger, R.G., Regan, R.S., Prudic, D.E., Barlow, P.M., GSFLOW Coupled groundwater and surface-water FLOW model based on the integration of the precipitation-runoff modeling system (PRMS) and the modular ground-water flow model (MODFLOW-2005), USGS Techniques and Methods, 6-D1. USGS: Reston, Virginia. 48

70 Marti, E., Fisher, S., Schade, J., Grimm, N., Flood frequency and stream riparian linkages in arid lands, In: Jones, J., Mulholland, P. (Eds.), Streams and Groundwater. Academic Press: San Diego, USA, pp Massmann, G., Simmons, C.T., Love, A., Ward, J., Smith, A., On variable density surface water-groundwater interaction: A theoretical analysis of mixed convection in a stably-stratified fresh surface water-saline groundwater discharge zone. Journal of Hydrology Massoudieh, A., Bombardelli, F.A., Ginn, T.R., A biogeochemical model of contaminant fate and transport in river waters and sediments. Journal of Contaminant Hydrology 112(1-4) McCallum, J.L., Cook, P.G., Brunner, P., Berhane, D., Solute dynamics during bank storage flows and implications for chemical base flow separation. Water Resources Research 46(7). McDonald, M.G., Harbaugh, A.W., A modular three dimensional finitedifference ground-water flow model, U.S. Geological Survey Technique of Water-Resources Investigations, Book 6, Chap. A1. USGS: Reston, Virginia. McKenzie, J.M., Voss, C.I., Siegel, D.I., Groundwater flow with energy transport and water-ice phase change: Numerical simulations, benchmarks, and application to freezing in peat bogs. Advances in Water Resources 30(4) MDBC, Basin Salinity Management Strategy Murray-Darling Basin Commision: Canberra, Australia. Meire, D., De Doncker, L., Declercq, F., Buis, K., Troch, P., Verhoeven, R., Modelling river-floodplain interaction during flood propagation. Natural Hazards 55(1) Mensforth, L., Water use strategy of Melaleuca halmaturorum in a saline swamp / by Lisa Jane Mensforth, Dept. of Botany. The University of Adelaide: Adelaide. Miller, C.T., Gray, W.G., Hydrogeological research: just getting started. Ground Water 40(3) Misut, P.E., Voss, C.I., Simulation of seawater intrusion resulting from proposed expanded pumpage in New York City, USA, In: Cass, T.M., George, F.P. (Eds.), Developments in Water Science. Elsevier, pp Morel-Seytoux, H.J., Daly, C.J., A discrete kernel generator for stream aquifer studies. Water Resources Research Morita, M., Yen, B., Modeling of conjunctive two-dimensional surface three-dimensional subsurface flows. Journal of Hydraulic Engineering 128(2) Mumba, M., Thompson, J., Hydrological and ecological impacts of dams on the Kafue Flats floodplain system, southern Zambia. Physics and Chemistry of the Earth 30(6)

71 Nield, S., Townley, L., Barr, A., A framework for quantitative analysis of surface water-groundwater interaction: Flow geometry in a vertical section. Water Resources Research Niswonger, R.G., Prudic, D.E., Documentation of the streamflow-routing (SFR2) package to include unsaturated flow beneath streams a modification to SFR1, Techniques and Methods, Book 6,Chap. 13, Section A. USGS: Reston, Virginia. Noe, G.B., Zedler, J.B., Variable rainfall limits the germination of upper intertidal marsh plants in southern California. Estuaries 24(1) Oldenburg, C.M., Pruess, K., Dispersive transport dynamics in a strongly coupled groundwater-brine flow system. Water Resources Research Oostrom, M., Dane, J.H., Gu ven, O., Hayworth, J.S., Experimental investigation of dense solute plumes in an unconfined aquifer model. Water Resources Research 28(9) Oostrom, M., Hayworth, J.S., Dane, J.H., Guven, O., Behaviour of dense aqueous phase leachate plumes in homogeneous porous media. Water Resources Research Orlob, G.T., Ghorbanzadeh, A., Impact of water resource development on salinization of semi-arid lands. Agricultural Water Management 4(1-3) Oswald, S.E., Kinzelbach, W., Three-dimensional physical benchmark experiments to test variable-density flow models. Journal of Hydrology 290(1-2) Overton, I.C., Jolly, I.D., Slavich, P.G., Lewis, M.M., Walker, G.R., Modelling vegetation health from the interaction of saline groundwater and flooding on the Chowilla floodplain, South Australia. Australian Journal of Botany Panday, S., Huyakorn, P., A fully coupled physically-based spatiallydistributed model for evaluating surface/subsurface flow. Advances in Water Resources Partington, D., Brunner, P., Simmons, C.T., Therrien, R., Werner, A.D., Dandy, G.C., Maier, H.R., A hydraulic mixing-cell method to quantify the groundwater component of streamflow within spatially distributed fully integrated surface water-groundwater flow models. Environmental Modelling and Software 26(7) Peck, A.J., Salinization of non-irrigated soils and associated streams: a review. Australian Journal of Soil Research 16(2) Perkins, S.P., Koussis, A.D., Stream-aquifer interaction model with diffusive wave routing. Journal of Hydraulic Engineering 122(4) Person, M., Konikow, L.F., Recalibration and predictive reliability of a solute-transport model of an irrigated stream-aquifer system. Journal of Hydrology 87(1-2)

72 Peter, A., A plea for the restoration of Alpine rivers: Basic principles derived from the Rhone-Thur Case Study, In: Bundi, U. (Ed.), Alpine Waters. Springer: Berlin, Germany, pp Peyrard, D., Sauvage, S., Vervier, P., Sanchez-Perez, J.M., Quintard, M., A coupled vertically integrated model to describe lateral exchanges between surface and subsurface in large alluvial floodplains with a fully penetrating river. Hydrological Processes 22(21) Pinder, G.F., Bredehoeft, J.D., Cooper, H.H., Determination of aquiferdiffusivity from aquifer response to fluctuations in river stage. Water Resources Research Pinder, G.F., Sauer, S.P., Numerical simulation of flood wave modification due to bank storage effects. Water Resources Research 7(1) Prudic, D.E., Documentation of a computer program to simulate streamaquifer relations using a modular, finitedifference, ground-water flow model, Open-File Report USGS: Reston, Virginia. Prudic, D.E., Konikow, L.F., Banta, E.R., A new streamflow-routing (SFR1) package to simulate streamaquifer interaction with MODFLOW-2000, Open-File Report USGS: Reston, Virginia. Qu, Y., Duffy, C.J., 2007a. A semidiscrete finite volume formulation for multiprocess watershed simulation. Water Resources Research 43(8). Qu, Y., Duffy, C.J., 2007b. A semidiscrete finite volume formulation for multiprocess watershed simulation. Water Resources Research 43(8) W Raymond, J., Therrien, R., Gosselin, L., Borehole temperature evolution during thermal response tests. Geothermics 40(1) Reilly, T.E., Goodmann, A.S., Analysis of saltwater upcoming beneath a pumping well. Journal of Hydrology Rogers, B.R., Fools rush in, Part 3:Selected dryland areas of the world. Arid Lands Newsletter Rosenberry, D.O., Winter, T.C., Dynamics of water-table fluctuation in an upland between two prairie potholes wetlands in North Dakota. Journal of Hydrology Rosenbom, A.E., Therrien, R., Refsgaard, J.C., Jensen, K.H., Ernstsen, V., Klint, K.E.S., Numerical analysis of water and solute transport in variablysaturated fractured clayey till. Journal of Contaminant Hydrology 104(1-4) Ross, M.A., Tara, P.D., Geurink, J.S., Stewart, M.T., FIPR hydrologic model users manual and technical documentation, CMHAS Water Resources Report FIPR University of South Florida: Tampa. Rushton, K.R., Tomlinson, L.M., Possible mechanisms for leakage between aquifers and rivers. Journal of Hydrology

73 Saeijs, H.L.S., van Berkel, M.J., Global water crisis: the major issue of the 21st century, a growing and explosive problem. European Water Pollution Control Said, A., Stevens, D.K., Sehlke, G., Estimating water budget in a regional aquifer using HSPF-MODFLOW integrated model. Journal of the American Water Resources Association 41(1) Saiko, T.A., Zonn, I.S., Irrigation expansion and dynamics of desertification in the Circum-Aral region of Central Asia. Applied Geography 20( ). Sauvage, S., Teissier, S., Vervier, P., Am eziane, T., Garabetian, F., Delmas, F., Caussade, B., A numerical tool to integrate bio-physical diversity of a large regulated river: hydro-biogeochemical bases; the case of the Garonne River (France). River Research Applications Schincariol, R.A., Schwartz, F.W., Mendoza, C.A., Instabilities in variable density flows: stability and sensitivity analyses for homogeneous and heterogeneous media. Water Resources Research 33(1) Schmid, W., Hanson, R.T., III, T.M.M., Leake., S.A., User s guide for the Farm process (FMP) for the U.S. Geological Survey s modular three-dimensional finitedifference ground-water flow model, MODFLOW-2000, USGS Techniques and Methods 6-A17. USGS: Reston, Virginia. Schofield, R., Thomas, D.S.G., Kirkby, M.J., Casual processes of soil salinisation in Tunisia, Spain and Hungary. Land Degradation and Development Schubert, J., Hydraulic aspects of riverbank filtration-field studies. Journal of Hydrology Segol, G., Classic Groundwater Simulations: Proving and Improving Numerical Models. Shikaze, S., Sudicky, E., Schwartz, F., Density-dependent solute transport in discretely-fractured geologic media: is prediction possible? Journal of Contaminant Hydrology 34(10) Simmons, C.T., Variable density groundwater flow: from current changes to future possibilities. Hydrogeoly Journal Simmons, C.T., Narayan, K.A., Modelling density-dependent flow and solute transport at the Lake Tutchewop saline disposal complex, Victoria. Journal of Hydrology 206(3-4) Simmons, C.T., Narayan, K., Mixed convection processes below a saline disposal basin. Journal of Hydrology Simmons, C.T., Narayan, K.A., Woods, J.A., Herzceg, A.L., Groundwater flow and solute transport at the Mourguong salinewater disposal basin, southeastern Australia. Hydrogeoly Journal

74 Singh, V., Bhallamudi, S.M., Complete hydrodynamic borderstrip irrigation model. Journal of Irrigation and Drainage Engineering 122(4) Slavich, P.G., Walker, G.R., Jolly, I.D., Hatton, T.J., Dawes, W.R., Dynamics of Eucalyptus largiflorens growth and water use in response to modified watertable and flooding regimes on a saline floodplain. Agricultural Water Management 39(2-3) Smith, A., Townley, L., Influence of regional setting on the interaction between shallow lakes and aquifers. Water Resources Research Sophocleous, M.S., Perkins, P., Methodology and application of combined watershed and ground-water models in Kansas. Journal of Hydrology 236(3-4) Sophocleus, M., Interactions between groundwater and surface water: the state of science. Hydrogeoly Journal Sparks, T., Integrated modelling of 2-D surface water and groundwater flow with contaminant transport, 31st IAHR congress: Seoul, pp Steppuhn, H., Pre-irrigation of a severely-saline soil with in-situ water to establish dryland forages. American Society of Agricultural Engineers 44(6) Steppuhn, H., Wall, K.G., Canada's salt tolerance testing laboratory. Canadian Agricultural Engineering 41(3) Sudicky, E., Jones, J., Park, Y.-J., Brookfield, A., Colautti, D., Simulating complex flow and transport dynamics in an integrated surface-subsurface modeling framework. Geosciences Journal 12(2) Sulis, M., Paniconi, C., Rivard, C., Harvey, R., Chaumont, D., Assessment of climate change impacts at the catchment scale with a detailed hydrological model of surface-subsurface interactions and comparison with a land surface model. Water Resources Research 47(1). Swain, E.D., Wexler, E.J., A coupled surfacewater and ground-water flow model (MODBRANCH) for simulation of stream-aquifer interaction, USGS Techniques of Water-Resources Investigations, Book 6, Chap. A6. USGS: Reston, Virginia. The International HYDROCOIN Project, Level 1: Code verification, SKI and OECD/NEA, OECD/NEA: Paris, p Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-Dimensional Numerical Model Describing Fully- Integrated Subsurface and Surface Flow and Solute Transport. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. Therrien, R., Sudicky, E.A., Three-dimensional analysis of variablysaturated flow and solute transport in discretely-fractured porous media. Journal of Contaminant Hydrology 23(1-2)

75 Townley, L., Davidson, M., Definition of a capture zone for shallow water table lakes. Journal of Hydrology Townley, L., Trefry, M., Surface water-groundwater interaction near shallow circular lakes: Flow geometry in three dimensions. Water Resources Research van der Kwaak, J., Loague, K., Hydrologic-response simulations for the R-5 catchment with a comprehensive physics-based model. Water Resources Research 37(4) van der Kwaak, J.E., Numerical simulation of flow and chemical transport in integrated surface-subsurface hydrologic systems. University of Waterloo: Waterloo, Canada. Volker, R.E., Rushton, K.R., An assessment of the importance of some parameters for seawater intrusion and a comparison of dispersive and sharpinterface modelling approaches. Journal of Hydrology Voss, C.I., SUTRA: A finite-element simulation model for saturated unsaturated fluid-density-dependent groundwater flow with energy transport or chemically reactive species solute transport, Report US Geology, Survey and Water Resources Investigations: Reston, USA. Walker, K.F., Thoms, M.C., Environmental effects of flow regulation on the lower River Murray, Australia. Regulated Rivers: Research and Management Weatherill, D., Graf, T., Simmons, C.T., Cook, P.G., Therrien, R., Reynolds, D.A., Discretizing the fracture-matrix interface to simulate solute transport. Ground Water 46(4) Weatherill, D., Simmons, C.T., Voss, C.I., Robinson, N.I., Testing densitydependent groundwater models: Two-dimensional steady state unstable convection in infinite, finite and inclined porous layers. Advances in Water Resources 27(5) Weng, P., Sanchez-Perez, J., Sauvage, S., Vervier, P., Giraud, F., Hydrological modelling to characterise the riparian wetland of a large alluvial river (Garonne River, France). Hydrological Processes Wilcox, L.E., Bowman, R.S., Shafike, N.G., Evaluation of Rio Grande management alternatives using a surfacewater/ground-water model. Journal of the American Water Resources Association 43(6) Winter, T.C., Relation of streams, lakes, and wetlands to groundwater flow systems. Hydrogeoly Journal Winter, T.C., Harvey, J.W., Lehn Franke, O. and Alley, W.M., Groundwater and surface water- a single resource, Circular US Geological Survey, US Department of the Interior. Woessner, W.W., Stream and fluvial plain groundwater interactions: Rescaling hydrogeologic thought. Ground Water 38(3)

76 Wooding, R.A., Tyler, S.W., White, I., Convection in groundwater below an evaporating salt lake. 1. Onset of instability. Water Resources Research 33(6) Worcester, B.K., Brun, L.J., Doering, E.J., Classification and management of saline seeps in western North Dakota. North Dakota Farm Research 33(1) 3-7. Wurster, F.C., Cooper, D.J., Sanford, W.E., Stream/aquifer interactions at Great Sand Dunes National Monument, Colorado: influences on interdunal wetland disappearance. Journal of Hydrology Zedler, J.B., Covin, J., Nordby, C., Williams, P., Boland, J., Catastrophic events reveal the dynamic nature of salt-marsh vegetation in Southern California. Estuaries 9(1) Zedler, J.B., West, J.M., Declining diversity in natural and restored salt marshes: A 30-year study of Tijuana Estuary. Restoration Ecology 16(2) Zobrist, J., Water chemistry of Swiss Alpine rivers, In: Bundi, U. (Ed.), Alpine Waters. Springer: Berlin, Germany, pp

77 3 Modelling the impacts of river stage manipulation on a river-floodplain system Overview Chapter 3 investigates and quantifies the impacts of river stage manipulation, as a salt interception measure, on floodplain salinity and its implications for vegetation health. To do this, twelve hypothetical scenarios are defined and evaluated using the calibrated fully integrated, physically-based model, HydroGeoSphere. As the study site (Site C in Clark s floodplain) is complex and strongly influenced by another factor (groundwater lowering via the SIS), one scenario without groundwater lowering was included as well. It is shown that groundwater lowering is able to mitigate floodplain salinity by creating a divide which prevents saline water from reaching the floodplain. It appears that the SIS is successful in intercepting saline groundwater that would otherwise have entered the river. Furthermore, it is discussed how the impacts of river stage manipulation are less compared to groundwater lowering over the whole floodplain. Also, the effects are limited to the extent that river stage manipulation can change the hydraulic gradient. It is shown that a higher river stage leads to relatively less solute mass in the floodplain aquifer. The reason is the reduction of flux from the regional groundwater to the floodplain aquifer due to the enhanced hydraulic head towards the floodplain aquifer. Moreover, mobilization of the solute mass stored in the unsaturated zone is quantified and the threedimensional distribution of solute mass mobilization in the floodplain aquifer is visualized. However, the impact on floodplain salinity attributed to river stage manipulation is restricted to the near-river zone which can be beneficial for riparian vegetation. Overall, river stage 56

78 manipulation may be considered as a short term management technique as it cannot entirely change the natural condition of the floodplain. Therefore, if longer term strategies are required, it may be possible to implement these salt interception measures periodically. This chapter was published as a paper in the journal Environmental Modelling and Software (Paper 2). The manuscript was co-authored by my Principal Supervisor, Prof. Simon Beecham, and advisor, Dr. Ali Hassanli. Also, Mr. Ian Jolly and Dr. Kate Holland (research scientist at CSIRO) and Dr. Juliette Woods (research fellow at Flinders University) contributed to this paper. The format of the paper has been changed to be consistent with the rest of this thesis. 57

79 Paper 2: Modelling the Impacts of River Stage Manipulation on a Complex River-Floodplain System in a Semi-Arid Region Published as: Alaghmand, S., Beecham, S., Jolly, I. D., Holland, K. L., Woods, J. A. and Hassanli, A. (2014), Modelling the Impacts of River Stage Manipulation on a Complex River-Floodplain System in a Semi-Arid Region, Environmental Modelling and Software, Elsevier, 59, pp Abstract: This paper investigates the complex interaction between a river and a saline floodplain in a semi-arid environment strongly influenced by groundwater lowering using a fully integrated physically-based numerical model. The main objective is to quantify the impacts of river stage manipulation on freshening of the shallow floodplain groundwater through bank storage. It is shown that river stage rises produce a relatively less saline floodplain aquifer with a larger freshwater lens. First, an increase in river stage reduces saline groundwater recharge to the floodplain. Second, the enhanced bank storage is able to freshen the groundwater near the river banks during high-flow pulses by mixing fresh water with saline groundwater. Overall, it was found that river stage manipulation may be considered as a short term salt management technique. However, if longer term strategies are required, it may be possible to implement these salt interception measures periodically. 3.1 Introduction Floodplain salinization in arid and semi-arid regions In arid and semi-arid environments, groundwater can be a major component of the water cycle (Ghazavi et al., 2012; Jolly et al., 2008). In these regions rainfall is typically seasonal, highly variable and significantly less than the evapotranspiration rate; therefore little, if any, diffuse groundwater recharge can occur. These factors create a natural tendency for salt accumulation in soils and groundwater. In floodplain environments, periodic natural overbank floods may prevent the development of soil and groundwater salinity (Meire et al., 2010; 58

80 Restrepo et al., 1998; Zimmermann et al., 2006). Under natural conditions, arid and semi-arid floodplains occasionally experience periods of higher salinity as a consequence of high evaporation conditions and the variability of natural overbank floods, which provide dilution and flushing of the stored salt. However, due to the impacts of human population expansion and associated changes in land use, surface water regulation, and water resource depletion, arid and semi-arid floodplains, such as those in south-eastern Australia, are now often experiencing extended periods of low surface water flows and high soil salinity (Allison, 1990; Holland et al., 2013; Jolly et al., 2008). Consequently, the dynamic equilibrium of salinization and leaching is interrupted (Jolly et al., 1993). This can result in the reduced leaching of accumulated salt from root zones, thereby causing the dieback of environmentally important riparian vegetation, such as red gum (Eucalyptus camaldulensis) and black box (Eucalytpus largiflorens) and a decline in river water quality (Allison, 1990; Herczeg, 1993; Jolly et al., 1996; Peck, 1973, 2003) Physical processes Over the years there has been a growing awareness that the floodplain is a key area where a number of hydrogeological processes operate that have the potential to influence future outcomes of salinity management activities and river operations (Doble et al., 2006; Eslamian and Nekoueineghad, 2009; Evans et al., 2013; Ghazavi et al., 2012; Holland et al., 2009c). Therefore, studies have been conducted to develop models to quantify the impact of hydrological changes in river flow and floodplain ecology (Straatsma et al., 2013). A number of processes affect the flux exchange between a river and a floodplain in arid and semi-arid environments. These include rainfall, regional groundwater recharge, bank storage, evapotranspiration and groundwater extraction. Recharge from rainfall is often negligible in arid and semi-arid regions (Rassam et al., 2013). Floodplains are generally topographically low in the landscape. Hence, the main recharge process in a floodplain aquifer is often groundwater flow from surrounding regional aquifers (Doble et al., 2006; Evans et al., 2013; Ghazavi et al., 2012). In Australia, the regional groundwater is usually naturally saline and often is the main source of solute movement towards the floodplain landscape. 59

81 Groundwater recharge may be increased due to increased irrigation practices in the surrounding highland and this can lead to groundwater mounds (Figure 3.1). (a) (b) Figure 3.1 Schematic of the SW-GW interaction across the Lower Murray River area before (a) and after (b) human-induced activities including weir and lock installations and irrigation practices In arid and semi-arid regions, bank storage is an important process in the interaction between the surface and groundwater domains especially in rivers with high riverbed and riverbank hydraulic conductivities. Bank recharge represents a gain to the groundwater system. Three types of groundwater recharge were hypothesized by Jolly (2004) include bank recharge, diffuse recharge and localized recharge. Diffuse and localized recharge may occur during overbank flow which is not the focus of this paper. Bank storage is a dynamic phenomenon in which aquifer recharge occurs during periods of river stage rise followed by aquifer discharge to the river when the river stage reverts to a normal lower level 60

82 (Ghazavi et al., 2012). During river stage recession, groundwater discharges to the river. The discharged groundwater usually has a solute concentration intermediate between that of the river and that of regional groundwater (McCallum et al., 2010). The net result of these processes at any point in space and time can lead to either a gaining or a losing river (Rassam, 2011). The observed response in rates of aquifer-floodplain exchange to changes in river stage strongly depends upon the state of connection between the two domains (Li et al., 2009). The significance of bank storage depends primarily upon the size of the river floodplain and its hydraulic and geometric properties (Doble et al., 2012; Knight and Rassam, 2007). Bank storage results in freshening of groundwater located near banks during high-flow pulses through the mixing of fresh river water with saline groundwater. For example, Holland et al. (2009a) showed that improvement in floodplain tree health was proportional to the extent of bank storage at different locations around a floodplain environment. Groundwater evapotranspiration combines two processes: evaporation from groundwater lying close to the ground surface and transpiration from plants that use groundwater. In lowland gaining river-floodplain systems, groundwater flowing from the regional aquifer moves through the floodplain before either flowing to the river or being attenuated by evapotranspiration. A shallow groundwater presence within floodplains usually means that evapotranspiration rates are significant (Doble et al., 2006). Groundwater is often extracted via production wells for various purposes such as irrigation, water supply and groundwater lowering (as a salt interception measure). Depending on the size and hydrogeology of the floodplain and pumping rate of the production wells, groundwater lowering can significantly influence the SW-GW interactions (Rassam et al., 2013). The unsaturated zone is often cited as the salt storage location during inter flood periods. This is particularly the case in arid and semi-arid regions where the evapotranspiration rate is much higher than rainfall, which creates unsaturated solute storage zones in some areas of the floodplain. This is accelerated with changes in land use (such as irrigation recharge in the highland and floodplain), surface water regulation (such as raised groundwater level in the floodplain and 61

83 reduction of high flow pulses) and water resource depletion (Figure 3.1). Solute mass stored in the unsaturated zone appears to be correlated with the underlying groundwater salinity (Jolly, 2004) Research challenges Research and investigations into floodplain processes have mostly occurred over the last decade (Evans et al., 2013). Several studies have described some of the challenges of modelling SW-GW interactions in arid and semi-arid floodplains (Rassam et al., 2013). In addition, Rassam et al. (2013) highlighted the importance of incorporating SW-GW interactions into river management models that are used by water managers. In fact, uncertainty reduction will be very worthwhile since worldwide there is significant investment in water projects (Straatsma et al., 2013). The level of understanding of arid and semi-arid floodplain environments is still relatively basic, particularly in relation to SW-GW interactions in floodplains due to the highly complex nature of the floodplain environment (Eslamian and Nekoueineghad, 2009; Evans et al., 2013). McEwan et al. (2006) and Jolly et al. (2008) emphasized that while SW-GW interactions in temperate regions have been investigated in several studies (Hoehn and Scholtis, 2011; Kollet et al., 2010; Meire et al., 2010), floodplains in arid and semi-arid regions have received far less attention. However, many SW-GW modelling studies model water flow only and neglect solute transport (Crowe et al., 2004; Restrepo et al., 1998; Walton et al., 1996). Indeed, despite the fact that aquifer and surface waters are hydraulically interconnected, they are often modelled as two separate systems and are analysed independently (Schmid et al., 2006). The first published studies featuring fully coupled groundwater flow and solute modelling in relation to floodplain and river ecology have been undertaken relatively recently (Bauer et al., 2006a; Bauer et al., 2006b; Langevin et al., 2005; Zimmermann et al., 2006). But, these studies generally did not model unsaturated zone processes, which are important to the prediction of ecological responses (Jolly, 2004). Another limitation of such studies is that many assumed steady-state GW and/or SW flow conditions, and these rarely exist in most arid and semi-arid floodplains, as these typically undergo periodic cycles of wetting and drying, resulting in transient SW-GW interactions (Jolly et al., 2008). Another issue is the high data requirements for this type of model as well as the lack of understanding 62

84 of the key role of salinity in arid and semi-arid floodplains (Hart et al., 1991). There is a clear need to develop modelling capabilities for the movement of salt to, from, and within floodplains (Evans et al., 2013; Jolly et al., 2008). This can be addressed through developing a 3D physically-based fully integrated surfacesubsurface numerical model with variable saturation and solute transport simulation capabilities (Alaghmand et al., 2013b) Objective Various management strategies can be used to maintain floodplain health. These include pumping saline groundwater, injection of fresh water, localised artificial flooding and environmental irrigation. These management measures require an understanding of SW-GW interaction at a fine scale, and the true ecological impact of land management decisions requires knowledge of the floodplain salinization risk. This paper investigates the complex interaction between a river (the Murray River in South Australia) and a saline floodplain (Clark s Floodplain) in a semi-arid area using a fully integrated physically-based numerical model featuring variable saturation and solute transport simulation capabilities. Clark s Floodplain is chosen as the study area due to the availability of sufficient recorded data which allows the development of a detailed unsteady-state (dynamic) model. However, the study period, which is from 1/01/2005 to 2/09/2010, is limited to just under five years and the scenarios are representative of transient behaviour that may not illustrate the system moving to new equilibrium positions. In addition, the study period corresponds to very dry conditions and so rainfall fluxes and river stage variations could have been much less than long-term variations. The main objective is therefore to quantify the impacts of river stage manipulation on freshening of the shallow floodplain groundwater through bank storage. The river-floodplain system is complex as the floodplain aquifer is strongly influenced by a Salt Interception Scheme (SIS) that involves groundwater lowering. Hence, various scenarios are defined to understand the combined and individual impacts of river stage manipulation and groundwater lowering on the flow and the solute dynamic of the floodplain aquifer. The hypothesis that is tested here is that higher river stages lead to a relatively less saline floodplain aquifer by increasing the fresh river water flux to the floodplain aquifer and reducing the saline groundwater flux from the highland to the floodplain aquifer. 63

85 3.2 Materials and methods A fully integrated surface-subsurface numerical model is developed for Clark s Floodplain, as described in detail below. The model is calibrated to data from a time period that includes operation of the SIS production wells. Scenarios are used to determine the relative impacts of river stage manipulation on water and solute balances within the floodplain aquifer Study site Clark s Floodplain is located on the Lower Murray River in South Australia (34 21'S, 'E) (Figure 3.2) next to the Bookpurnong Irrigation District. The study site is located in a semi-arid region of South Australia, with annual rainfall varying between 200 and 300 mm and annual areal potential evaporation of 1800 mm. Data from Loxton meteorological station shows that local evaporation rates were continuously higher than rainfall depths between 2005 and 2010 (BOM, 2013). At the study site the Coonambidgal Clay (typically consisting of clays and silts) ranges from 2 to 7 m thick, while the Monoman Formation (coarse-grained quartz sands) is approximately 7 m thick (Figure 3.2). The highland adjacent to the floodplain consists of a layer of Loxton Sands (Upper and Lower units) up to 35 m in depth. The whole area is underlain by the Loxton Sand and Bookpurnong Beds, the latter acts as an aquitard basement to the shallow aquifer that includes the Monoman Formation and Loxton Sands (AWE, 1999; Barnett et al., 2002; Doble et al., 2006). For further details on the hydrogeology of the study site the reader is referred to Doble et al. (2006) and Alaghmand et al (2013a). The increased groundwater recharge contributed by the Bookpurnong Irrigation District has locally raised the water-table in the Loxton Sands (Telfer, 1999). The increased groundwater gradient between the Loxton Sands aquifer and the Murray River has led to greater salt flux from the saline regional aquifer into the floodplain and river. Groundwater salinity in the floodplain has also increased due to a lack of floods that could potentially freshen the groundwater via bank storage. Black box and red gum tree communities have been most affected by the salinization of the floodplain (Doble et al., 2006). Groundwater salinity in the Loxton Sands-Monoman Formation aquifer is typically in excess of 50,000 µs cm -1, while irrigation recharge salinity is typically 8,000 µs cm -1. In an effort to 64

86 mitigate such impacts, salt interception schemes (SISs) have been implemented at various sites along the Lower Murray River, which intercept saline groundwater before it reaches the river (Holland et al., 2009b). Two of the SIS production wells, 32FP and 34FP, are located at the study site and these have significant impacts on the SW-GW interactions. Meteorological, hydraulic and hydrogeological data for the Clark s Floodplain site are well-documented for the modelled time period (1/1/2005-2/09/2010) and were used to inform initial and boundary conditions for the model. The time period was chosen to include the non-flooding conditions and a flood event which occurred in Recorded meteorological data, such as rainfall and potential evaporation, were obtained from Loxton meteorological station (BOM, 2013). Murray River flows at the study site were obtained from Lock 4 water level station, which is situated immediately upstream of Clark s Floodplain (WaterConnect, 2013) (Figure 3.2). River salinity observations were obtained from Clarke s Sandbar and Rilli Island stations located upstream and downstream of the study site, respectively (WaterConnect, 2013). Floodplain potentiometric head and solute concentration observations were obtained from groundwater observation wells located along two transects that extend from the river to the floodplain perimeter: B1, B2 and B3 on transect 1; and B4, B5 and B6 on transect 2 (Figure 3.2) which were adopted from Berens et al. (2009) and Holland. Built in 2004 and 2005, the observation wells were designed to monitor groundwater levels and salinity at depths from 2 m up to 10 m, since salinity is anticipated to occur at the top of water table (Berens et al., 2009) Numerical model Characterisation of near-river-aquifer systems is complex because of the nature of SW-GW interaction processes, and the uncertainty of land cover and aquifer properties, which can produce significant errors in hydrodynamic models outputs (Eslamian and Nekoueineghad, 2009; Sophocleous, 2010; Straatsma et al., 2013). This can be addressed using a fully-integrated, physically-based numerical model (Alaghmand et al., 2013b). Due to the required capabilities, the available observed input data, the scale of the study, and the required robustness and stability of the numerical methods, the HGS model (Therrien et al., 2006) was selected for this 65

87 research. HGS is a three-dimensional numerical model describing fully-integrated surface and subsurface flow and solute transport. HGS models the flow of water through unsaturated porous media by numerical solution of the Richards equation. The van Genuchten (1980) or Brooks-Corey (Brooks and Corey, 1964) relationships are used to relate pressure head to saturation and relative hydraulic conductivity. Surface water flow is modelled using two-dimensional depthaveraged flow. Saturated groundwater flow is modelled by numerical solution of the groundwater flow equation. Two surface and subsurface coupling approaches are available in HGS, namely the common node approach (based on continuity of hydraulic head between two domains) and the dual node approach (based on a first-order exchange coefficient), with the latter being used in this study. Transpiration from vegetation occurs within the root zone of the subsurface and is a function of the leaf area index (LAI), nodal water (moisture) content and a root distribution function (RDF) over a prescribed extinction depth. Evaporation from the soil surface and subsurface soil layers is a function of nodal water content and an evaporation distribution function (EDF) over a prescribed extinction depth. The model assumes that evaporation occurs along with transpiration, resulting from energy that penetrates the vegetation cover. For further details on the code and a recent software review the reader is referred to Therrien et al., (2006) and Brunner and Simmons (2012). HGS requires pre- and post-processor tools in order to handle input preparation (complex topography and grid) and visualization of the outputs. In this study, Grid Builder (McLaren, 2005) and Groundwater Modelling System (GMS) (AquaVeo, 2011) were used to generate the model grid. GMS was also used to visualize and interpret the model outputs. In this study, HGS used the control volume finite element approach to solve surface and subsurface flow and transport. The model was a transient model setup for a period of 2070 days (from 1/01/2005 to 2/09/2010) using an initial time step of 0.1 days, a maximum time step of 1 day and a maximum time step multiplier of The model solves non-linear equations for variably-saturated subsurface flow, surface flow and solute transport. To solve the non-linear equations, HGS uses the Newton-Raphson linearization method. Newton iteration parameters include Newton maximum iterations (25), Jacobian epsilon (10.0 d -5 ), Newton absolute convergence criteria (1.0 d -5 ), Newton residual convergence criteria (1.0 d -3 ) and flow solver maximum iterations (1.0 d 5 ). 66

88 Figure 3.2 a: Location of Clark s floodplain in Australia (shown in purple), b: Perimeter of the geometry model (shown in red), c: 3D visualization of the geometry of the study site including the soil types and observation (B1, B2, B3, B4, B5, B6, 31F, 33F and 35F) and SIS production wells (32F and 34F) (Z magnification= 8). The cover image is adopted from GoogleMaps Model set up Geometry grid The model domain perimeter is shown in Figure 3.3a. The model spatial discretisation is based on a LiDAR Digital Elevation Model of the study site with a 10 m grid resolution. The resulting grid consisted of 78,624 nodes and 143,500 elements. As shown in Figure 3.3a, the geometric grid covers 61.3 ha of Clark s Floodplain from the floodplain slope break to the Lower Murray River main channel. This includes two SIS production wells (32F and 34F) and nine observation wells (Figure 3.3a). In this case, the length of the river bank is 570 m and the distance from the river bank to the SIS wells varies between 480 m and 650 m. 67

89 Parameters Three soil types were represented, namely a continuous 10 metre-thick layer of Monoman Formation sand, overlaid by a spatially variable, 2 to 6 metre-thick layer of semi-confining heavy Coonambidgal Clay and Upper Loxton Sand in the adjacent highland (Figure 3.3). Soil properties, such as hydraulic conductivity (isotropic), porosity, residual saturation and specific storage, were obtained from Carsel and Parrish (1988) and Doble et al. (2006). Van Genuchten function parameters (n and alpha) (van Genuchten, 1980) were adopted from Jolly et al. (2008) who adjusted and proposed these parameters for the Lower Murray River soil types. Longitudinal and transverse solute dispersivity values were estimated through model calibration. The hydraulic properties of the surface domain (river bed and floodplain corridor) have significant differences and so these were divided in the model into main channel (river) and floodplain. Furthermore, the vegetation coverage of the floodplain was divided into two different categories (Eucalyptus trees and grass) and evapotranspiration parameter values for both categories were adopted from Hingston et al. (1997), Banks et al. (2011) and Verstrepen (2011). Table 3.1 summarises the parameters values used in the numerical model. Table 3.1 Parameter values of the model for the study site Model parameter Value Units Subsurface domain Monoman Coonambidgal Upper Sand Clay Loxton Sand Porosity % Hydraulic conductivity m d -1 Specific storage 1.6 x x x 10-4 m -1 Evaporation limiting saturation (min) Evaporation limiting saturation (max) Longitudinal dispersivity m Transverse dispersivity m Residual water content a m -1 n Evapotranspiration Eucalyptus Grass Tree canopy evaporation 4.5 x x 10-4 m Evaporation extinction depth defined by quadratic decay Evaporation distribution m

90 function Transpiration extinction depth defined by quadratic decay Root distribution function m Leaf area index m 2 m -2 Transpiration fitting parameter c Transpiration fitting parameter c Transpiration fitting parameter c3 1 1 Transpiration limiting saturation (at wilting point) Transpiration limiting saturation (at field capacity) Transpiration limiting saturation (at oxic limit) Transpiration limiting saturation (at anoxic limit) Initial interception storage 3.0 x x 10-4 m Surface domain River Floodplai n Friction (x-plane) 5.0 x x 10-2 T L -1/3 Friction (y-plane) 5.0 x x 10-2 T L -1/3 Rill storage height 1.0 x x 10-2 m Coupling length 1.0 x x 10-2 m Obstruction storage height x 10-3 m Boundary conditions The locations where boundary conditions were specified are given in Figure 3.3a. Two types of boundary conditions were used in the model including first-type (Dirichlet) boundaries of prescribed head/concentration and second-type (Neumann) boundaries of prescribed flow/solute flux. In the subsurface (porous media) domain, a constant first type (Dirichlet) boundary condition of 12 m AHD (Australian Height Datum) constant head was specified at the north-eastern part of the domain. This condition was adopted based on potentiometric contours (AWE, 2013). The observed river levels for the surface domain were set at the river side of the model using a time-varying first-type (Dirichlet) boundary condition. In this regard, the observed water levels downstream of Lock 4 were applied to the river nodes of the model (WaterConnect, 2013). To represent the solute boundary conditions, a first-type (Dirichlet) constant concentration boundary condition was assigned. The observed groundwater concentrations at the observation wells in the river and the floodplain ranged from 300 µs cm -1 to 50,000 µs cm -1 (Holland et al., 2013). Hence, constant values were

91 applied at the subsurface outer boundary (representing regional groundwater in the highland aquifer) and the river nodes accordingly. Two SIS production wells (30F and 32F) were represented in the model using recorded pumping rates and durations obtained from Berens et al. (2009). Rainfall was modelled for the entire model surface domain beginning on day 1 using a time-varying second-type (Neumann) boundary condition according to recorded data (BOM, 2013). Evapotranspiration was dynamically modelled as a combination of evaporation and transpiration processes by removing water from all model cells of the surface and subsurface flow domains within the defined zone of the evaporation and root extinction depths Initial conditions The initial conditions for the calibration model were obtained from a steady-state flow and transport model that represented the status of the river-floodplain system prior to the study period (2005). The initial model was ran for long enough (30 years) to reach the equilibrium condition (Barnett et al., 2012). Hydraulic head and solute concentration outputs from the initial model compared favourably with observations from six observation wells on Clark s Floodplain recorded in 2005 which was available through Holland et al. (2013). Also, the status of the solute concentration distribution at the beginning of the study period was checked with the general observed solute distribution pattern in the floodplain. This can be considered as two zones: a relatively fresh GW zone within 50 m distance of the river banks (B1: 6,500 µs cm-1 and B4: 1,200 µs cm-1); and a saline zone (B2: 53,000 µs cm-1, B3: 54,000 µs cm-1, B5: 50,900 µs cm-1 and B6: 52,000 µs cm-1) for the rest of the floodplain (See Figure 3.6c) Coupled flow and transport calibration Calibration was undertaken using an iterative trial-and-error method. In order to minimize the uncertainty associated with parameters such as hydraulic conductivity, porosity, dispersivity (longitudinal and transverse) and leaf area index, these were altered within known ranges and reasonable limits in order to achieve an acceptable match to observations of hydraulic head and solute concentrations pattern. Two different approaches were employed for the flow and solute calibrations. While, the aim of the calibration process for flow is to match 70

92 the absolute groundwater heads at the observation wells, the solute is calibrated to the observed concentration patterns. This is because concentration patterns are much more sensitive to local-scale geological heterogeneity than are hydraulic heads, and models may have difficulty reproducing the concentrations or their temporal variability at single observation wells (Barnett et al., 2012). (a) (b) Figure 3.3 a: Configuration of the model boundary conditions, b: Configuration of the vegetation and soil layers of Clark s Floodplain along transect 1 (Z magnification= 3). Observation wells are shown as black columns Numerical model performance evaluation In order to obtain a reasonable evaluation of the numerical model performance, several factors need to be taken into account. These may include the field of application, characteristics of the model, available observed data, information and knowledge of the problem, and the specific objectives of the modelling exercise (Bennett et al., 2013; Jakeman et al., 2006; Matthews et al., 2011). Moreover, environmental models typically have multiple interacting drivers with uncertain 71

93 properties (Bennett et al., 2013; Rassam et al., 2013). Hence, multiple evaluation metrics need to be used for a comprehensive evaluation of the numerical model. Otherwise, a single performance criterion approach may lead to counterproductive results such as favouring models that do not reproduce important features of a system (Bennett et al., 2013; Gupta et al., 2012). A number of quantitative approaches are available to assess the model performance (Bennett et al., 2013). For instance, direct value comparison methods aim to test whether the modelled values show similar characteristics as a whole to the observed values. In this case, the means of the modelled and observed data sets are compared and expressed as Means Differences in Table 3.2. Clearly the ideal value would be zero. Furthermore, some model performance evaluation methods, such as the residual method, involve coupling observed and modelled values. In the residual method the difference between modelled and observed data are calculated. Of the many possible numerical calculations on model residuals, Mean Square Error (MSE) and Root Mean Square Error (RMSE) are considered here. The ideal value for both of these metrics is zero. Another model performance evaluation metric involves preserving the data patterns. This method tests the ability of the model to preserve the patterns of observed and modelled data. The Coefficient of Determination (r 2 ) is one of the metrics in this category which indicates how variation of one variable is explained by a second variable. This is commonly used to measure the efficiency of a model and values range between 0 and 1. Another Coefficient of Determination which is popular in hydrologic modelling is the Nash-Sutcliffe Model Efficiency (NSE) (Nash and Sutcliffe, 1970). This ranges from to 1 and indicates how well a model explains the variance in the observations, compared with their mean as the prediction. The ideal value for both of these metrics is one. A detailed description of the qualitative and quantitative methods of characterising performance of environmental models is provided by Bennett et al. (2013). In addition to the above mentioned quantitative model performance evaluation methods, visual performance measures have been developed to mimic how the eye evaluates proximity between observed and modelled values (Ehret and Zehe, 2011; Ewen, 2011). This type of qualitative method avoids traducing model errors simply in terms of difference of magnitude and, also includes time shifts. In fact, 72

94 qualitative assessments are important in complex models as they enable the modeller to sketch out trends and system behaviour rather than producing actual values for variables (Bennett et al., 2013) Scenarios The calibrated model represents the observed SW-GW interaction from 1/1/2005 to 2/09/2010. During the study period the SIS was in operation and there was no river manipulation (hereafter referred to as the Only-SIS scenario). To predict SW-GW interactions induced by river stage manipulation, the calibrated model was re-run while imposing various river stage elevations. Hence, twelve hypothetical river stage manipulation scenarios were defined for water stage increases of 0.5 m, 1.0 m and 1.5 m and for a decrease of 0.5 m. Each of these water stage changes were modelled for 1 month, 2 months, and 3 months in each year. Model simulations covered the period between 1/1/2005 and 2/9/2010 (2070 time steps). The response of the floodplain aquifer to the various scenarios was observed in terms of hydraulic heads and solute dynamics. River stage manipulation was not the only stress on the model during the study period as groundwater lowering also occurred via the operation of the SIS production wells. Hence, one scenario without groundwater lowering (hereafter referred to as the No-salt management scenario) was included as well. Figure 3.4 shows the hydrographs for the manipulated river stage elevations for the defined scenarios. 3.3 Results and discussion Results of the numerical model are discussed in five sections. First the results of the calibrated model are demonstrated and discussed along with No-salt management scenario. This is followed by discussion of the model results in terms of water balance and solute balance for the defined scenarios. Then, solute mass in the unsaturated zone is analysed and finally, the ecological implications of river stage manipulation are discussed. 73

95 (a) (b) (c) Figure 3.4 Time-varying river stage boundary conditions for scenarios featuring river stage rise durations of (a) one month, (b) two months, and (c) three months Calibrated model The calibrated model represents the observed behaviour of the river-floodplain system in terms of water and solute dynamics over the period 1/01/2005 to 2/09/2010. The numerical model performance in terms of groundwater head was tested both qualitatively and quantitatively. The observed and modelled series of hydraulic heads and solute concentrations at observation wells B1, B2, B3, B4, B5 and B6 were also compared visually (Figure 3.5). Moreover, quantitative evaluation was undertaken using the model performance evaluation metrics discussed in Section 2.4 including Means Difference, Coefficient of Determination (r 2 ), Mean Sum of Error (MSE), Root Mean Squared Error (RMSE) and Nash-Sutcliffe Model Efficiency (NSE) (Table 3.2). Considering Figure 3.5 and Table 3.2, it appears that the numerical model properly reproduced the observed groundwater dynamic during the study period. 74

96 The modelled groundwater salinity distribution result displays the presence of freshwater along the eastern margin abutting the river channel and a saline zone in the rest of the floodplain aquifer (Figure 3.6a). The observed groundwater salinities at the location of the observation wells are shown in Figure 3.6c (Holland et al., 2013). This shows that the observation wells in the vicinity of the river bank (B1 and B4) have significantly lower salinity than the other four observation wells located further away on the floodplain (B2, B3, B5 and B6). Moreover, an EM31 survey was conducted in November 2007 and reported by Berens et al. (2009). Depending on subsurface conductivity, the EM31 has a limited penetration depth of approximately 4-6 m, and yields a bulk conductivity representation of that shallow interval. Variables that may typically influence the results of the EM31 survey include groundwater depth and salinity, variations in soil moisture and salinity, and the clay content. However, with the Murray River floodplains consisting mainly of sands and localised clays of similar porosity, the water content in the saturated environment is most likely consistent, leaving salinity as the main driver of conductivity. Hence, the EM31 results can be a proper indicator of groundwater salinity at the study site. This is shown in Figure 3.6b. The modelled groundwater salinity distribution (Figure 3.6a) and conductivity distribution obtained from the EM31 survey in November 2007 (Figure 3.6b) present a good agreement. Overall, it is confirmed that the calibrated model is able to reproduce the solute dynamic of the surface-groundwater interaction processes in an acceptable manner as it is consistent with the observed data. Table 3.2 Model performance evaluation metrics (Means Difference; MSE = Mean Square Error; RMSE = Root Mean Square Error; r 2 = Coefficient of Determination; NSE = Nash-Sutcliffe Model Efficiency coefficient) Observation Well Means difference (m) MSE (m) RMSE (m) r 2 NSE B B B B B B

97 Figure 3.5 Modelled and observed groundwater heads at the observation wells. River stage and modelled and observed groundwater heads are shown as blue lines, black lines and red dots, respectively. The light blue pattern represents the periods during which the SIS production wells were in operation. Figure 3.7 shows the groundwater balance for Only-SIS and No-salt management scenarios. As rainfall recharge is unlikely to happen at the study site, the river and regional groundwater are the dominant recharge features in both models. As a general trend, the bank storage strongly responds to the river stage fluctuation, while regional groundwater recharge is a function of the operation of the SIS production wells. Both bank storage and regional groundwater are larger during the operation of the SIS production wells (Figure 3.7a). This shows the boosted hydraulic gradient towards the SIS production wells during their operation. In Only-SIS model, when the SIS is shut down for a short period (November April 2007), the dominant recharge feature is regional groundwater. In this period 76

98 EC (µs.cm -1 ) the bank storage is at a minimum due to the reversed hydraulic head towards the river. On the other hand, in No-salt management scenario only two major river bank recharges occur and these are responses to the two high flows in December 2006 and January (a) (b) 60,000 50,000 40,000 30,000 20,000 10, B1 B2 B3 B4 B5 B6 (c) Figure 3.6 a: Modelled groundwater salinity distribution (November 2007, time step 650 days), b: Conductivity distribution, EM31 survey in November 2007 (Berens et al., 2009), c: Recorded groundwater salinity during the study period (Holland et al., 2013). The main discharge processes are evapotranspiration, bank discharge and groundwater extraction via the SIS in Only-SIS scenario. In No-salt management scenario, water discharge via evapotranspiration is slightly higher than Only-SIS scenario. This is because of the relatively shallower groundwater table in No-salt management model. Generally, during the operation of the SIS production wells, the floodplain aquifer has a gaining regime while in No-salt management scenario 77

99 it is mostly losing. The discharge via the river bank occurs continuously in No-salt management scenario. This is due to the higher groundwater table in the floodplain aquifer except during the two river high flows. But in Only-SIS scenario, only two major groundwater discharges were observed. The first of these was prior to the commencement of the SIS operation when the river stage was lower than the groundwater table. The second was three months after the SIS was shut down. Both of the major groundwater discharges were diminished when the SIS commenced (July 2005 and May 2007). These are consistent with the observed and modelled groundwater head dynamics, as shown in Figure 3.5. It seems that the SIS operation lowers the groundwater table and enhances fresh river water recharge to the floodplain aquifer on one side and saline regional groundwater recharge on the other side of the production wells. (a) (b) Figure 3.7 Groundwater balance for the calibrated model (a) and No-SIS scenario (b) Figure 3.8 illustrates the solute mass balance for Only-SIS and No-salt management scenario. The results show that the stored solute mass in the floodplain aquifer is reduced during the study period of Only-SIS scenario by 4% (Figure 3.8a). Although without the SIS operation, the stored solute mass would have increased by 5% (Figure 3.8b). It seems that groundwater lowering via the SIS production wells may lead to a less saline floodplain aquifer. This happens through two mechanisms, namely the extraction of some portion of the saline groundwater (Figure 3.9) and the reversal of the hydraulic head towards the floodplain aquifer. The 5-15 ton.d -1 solute mass lowering from the wells is not substantial in comparison with the total stored solute mass in the floodplain 78

100 aquifer (around 60,000 tonnes in Figure 3.8). Therefore, the main mechanism is reversing the hydraulic head towards the floodplain aquifer. In fact, the SIS production wells create a divide which stops saline water from reaching the floodplain by lowering the groundwater table. In other words, groundwater lowering enhances the fresh water lens on one side and keeps the saline groundwater on the other side. But, in the absence of the SIS production wells, there is no process to prevent the saline groundwater from reaching the river and the only discharge process is via the river bank. Hence, the recharged solute mass from regional groundwater is stored in the floodplain aquifer or discharges to the river and leads to a more saline floodplain. Figure 3.10 shows the impact of the SIS production wells operating along transect 1. Regarding Figure 3.8a, two major solute mass discharges to the river bank are observed in June 2005 and March 2007 (up to 2 ton.d -1 ). Considering that approximately 800 ton.d -1 of river salt load was recorded at Lock 4 just upstream of the study site during the same periods (WaterConnect, 2013), the solute mass discharge via the river bank should not have a significant impact on the river water quality. This indicates that the dominant solute mass discharge process is saline groundwater extraction via the SIS production wells. In addition, in No-salt management scenario (Figure 3.8b), saline groundwater constantly discharges to the river at a higher rate (up to 4 ton.d -1 ). This would present a risk to river water quality over the long term. (a) (b) Figure 3.8 Solute mass balance for Only-SIS scenario (a) and No-salt management scenario (b) 79

101 Figure 3.9 Solute mass extracted via the SIS production wells during the study period Figure 3.10 Groundwater salinity along transect 1 for Only-SIS and No-salt management scenarios (Z magnification: 3) at time step 2070 days (2/09/2010) Water balance The river stage manipulation scenarios are now considered. One of the main starting points for analysis of the flow dynamics in the surface-groundwater system is the water balance. Hence, the outputs from the numerical model for 80

102 each round of river stage manipulation (hereafter referred to as trials) that are considered here include groundwater table dynamics, change in water storage in the floodplain aquifer (state of gaining or losing floodplain), flux exchange between the two domains (bank storage) and recharge from regional groundwater. The dynamics of the GW heads at the observation wells along transect 1 (wells B1, B2 and B3) are shown in Figure As expected, a rising river stage creates higher gradients from the river to the floodplain aquifer. Obviously, the GW dynamic is much more enhanced near the river bank rather than further away. This is noticeable in Figures 11a and 11b where observation well B4 shows a greater response to the river stage manipulations. Also, longer operation of the SIS wells has a greater effect on GW heads. This can be seen in Figure 3.11c whereby the same river stage rise scenario (here 1.5 m) with a longer trial duration shows a higher average groundwater hydraulic head in the floodplain aquifer at observation well B6. In other words, the extent of the floodplain aquifer response increases as the duration of SIS operations increases. Figure 3.12a shows the change in water storage in the floodplain aquifer during the study period. When this parameter is positive it represents a gaining floodplain while a negative value indicates that the floodplain has a losing regime. The change in water storage generally depends on the conductance (which does not vary over time) and the time-varying head gradient between the river and groundwater. According to Figure 3.9, the floodplain aquifer was approaching a gaining condition due to the river stage rise just before commencement of the SIS operation in July But, commencement of groundwater extraction via the SIS bores quickly formed a losing floodplain. This explains the formation of the pumping drawdown cone. The floodplain aquifer was in a losing condition until the first trial in November 2006, except for a short period in December 2005 corresponding to a high river flow. The response of the floodplain aquifer to each of the trials is evident in Figure 3.12a. As a general pattern, each rising scenario leads to a gaining floodplain during each trial and a losing floodplain after the trials. Clearly, the magnitude of the response to each trial is proportional to the height of the river stage rise. In 81

103 contrast, the 0.5 m drop scenario creates a losing floodplain during the trial and a gaining one after the trial. Due to the complexity of the study site, the floodplain response to each of the four trials is different. The first trial is coincident with the period during which the SIS was shut down. Hence, one of the main discharge components was absent. This created the most enhanced gaining floodplain condition (maximum value in Figure 3.12a). In fact, in this period the floodplain aquifer was recharged from the river and from regional groundwater. However, during the 2 nd trial a sudden decrease in the pumping rates of the SIS production wells occurred (from 5.5 l.s -1 to 2.2 l.s -1 ). This led to the lowest gaining condition among the four trials. The responses to the 3 rd and 4 th trials are the result of both the operation of the SIS production wells and river stage manipulation. For the 0.5 m drop scenario, the highest losing and gaining conditions happened during the 2 nd and 4 th trials, respectively. During the 2 nd trial the hydraulic head increased towards the river due to a decrease in the pumping rate of the SIS production wells. For the 4 th trial, the gaining condition was coincident with a river high flow. Moreover, the highest losing condition (minimum values in Figure 3.12a) occurred just before resumption of the SIS operation, but this was not due to river stage manipulation. This is partly attributed to the resumption of the SIS production wells and partly to the river stage decrease during that period. (a) (b) 82

104 (c) Figure 3.11 Dynamics of GW heads at the observation wells on Transect 2, a: 1.5 m rise for 3 months, b: 0.5 m drop for 3 months and c: 1.5 m rise for one, two and three months at observation well B6. (a) (b) 83

105 (c) Figure 3.12 a: Change in water storage in the floodplain aquifer, b: Flux exchange between the river and the floodplain aquifer and c: Floodplain aquifer recharge from regional groundwater. All the results shown here are for the three month scenarios. The blue and yellow patterns represent groundwater lowering and river stage manipulation, respectively. Figure 3.12b shows the flux exchange between the river and floodplain aquifer during the study period for the 3 month scenarios. This shows that the increase of bank storage is proportional to the rise in river stage. For example, as more water enters the floodplain aquifer, consequently more water returns back to the river during the river stage recession and this results in a greater flux for the 1.5 m scenario than for the other scenarios. This is due to the head difference that is formed with each river stage manipulation. After each trial, a minimum bank storage can be observed. It can also be seen that the volume of water that enters the aquifer from the river during the trial is greater than that which is subsequently discharged to the river after the trial. This is partly due to bank storage and also to a loss of water to evapotranspiration since each trial provides more water available for evapotranspiration. For example, as more water enters the floodplain aquifer, consequently more water returns back to the river during the river stage recession and this results in a greater flux for the 1.5 m scenario than for the other scenarios. 84

106 During the river drop scenario a different behaviour can be observed. In that period the floodplain is in a losing condition and the hydraulic gradient is changed towards the river. But, after each trial, there is a lag-time before the system reaches an equilibrium. During this lag-time, a significant amount of water moves from the river to the floodplain aquifer. The time required for the system to reach the equilibrium appears to be more closely related to the duration of the trial rather than to the river stage level changes. It seems the general pattern of flux exchange is similar for all four trials and the SIS operation does not have a significant influence on the pattern. Even, the slight increase in bank storage during the 1 st round of SIS operation (July November 2006) strongly corresponds to the river stage fluctuation. In other words, river stage manipulation has more influence than SIS operation on the groundwater head. Another important component of the water balance is floodplain aquifer recharge from regional groundwater, which is shown in Figure 3.12c. Recharge from regional groundwater is strongly attributed to the operation of the SIS production wells. In fact, recharge increases by three to four-fold during SIS operation compared to either before their operation (before July 2005) or when the SIS wells were shut down (from November 2006 to April 2007). However, a higher river stage rise may decrease the flux from the regional groundwater to the floodplain aquifer due to the enhanced hydraulic head towards the floodplain aquifer during the river stage rise trials. For example, the 0.5 m drop scenario leads to a slight increase in regional groundwater recharge Solute mass balance In order to analyse the spatial and temporal solute dynamics of the floodplain aquifer in the context of SW-GW interaction, the following modelled outputs were considered: (a) total stored solute mass in the floodplain aquifer; (b) change in total stored solute mass in the floodplain aquifer; (c) stored solute mass in the unsaturated zone; and (d) solute concentrations at observation wells B1 and B4. The total stored solute mass in the floodplain aquifer is shown in Figure 3.13a. Decreases in stored solute mass in the floodplain aquifer are mainly attributed to 85

107 saline groundwater extraction by the SIS production wells. This is why the total stored solute mass generally decreases for all scenarios. However, it may be seen that a higher river stage results in relatively less solute being accumulated in the floodplain aquifer. Since the main source of solute mass entering the floodplain aquifer is regional saline groundwater, an increase in river stage can reduce the rate of saline groundwater entering the floodplain aquifer. On the other hand, the 0.5 m drop scenario results in relatively more solute accumulation. This is due to the increased recharge of the floodplain aquifer from saline regional groundwater, which is consistent with the results shown in Figure 3.12c. The results indicate that for a higher stage rise, comparatively less solute mass accumulates in the floodplain aquifer. This is because of the increase in head gradient from the river towards the floodplain aquifer. Figure 3.13.b illustrates the change in stored solute mass in the floodplain aquifer over the study period. As a general pattern, when the SIS operates, the change in stored solute mass is negative (more solute leaves the system than enters). Higher river stage rises may enhance this process by increasing the hydraulic gradient towards the floodplain aquifer. In contrast, when the floodplain aquifer is losing, it means that more saline groundwater from the regional aquifer enters the floodplain. This can be seen in the 0.5 m drop scenario. Changes in solute concentration along transect 1 due to changes in river stage for the 0.5 m drop and +1.5 m rise three month scenarios during the 3 rd trial are presented in Figure Clearly, a 0.5 m drop scenario produces a more saline aquifer, while the 1.5 m rise would produce the least saline aquifer. Also, the extent of the freshwater lens is longer in the 1.5 m rise scenario. Considering Figures 11 and 14, it appears that the river stage manipulation operation is more effective in the vicinity of the river bank. This can be beneficial in arid and semiarid floodplains where riparian vegetation health depends on the availability of freshwater, as is the case in the study area. It is also expected that larger increases in river stage will result in (a) dilution propagating further inland and (b) dilution being more pronounced at the river-aquifer interface. 86

108 (a) (b) Figure 3.13 a: Total solute mass in the floodplain aquifer and b: Change in stored solute mass in the floodplain aquifer. Both are for the three month scenarios. The blue and yellow patterns represent groundwater lowering and river stage manipulation, respectively. 87

109 Figure 3.14 Spatial distribution of modelled solute concentration along transect 1 during the 3 rd trial (time step 1480 (20/01/2009)) for the 1.5 m rise and 0.5 m drop three month scenarios. Observation wells are shown in black Solute mass in the unsaturated zone It was found that at the beginning of the study period, 13% of the total solute mass was stored in the unsaturated zone. Two main drivers are influencing the solute dynamic in the floodplain aquifer including groundwater lowering and river stage manipulation. Here, the impact of each of these drivers on the accumulated or mobilized solute mass in the unsaturated zone is analysed. Figure 3.15 compares the solute mass in the unsaturated zone in No-salt management and Only-SIS along scenarios along transect 2. As expected, a significant amount of solute mass is mobilized due to the operation of the SIS production wells. This may be because the SIS operation lowers the groundwater table which leads to an overall less saline unsaturated zone. Moreover, Only-SIS scenario shows a much larger freshwater lens compared to No-salt management scenario. Again, it seems that groundwater lowering is able to maintain a less saline floodplain aquifer by mitigating regional groundwater recharge on one hand and by attracting more freshwater via the river bank on the other hand. 88

110 Furthermore, the results confirm that Only-SIS scenario at the last time-step (2070) shows a mobilization of 6% of the solute mass from the unsaturated zone. On the other hand, No-salt management scenario shows 5% more solute mass accumulation in the unsaturated zone compared to the model in the 1 st time step. (a) (b) Figure 3.15 Visualization of the solute concentration distribution in the floodplain aquifer for Only-SIS (a) and No-salt management scenarios (b) along transect 2 89

111 Figure 3.16 shows the spatial distribution of the solute concentration in the floodplain aquifer for the three month long 1.5 m rise and 0.5 m drop scenarios at time-step 1120 days (just after the 2 nd trial) along transect 2. Comparing with Figure 3.15, it appears that river stage manipulation has less influence than groundwater lowering floodplain salinity, with the influence being limited to the near-river zone. The higher hydraulic gradient from the river to the floodplain creates a relatively larger freshwater lens. The results indicate that the total solute mass in the unsaturated zone under the 1.5 m rise scenario is 7% less than for the 0.5 m drop scenario after the 2 nd trial. In fact, the 0.5 m drop scenario leads to more saline regional groundwater recharges to the floodplain (see Figure 3.13b). (a) 90

112 (b) Figure 3.16 Visualization of the solute concentration distribution in the floodplain aquifer for the three month long +1.5 m (a) and -0.5 m (b) scenarios at time-step 1120 days (just after the 2 nd trial) along transect 2. In order to demonstrate the impact of river stage manipulation on the solute mass accumulation in the unsaturated zone, time-steps 0 (beginning of the study period) and 1120 days (after the 2 nd trial) in the three month 1.5 m rise scenario are analysed. Figure 3.17 shows the spatial distribution of solute mass mobilization attributed to river stage manipulation (three month, +1.5 m scenario) at time-step 1120 days. The model results show that up to 14 kg.m -3 of solute mass is removed from the unsaturated zone. It appears that the greatest solute mass mobilization occurs at a distance of around m from the river. In fact, the extent that river stage manipulation may significantly affect the solute mobilization is limited to the extent that it could change the head gradient, which in this case is the boundary of the fresh and saline zones. The solute mass mobilization in the rest of the floodplain aquifer is restricted. This is because of the relatively lower amount of solute stored at the river bank (0-50 m). On the other hand, as the main source 91

113 of solute comes from the highland aquifer, less change in solute in the unsaturated zone can be expected in the highland. Figure 3.17 Visualization of distribution of solute mass mobilization in the floodplain aquifer for the three month long, +1.5 m scenario at time-step 1120 days (just after the 2 nd trial) along transect Ecological implications Knowledge of the interaction between groundwater and surface water bodies is vital for assessing the role of riparian floodplain processes on water quality and groundwater level dynamics (Eslamian and Nekoueineghad, 2009; Rassam, 2005). The ecological implications of river stage manipulation were not the foremost objective of this research. However, it can be inferred that each river stage rise trial may lead to soil water freshening and this may lead to a riparian tree response since each trial makes more fresh water accessible for riparian trees. This is reinforced by the results of Berens et al. (2009), Holland et al. (2009a) and Holland et al. (2013) who showed that introduction of sufficient fresh water through sources such as artificial inundation and bank storage to a saline floodplain aquifer can to some extent maintain species richness and diversity on the floodplain. For instance, the availability of relatively more fresh water (through a freshened soil profile) may lead to epicormic growth in trees. However, the ecological response mainly depends on the vegetation condition at the beginning of the trial (Holland et al., 2013). Therefore, due to the overall saline 92

114 nature of arid and semi-arid floodplains, this response is unlikely to be sustained unless there is a regular recurrence of the trial cycles. 3.4 Conclusion A fully integrated, physically-based, numerical model (HydroGeoSphere) of surface water-groundwater flow and solute transport at Clark's Floodplain was developed and calibrated against observed data, which included river stages, floodplain aquifer heads and solute concentrations. The calibration results showed that the model was capable of reproducing the dynamics of both the flow and the solute. The calibrated model was applied to investigate the relative impacts of river stage manipulation on surface water-groundwater interactions that were also being influenced by Salt Interception Scheme (SIS) groundwater lowering measures. Twelve hypothetical scenarios were defined including river stage changes of +0.5, +1.0, +1.5, and -0.5 m. Using the calibrated model (Only-SIS scenario), it was shown that groundwater lowering via the SIS production wells was able to mitigate the saline regional groundwater recharge on the one hand and enhance river bank storage on the other hand, thereby leading to a less saline floodplain aquifer. In fact, by lowering the groundwater table, the SIS production wells created a divide which stopped saline water from reaching the floodplain. In this situation, even if groundwater discharge to the river via the river bank did occur, it would not have a significant impact on the river water quality. However, without the groundwater lowering operation, there would be no process to prevent the saline groundwater from reaching the river and the only discharge process would be via the river banks. It was demonstrated that an absence of groundwater lowering measures would have led to continuous saline groundwater discharge to the river which could be a risk to the river water quality over the long term. In other words, the SIS was successful in intercepting saline groundwater that would otherwise have entered the river. The hypothetical scenarios were used to demonstrate the impact of river stage manipulation on the SW-GW interaction in the river-floodplain system. In terms of water balance, it was shown that each rising river manipulation scenario led to a gaining floodplain which was proportional to the height and duration of each 93

115 trial. The floodplain aquifer became a losing feature after each rising trial for a duration almost equal to the duration of the trial. This pattern was reversed for the 0.5 m drop scenario. It was shown that river stage rise formed higher gradients from the river to the floodplain aquifer and this was much more pronounced in the near-river zone (up to 100 m from the river). However, longer durations of river stage manipulation could potentially extend the influence further into the floodplain. Furthermore, it was demonstrated that a higher river stage rise was able to reduce the flux from the regional groundwater to the floodplain aquifer due to the enhanced hydraulic head towards the floodplain aquifer. Consequently, a higher river stage led to relatively less solute mass in the floodplain aquifer. In contrast, scenarios that involve lowering the river stage enhanced the flux from the saline regional groundwater to the floodplain aquifer. This resulted in a more saline floodplain aquifer. Among the dominant drivers in this study, groundwater extraction had a greater influence on solute mass mobilization in the unsaturated zone. However, river stage rise can also be effective but is limited to the near-river zone (up to 100 m). Moreover, the results of this study indicate that river stage manipulation may be able to decrease the soil and groundwater salinities in the near-river zone which can improve the health of riparian vegetation. However, its impacts are spatially and temporally limited and river stage manipulation cannot entirely change the natural condition of the floodplain. Hence, it may be considered as a short term management technique. However, if longer term strategies are required, it may be possible to implement these salt interception measures periodically. According to the results of this study, it can be concluded that bank storage is one of the main drivers of surface-groundwater interactions in a saline, semi-arid floodplain, particularly in non-flooding conditions. Also, induced bank storage through river level manipulation may lead to a less saline floodplain aquifer. This happens through two mechanisms. First, an increase in river stage lowers the gradient from the regional groundwater aquifer to the floodplain which reduces saline groundwater recharge to the floodplain. Second, modelling supports observational data that bank storage is able to freshen the groundwater near the 94

116 river banks during high-flow pulses by mixing fresh water with saline groundwater. The applicability of these findings to other areas depends on floodplain topography, soil salinity, groundwater condition (i.e. gaining or losing), geomorphology of the floodplain aquifer (i.e. hydraulic conductivity), river characteristics (i.e. bed conductance, bank slope, depth) and vegetation condition. Further studies are recommended to investigate other potential drivers and salt interception measures, such as artificial floodplain inundation. 95

117 References Alaghmand, S., Beecham, S., Hassanli, A., 2013a. Impacts of Groundwater Extraction on Salinization Risk in a Semi-Arid Floodplain. Nat. Hazards Earth Syst. Sci. 13(12) Alaghmand, S., Beecham, S., Hassanli, A., 2013b. A review of the numerical modelling of salt mobilization from groundwater-surface water interactions. Water Resources 40(3) Allison, G.B., Cook, P.G., Barnett, S.R., Walker, G.R., Jolly, I.D., Hughes, M.W., Land clearance and river salinization in the western Murray Basin, Australia. Journal of Hydrology AquaVeo, GMS: Provo, UT. AWE, Clarks Floodplain Investigations. Prepared for the Loxton to Bookpurnong Local Action Planning Committee. Australian Water Environments Report Adelaide. AWE, Loxton Bookpurnong SIS Atlas. Australian Water Environment: Adelaide. Banks, E.W., Brunner, P., Simmons, C.T., Vegetation controls on variably saturated processes between surface water and groundwater and their impact on the state of connection. Water Resources Research 47(11). Barnett, B., Townley, L.R., Post, V., Evans, R.E., Hunt, R.J., Peeters, L., Richardson, S., Werner, A.D., Knapton, A., Boronkay, A., Australian groundwater modelling guidelines, In: report, W. (Ed.). National Water Commission: Canberra. Barnett, S.R., Yan, Y., Watkins, N.R., Woods, J.A., Hyde, K.M., Murray Darling Basin Salinity Audit: Groundwater Modelling to Predict Future Salt Loads to the River Murray in South Australia, Report DWR 2001/017. Department for Water Resources Adelaide. Bauer, P., Gumbricht, T., Kinzelbach, W., 2006a. A regional coupled surface water/groundwater model of the Okavango Delta, Botswana. Water Resources Research 42(4) W Bauer, P., Held, R.J., Zimmermann, S., Linn, F., Kinzelbach, W., 2006b. Coupled flow and salinity transport modelling in semi-arid environments: The Shashe River Valley, Botswana. Journal of Hydrology 316(1 4) Bennett, N.D., Croke, B.F.W., Guariso, G., Guillaume, J.H.A., Hamilton, S.H., Jakeman, A.J., Marsili-Libelli, S., Newham, L.T.H., Norton, J.P., Perrin, C., Pierce, S.A., Robson, B., Seppelt, R., Voinov, A.A., Fath, B.D., Andreassian, V., Characterising performance of environmental models. Environmental Modelling and Software

118 Berens, V., White, M., Souter, N., Bookpurnong Living Murray Pilot Project: A trial of three floodplain water management techniques to improve vegetation condition. Department of Water, Land and Biodiversity Conservation: Adelaide. BOM, Bureau of Meteorology (BOM). Brooks, R.J., Corey, A.T., Hydraulic Properties of Porous Media, Hydrology Paper No. 3. Fort Collins, Colorado State University. Brunner, P., Simmons, C.T., HydroGeoSphere: A Fully Integrated, Physically Based Hydrological Model. Ground Water 50(2) Carsel, R.F., Parrish, R.S., Developing joint probability distributions of soil water retention characteristics. Water Resources Research 24(5) Crowe, A.S., Shikaze, S.G., Ptacek, C.J., Numerical modelling of groundwater flow and contaminant transport to Point Pelee marsh, Ontario, Canada. Hydrological Processes 18(2) Doble, R., Brunner, P., McCallum, J., Cook, P.G., An analysis of river bank slope and unsaturated flow effects on bank storage. Ground Water 50(1) Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2) Ehret, U., Zehe, E., Series distance - An intuitive metric to quantify hydrograph similarity in terms of occurrence, amplitude and timing of hydrological events. Hydrology and Earth System Sciences 15(3) Eslamian, S., Nekoueineghad, B., A review on interaction of groundwater and surface water. International Journal of Water 5(2) Evans, R., Hoxley, G., Collett, K., Mallee Salinity Workshop: Chapter 2 - Floodplain processes. Mallee Catchment Management Authority. Ewen, J., Hydrograph matching method for measuring model performance. Journal of Hydrology 408(1-2) Ghazavi, R., Vali, A.B., Eslamian, S., Impact of Flood Spreading on Groundwater Level Variation and Groundwater Quality in an Arid Environment. Water Resources Management 26(6) Gupta, H.V., Clark, M.P., Vrugt, J.A., Abramowitz, G., Ye, M., Towards a comprehensive assessment of model structural adequacy. Water Resources Research 48(8). Hart, B., Bailey, P., Edwards, R., Hortle, K., James, K., McMahon, A., Meredith, C., Swadling, K., A review of the salt sensitivity of the Australian freshwater biota. Hydrobiologia 210(1) Herczeg, A.L., Simpson, H.J., Mazor, E., Transport of soluble salts in a large semiarid basin: River Murray, Australia. Journal of Hydrology

119 Hingston, F.J., Galbraith, J.H., Dimmock, G.M., Application of the processbased model BIOMASS to Eucalyptus globules subsp. Globules plantations on ex-farmland in south Western Australia: I. Water use by trees and assessing risk of losses due to drought. Forest Ecology and Management Hoehn, E., Scholtis, A., Exchange between a river and groundwater, assessed with hydrochemical data. Hydrology and Earth System Sciences 15(3) Holland, K.L., Charles, A.H., Jolly, I.D., Overton, I.C., Gehrig, S., Simmons, C.T., 2009a. Effectiveness of artificial watering of a semi-arid saline wetland for managing riparian vegetation health. Hydrological Processes Holland, K.L., Doody, T.M., McEwan, K.L., Jolly, I.D., White, M., Berens, V., Souter, N.J., 2009b. Response of the River Murray floodplain to flooding and groundwater management: Field investigations, Water for a Healthy Country National Research Flagship. CSIRO: Adelaide, p. 65. Holland, K.L., Jolly, I.D., Overton, I.C., Walker, G.R., 2009c. Analytical model of salinity risk from groundwater discharge in semi-arid, lowland floodplains. Hydrological Processes Holland, K.L., Turnadge, C.J., Nicol, J.M., Gehrig, S.L., Strawbridge, A.D., Floodplain response and recovery: comparison between natural and artificial floods, Technical Report Series No. 13/4. Goyder Institute for Water Research: Adelaide. Jakeman, A.J., Letcher, R.A., Norton, J.P., Ten iterative steps in development and evaluation of environmental models. Environmental Modelling and Software 21(5) Jolly, I.D., Review of River Murray Floodplain Salinity Studies in South Australia and Their Relevance to the Victorian Mallee Floodplains. CSIRO Land and Water: Adelaide. Jolly, I.D., McEwan, K.L., Holland, K.L., A review of groundwater-surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1(1) Jolly, I.D., Walker, G.R., Hollingworth, I.D., Eldridge, S.R., Thorburn, P.J., McEwan, K.L., Hatton, T.J., The causes of decline in eucalypt communities and possible ameliorative approaches, In: walker, G.R., Jolly, I.D., Jarwal, S.D. (Eds.), Salt and Water Movement in the Chowilla Floodplain. CSIRO Division of Water Resources: Canberra, Australia. Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4) Knight, J.H., Rassam, D.W., Groundwater head responses due to random stream stage fluctuations using basis splines. Water Resources Research 43(6). Kollet, S.J., Maxwell, R.M., Woodward, C.S., Smith, S., Vanderborght, J., Vereecken, H., Simmer, C., Proof of concept of regional scale hydrologic 98

120 simulations at hydrologic resolution utilizing massively parallel computer resources. Water Resources Research 46(4). Langevin, C., Swain, E., Wolfert, M., Simulation of integrated surfacewater/ground-water flow and salinity for a coastal wetland and adjacent estuary. Journal of Hydrology 314(1 4) Li, H.T., Brunner, P., Kinzelbach, W., Li, W.P., Dong, X.G., Calibration of a groundwater model using pattern information from remote sensing data. Journal of Hydrology 377(1 2) Matthews, K.B., Rivington, M., Blackstock, K., McCrum, G., Buchan, K., Miller, D.G., Raising the bar? - The challenges of evaluating the outcomes of environmental modelling and software. Environmental Modelling and Software 26(3) McCallum, J.L., Cook, P.G., Brunner, P., Berhane, D., Solute dynamics during bank storage flows and implications for chemical base flow separation. Water Resources Research 46(7). McEwan, K.L., Jolly, I.D., Holland, K.L., Groundwater-Surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology, Science Report 53/06. CSIRO Land and Water Adelaide. McLaren, R.G., Grid Builder: A pre-processor for 2-D, triangular element, finite-element programs. Groundwater Simulations Group, University of Waterloo: Waterloo, Ontario. Meire, D., De Doncker, L., Declercq, F., Buis, K., Troch, P., Verhoeven, R., Modelling river-floodplain interaction during flood propagation. Natural Hazards 55(1) Nash, J.E., Sutcliffe, J.V., River flow forecasting through conceptual models part I - A discussion of principles. Journal of Hydrology 10(3) Peck, A.J., Hatton, T., Salinity and the discharge of salts from catchments in Australia. Journal of Hydrology Peck, A.J., Hurle, D.H., Chloride balance of some farmed and forested catchments in Southwestern Australia. Water Resources Research Peyrard, D., Sauvage, S., Vervier, P., Sanchez-Perez, J.M., Quintard, M., A coupled vertically integrated model to describe lateral exchanges between surface and subsurface in large alluvial floodplains with a fully penetrating river. Hydrological Processes 22(21) Rassam, D.W., Impacts of hillslope floodplain characteristics on groundwater dynamics: implications for riparian denitrification, The International Congress on Modelling and Simulation: Melbourne, Australia, pp Rassam, D.W., A conceptual framework for incorporating surface groundwater interactions into a river operation planning model. Environmental Modelling & Software 26(12)

121 Rassam, D.W., Peeters, L., Pickett, T., Jolly, I., Holz, L., Accounting for surface-groundwater interactions and their uncertainty in river and groundwater models: A case study in the Namoi River, Australia. Environmental Modelling and Software Restrepo, J.I., Montoya, A.M., Obeysekera, J., A Wetland Simulation Module for the MODFLOW Ground Water Model. Ground Water 36(5) Schmid, W., Hanson, R.T., III, T.M.M., Leake., S.A., User s guide for the Farm process (FMP) for the U.S. Geological Survey s modular three-dimensional finitedifference ground-water flow model, MODFLOW-2000, USGS Techniques and Methods 6-A17. USGS: Reston, Virginia. Sophocleous, M., Review: Groundwater management practices, challenges, and innovations in the High Plains aquifer, USA-lessons and recommended actions. Revue critique: Pratiques, défis et innovations dans le domaine des de la gestion des eaux souterraines de l'aquifère des Grandes Plaines (High Plains), aux Etats Unis d'amérique - Leçons et recommandations 18(3) Straatsma, M.W., van der Perk, M., Schipper, A.M., de Nooij, R.J.W., Leuven, R.S.E.W., Huthoff, F., Middelkoop, H., Uncertainty in hydromorphological and ecological modelling of lowland river floodplains resulting from land cover classification errors. Environmental Modelling and Software Telfer, A., Overton, I.C.,, The impact of irrigation on floodplain vegetation health adjacent the River Murray, In: Rutherford, I.D., Bartley, R. (Ed.), The Second Australian Stream Management Conference: Adelaide. Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-Dimensional Numerical Model Describing Fully- Integrated Subsurface and Surface Flow and Solute Transport. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. van Genuchten, M.T., A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Science Society of America Journal 44(5) Verstrepen, L., Evaluating rainwater harvesting on watershed level in the semi-arid zone of Chile, Bioscience Engineering. Universiteit Gent: Gent, p Walton, R., Chapman, R., Davis, J., Development and application of the wetlands Dynamic Water Budget Model. Wetlands 16(3) WaterConnect, River Murray Water Data Zimmermann, S., Bauer, P., Held, R., Kinzelbach, W., Walther, J.H., Salt transport on islands in the Okavango Delta: Numerical investigations. Advances in Water Resources 29(1)

122 4 Quantifying the impacts of artificial flooding on a river-floodplain interaction Overview Chapter 4 describes the application of artificial flooding on a saline semi-arid floodplain and the potential contributions to floodplain salinity and vegetation health. In this case, a total of four scenarios are defined to study the extent to which artificial flooding can affect floodplain salinity and to evaluate the ecological implications. A fully integrated, physically-based model is first calibrated against observed data obtained from two artificial flooding field trials conducted at the study site (Site A in Clark s floodplain). The model results show that the effects of artificial flooding are restricted to the inundated zone; therefore, the impacts are demonstrated in the inundated zone soil profile. It appears that during each artificial flooding event, groundwater is discharged to the river but its impact on river water quality is negligible. It may also be seen that artificial flooding is able to reduce root zone salinity beneath the flooded area and this can lead to improved vegetation health. However, these influences are generally limited spatially to the inundated zone and cannot permanently change the saline nature of the system. This is consistent with the EM31 survey results which are available from the artificial flooding field trials conducted at the study site. It is recommended that artificial flooding should be considered as an intervention technique when vegetation starts to decline. Hence, artificial flooding may be considered as a short term management technique and which may need to be applied periodically if it is to be used as an effective long term strategy. 101

123 This chapter is currently in review in the journal Environmental Modelling and Software (Paper 3). The manuscript was co-authored by my Principal Supervisor, Prof. Simon Beecham, and advisor, Dr. Ali Hassanli. Also, Mr. Ian Jolly and Dr. Kate Holland (research scientist at CSIRO) and Dr. Juliette Woods (research fellow at Flinders University and DEWNR) contributed to this paper. The format of the paper has been changed to be consistent with the rest of this thesis. 102

124 Paper 3: Quantifying the impacts of artificial flooding and groundwater lowering on a river-floodplain interaction in a semi-arid saline floodplain Published as: Alaghmand, S., Beecham, S., Woods, J.A., Jolly, I.D., Holland K.L. and Hassanli, A., Quantifying the impacts of artificial flooding and groundwater lowering on a river-floodplain interaction in a semi-arid saline floodplain, Environmental Modelling and Software (in review). Abstract: The natural flow regime of the Lower Murray River in South Australia has been altered due to abstraction, river regulation and climate change which resulted in severe decline in the condition of long-lived vegetation on the floodplain. This paper aims to quantify the relative impacts of artificial flooding on the flow and solute dynamics of the floodplain aquifer and its ecological implications. It is shown that artificial flooding can temporarily form less saline groundwater and soil profiles which in turn can improve water availability for vegetation. From an ecological point of view, artificial flooding delivers some of the same benefits as natural overbank floods. It appears that artificial flooding is an intervention technique that is limited spatially to the flooded area and temporally to the flood duration. Hence, artificial flooding may be considered as a short term management technique in arid and semi-arid saline floodplain. 4.1 Introduction Nearly all rivers in the world are to some extent influenced by human activities. River systems usually include floodplains and therefore an important component of the system is the interaction of surface water (SW) and groundwater (GW). Analysis of the interactions can range from pure hydraulic studies to assessment of water balances and ecological impacts. Even from a salinity perspective, floodplains often play a significant role. In many cases, particularly in arid and semi-arid areas, floodplain aquifers may act as a source of solute or as a pathway for solute mobilization towards the river (Alaghmand et al., 2013a). It is wellunderstood that there is a dynamic interaction between these two domains 103

125 hydraulically, ecologically and hydrologically (Karim et al., 2012; Motta et al., 2012; Russo et al., 2012; Zilli and Paggi, 2013). For example, river manipulation by constructing a weir or dam significantly affects areas both upstream and downstream of the structure. Upstream it can lead to consequences such as waterlogging of riparian trees due to elevated water-tables (Jolly et al., 2008) and permanent inundation of the floodplain which is detrimental for seed and egg bank germination (Brock et al., 2003). Downstream, it can also have impacts such as reduction of environmental flows (Eng et al., 2013; Tagliaferro et al., 2013), loss of wet-dry cycles leading to long lasting ecological effects (Baldwin and Mitchell, 2000), reduced groundwater replenishment and diminishing hydraulic connectivity between the river and the floodplain (less fresh water in the floodplain aquifer via bank recharge) that can cause floodplain salinization (Girard et al., 2003; Jolly et al., 1998; Lamontagne et al., 2005). Moreover, human agricultural activities on the floodplain and adjacent highland, such as excess drainage and groundwater extraction, may cause soil salinity issues, increased groundwater levels and groundwater quality declines that can influence the river flow quality and quantity (Brunner et al., 2007; Brunner et al., 2008; Nouri et al., 2013). However, many of these activities are unavoidable due to human population growth and the limited nature of available natural resources. So, it is essential to have a comprehensive and realistic understanding of the processes involved and the consequences of such scenarios. Numerical models are one of the most popular tools that can be used to explore the relative importance of each SW-GW interaction process (Rassam et al., 2013). For many years in modelling practice, the surface and subsurface domains were considered separately which is not a realistic assumption particularly in areas with dynamic and hydraulically well-connected SW and GW systems (Halford and Mayer, 2000; Rassam and Werner, 2008; Schmid et al., 2006; Swain and Wexler, 1996). In surface models, groundwater is often considered simply as a loss term depending on direction. In groundwater models, surface waters such as rivers and lakes are often modelled as a boundary condition (Rassam and Werner, 2008). The relatively new concept of fully-integrated numerical models was established by Brown (1995). Fully-integrated models solve both the surface and sub-surface domains simultaneously using a single matrix of equations. These equations are 104

126 derived from the principles of conservation of mass and momentum of water and/or solute. Sophocleous and Perkins (2000) highlighted that one of the advantages of fully-integrated modelling over either watershed or GW models is improving model calibration by increasing the calibration targets, which reduces parameter non-uniqueness. Jolly et al. (2008) stated that there is a clear need to develop modelling capabilities for the movement of salt to, from, and within wetlands to provide temporal predictions of wetland salinity which can be used to assess ecosystem outcomes. Moreover, Alaghmand et al. (2013b) concluded that a 3D fully-integrated physically-based surface-subsurface numerical model with the capabilities of saturated-unsaturated modelling and solute transport simulation is required for detailed modelling of river and floodplain interactions. High quality observed data are also an essential part of these studies to establish a reliable, validated model. The Lower Murray River is a typical example of a semi-arid area where changes to floodplain hydrology (river flow manipulation and irrigation practices in the adjacent highlands) have caused a decline in the condition of the dominant riparian tree species (Eucalyptus largiflorens, black box; E. camaldulensis, river red gum; and Acacia stenophylla, river cooba) due to groundwater and soil salinization (Maheshwari et al., 1995; Slavich et al., 1999). More than 75% of all trees along the Lower Murray River floodplain in South Australia were classified as unhealthy in (Smith and Kenny, 2005). The extent, duration and timing of floodplain inundation all have significant impacts on floodplain biota. These floodplain inundation effects have been influenced by regulation of the Murray River as well as diversions for consumptive use (Bunn and Arthington, 2002; George et al., 2005; Walker and Thoms, 1993). Overbank flooding is a key hydrologic process affecting floodplain aquifer flow and solute dynamics and ecological processes such as biogeochemical cycling and plant diversity (Naiman and Décamps, 1997). Overbank flooding historically occurred for a few days to weeks every few years for most natural rivers (Wolman and Leopold, 1957). Groundwater recharge to the floodplain aquifer can be even larger during overbank flooding as it provides another source of recharge to the floodplain aquifer in addition to rainfall. But in the Lower Murray River a system of major reservoirs and small locks and weirs were installed along the river during 105

127 the 1920s to regulate river flow and provide a navigable passage. This led to the raising of groundwater levels beneath the adjacent floodplains together with a reduction of frequency of medium to large floods by a factor of three (Alaghmand et al., 2014a; Ohlmeyer, 1991). This was combined with an absence of overbank flooding between 1994 and 2010 during the Millennium Drought (Leblanc et al., 2012; Timbal et al., 2010). One of the consequences was an increase in soil and groundwater salinity in the floodplains that has reduced water availability for riparian vegetation due to increased osmotic potentials (Busch and Smith, 1995; Jolly et al., 1993; Munns, 1993). Figure 4.1 shows the conceptual model of the river-floodplain system influenced by human-induced activities at the Lower Murray River. The main objective of this study is to quantify the interactions between a river (Lower Murray River) and a semi-arid saline floodplain (Clark s Floodplain), which is significantly influenced by artificial flooding and groundwater lowering. The research question is what are the combined impacts of artificial flooding and groundwater lowering on the floodplain aquifer flow and solute dynamics and what are the ecological implications. The individual impacts of each driver are investigated using a fully-integrated physically-based flow and solute numerical modelling approach. The impacts are discussed in terms of water and solute balances. Also, the ecological implications of artificial flooding are examined to see if vegetation health condition improves and is sustained after the flooding events. 4.2 Methods The numerical model employed here is based on a conceptual model of a river and a saline floodplain system at Clark s Floodplain located on the Lower Murray River in South Australia. Prior to European settlement there was a natural balance between the river and the saline regional groundwater. But after river flow manipulation by installation of weirs along the river, and irrigation practices in the adjacent highlands, the water table level in the floodplain has been increased. The raised groundwater level beneath the floodplain has led to increased rates of groundwater evapotranspiration. Because the groundwater is naturally saline, the increased ET results in floodplain salinization, which consequently affects the 106

128 health of floodplain vegetation (Alaghmand et al., 2014a). Therefore, artificial flooding has been examined as a salt management measure to mobilize some of the accumulated solute from the unsaturated zone and to increase water availability for the stressed vegetation. Figure 4.1 Conceptual model of the river-floodplain system influenced by humaninduced activities (Adopted from Alaghmand et al. (2014a)) Study site Clark s Floodplain is located on the Lower Murray River in South Australia (34 21'S, 'E), downstream of Lock and Weir No. 4 (Figure 4.2). The climate in the Lower Murray River is semi-arid with mild winters and long hot 107

129 summers. Annual potential pan evaporation ( mm) is over seven times the average annual rainfall ( mm) (BOM, 2013). Annual rainfall is highly variable, with Bureau of Meteorology records showing annual rainfall ranging between 87 and 556 mm since 1963 (Holland et al., 2009). The floodplain is located next to the Bookpurnong Irrigation District which was developed in Irrigation has created a localised groundwater mound in the regional Loxton-Parilla Sands aquifer, increasing the hydraulic gradient towards the floodplain and causing an increase in groundwater flux to the river and seepage of saline groundwater at the edge of the river valley (Holland et al., 2013). In order to maintain the River Murray water quality, a series of production wells along the floodplain were constructed in These wells are known as the Bookpurnong floodplain salt interception schemes (SIS) which aim to reduce the hydraulic gradient that drives the regional saline groundwater towards the River Murray by maintaining the water-table between the SIS bores at river level. The SIS bores (30F and 32F) were operational between August 2005 and November 2006 and between May 2007 and a flood that occurred in 2010/11 (Holland et al., 2009). Two of the SIS production wells are located in the study site, namely 30F and 32F, which are shown in Figure 4.2. The Lower Murray River floodplain in South Australia is typically vegetated by a mixture of Eucalyptus camaldulensis (red gum), Eucalytpus largiflorens (black box), Acacia stenophylla and Muehlenbeckia florulenta (lignum) (O'Malley and Sheldon, 1990). The existence of river red gums shows that the floodplain depression was naturally flooded every 3 to 5 years before regulated flow regimes were instituted (Holland et al., 2009). Soils in the study area generally consist of micaceous cracking clay deposits, known as the Coonambidgal Clay. This surficial clay layer can be up to 5 m thick, being thickest around relict and existing wetlands, with typically low hydraulic conductivity values between 0.05 and 0.1 m/day (Doble et al., 2006). The Coonambidgal Clay overlies the Monoman Formation, an aquifer unit composed of unconsolidated sand deposits with variable clay and silt content that is hydraulically connected to surface water bodies on the floodplain (Jarwal, 1996). The Monoman Formation can be up to 30 m thick in the study area with a hydraulic conductivity of m/day (Doble et al., 2006). Within the River Murray valley, the Monoman Formation aquifer is in 108

130 direct contact with the regional Loxton-Parilla Sands. Groundwater salinities in the study area can be divided into two zones, namely a relatively fresh water zone (<5,000 µs/cm) in parts of the floodplain aquifer nearest to the river and another zone where salinity levels are in excess of sea water (up to 50,000 µs/cm) in areas away from the river bank (Holland et al., 2009). In fact, the regional aquifer acts as the main source of solute flowing towards the floodplain and eventually the river. Figure 4.2 Configuration of the study site. Observation wells and SIS production wells are shown in red and blue circles, respectively. The yellow line indicates the floodplain depression. The green diamonds represent river salinity loggers. Inset map shows the location of the study site in Australia. A total of 12 observation wells were constructed at the study site in June 2005 to record the groundwater level and salinity using pumping/bailing methods (Figure 4.2). Also, soil samples were collected in 0.5 m increments from the unsaturated zone from all twelve wells on a bi-annual cycle until March Furthermore, a Geonics EM31 conductivity meter survey was conducted at the study site on six occasions; June 2005, February 2006, September 2006, February 2007, September 2007 and March 2008 (Holland et al., 2009). This EM technique uses the induction of an electromagnetic field to sense the soils ability to conduct or resist electrical current. Moreover, recorded meteorological data such as rainfall and 109

131 potential evaporation were obtained from the Loxton Research Centre meteorological station (ID: ) (BOM, 2013). The Murray River recorded flows were obtained from the water level station downstream of Lock 4, which is situated immediately upstream of Clark s Floodplain (ID: A ) (WaterConnect, 2013). Likewise, River salinity observations were obtained from the Clarke s Sandbar (ID: A ) and Rilli Island (ID: A ) stations located upstream and downstream of the study site, respectively (Figure 4.2) Field data This paper focuses on two rounds of artificial flooding (hereafter referred to as trials) that were conducted by White et al. (2009) at a depression located at the northern end of Clark s Floodplain near the river. Hence, all the field data were obtained from White et al. (2009) in order to build the study site model. An 80 m earthen bank was constructed at the southern end of the floodplain depression where a flood-runner normally fills it from the river (Figure 4.3). The construction of the earthworks formed an inundation zone that covered 3.7 ha of the depression with a 10.7 ML capacity. Fresh water from the river was pumped over the earthen bank in the depression. Two artificial floods (29.97 ML between 25 July and 30 August 2005 and ML between 25 September and 5 December 2006) kept the water level to the 12.0 m AHD (Australian Height Datum) contour, holding the water at 1 m depth in the deepest point throughout the duration of the trials. In addition, two of the SIS production wells were located in the study site and were in operation during the study period except from November 2006 to May During this period the wells were shut down due to technical problems. When they were in operation they were extracting the groundwater at a rate of 1-3 L/s. The groundwater lowering began operation mid-way through the first inundation and stopped mid-way during the second inundation due to a fault in the disposal pipeline. These irregularities in SIS operation affected groundwater levels, which accompanied with river levels rise caused difficulties in interpreting groundwater levels and salinity data due to each of the events. After the second trial of artificial flooding, when no water was pumped to the floodplain depression, a decrease in water levels across all wells was recorded, correlating with the decrease in river levels. 110

132 Figure 4.3 Clark s Floodplain depression during the second trial in August 2005 (a: River Murray water is pumped to the depression over the constructed earthen bank, b. the floodplain depression is filled with the river water) (Holland et al., 2009) Numerical model set-up HydroGeoSphere The HydroGeoSphere (HGS) model (Therrien et al., 2006) was selected for this study because of the hydrological processes involved, including bank recharge, surface recharge (artificial flooding, rainfall), possible river depletion, groundwater lowering and evapotranspiration (ET). Moreover, HGS was chosen over more commonly used numerical models such as MODFLOW. Brunner 2010 et al. (2010) showed that HGS is a more sophisticated model that includes a more realistic physical coupling between surface water and groundwater. This is particularly the case when considering the unsaturated zone, for accurate estimation of infiltration rates under a river and for inclusion of a transition zone (Brunner et al., 2010). Further details on the code and a recent review of software are provided by Therrien et al., (2006) and Brunner and Simmons (2012), respectively. Here, HGS used the control volume finite element method to solve the flow equations for all domains considered in the simulation, and the control volume finite element method to solve the transport equation. A dual node approach was used to couple the surface and sub-surface domains (Liggett et al., 2012). The model solves non-linear equations for variably-saturated subsurface flow, surface flow and solute transport. 111

133 Governing Equations Subsurface flow The following modified form of the Richards' equation is used to describe threedimensional transient subsurface flow in a variably-saturated porous medium: Ψ. (q) + г ex ± Q = S ω S s t + (φs w) t where г ex represents the volumetric exchange rate between the subsurface and surface domains. Exchange with water outside of the simulation domain, as specified from boundary conditions, is represented by Q, which is a volumetric fluid flux per unit volume. Also, S ω is the water saturation, S s is the specific storage, φ is the porosity, Ψ is the pressure head and q is the Darcy flux of water given by: q = K. k rw (Ψ + z) where K is the hydraulic conductivity, k rw is the relative permeability of the medium and z is the elevation head. Surface flow Surface flow is modelled using a two-dimensional depth-averaged diffusion-wave approximation of the Saint Venant equation for surface water flow:. (d s q s ) d s г s ± Q = (φ sh s ) t where d s is the depth of surface water flow, h s is the water surface elevation, г s is the volumetric exchange rate between the surface and subsurface domains, φ s is the surface water domain porosity and q s is the flux of water given as: q s = K s. K rs (d s z s ) where K rs is the relative permeability of the surface water domain, and K s is the conductivity which is derived using Manning s formula. Evapotranspiration This section is adopted from Alaghmad et al. (2014a). ET is calculated as a combination of transpiration and evaporation (Alaghmand et al., 2014a; Nouri et 112

134 al., 2012). The rate of transpiration (Tp) is estimated using the following relationships (Kristensen and Jensen, 1975): T p = f 1 (LAI) f 2 (θ) RDF [E p E can ] where LAI is the leaf area index, θ is the water content, Ep is the reference potential evapotranspiration, and Ecan is the canopy evaporation. The vegetation function (f1) correlates the transpiration (Tp) with the leaf area index (LAI) in a linear fashion and the moisture content (θ) function (f2) correlates Tp with the moisture state at the roots. The root zone distribution function (RDF) is defined by the relationship: RDF = c2 c1 rf(z)dz Lr rf(z)dz 0 where C1 and C2 are fitting parameters, Lr is the effective root length, z is the depth coordinate from the soil surface and rf(z) is the root extraction function, which typically varies logarithmically with depth. In HGS, evaporation from the soil surface and subsurface soil layers is a function of water content and an evaporation distribution function (EDF) over a prescribed extinction depth. The model assumes that evaporation (Es) occurs along with transpiration, resulting from energy that penetrates the vegetation cover and is expressed as (Therrien et al., 2010): E s = α (E p E can ) [1 f 1 (LAI)] EDF where α is a wetness factor that depends on the moisture content at the end of the energy-limiting stage and below which evaporation is 0. Solute Transport Three-dimensional transport of solutes in a variably-saturated porous matrix is described by the following equation:. ω m (qc θ s S ω D C) + [ω m θ s S ω RλC] + Ω ex ± Q c = ω m [ (θ ss ω RC t + θ s S ω RλC] 113

135 where C is the solute concentration and λ is a first-order decay constant. Solute exchange with the outside of the simulation domain, as specified by the boundary conditions, is represented by Q c. R and D are the retardation factor and hydrodynamic dispersion tensor, respectively (Bear, 1972; Freeze and Cherry, 1979). Also, Ω ex represents the mass exchange rate of solutes per unit volume between the subsurface and surface domains. ω m and θ s are the volumetric fraction of the total porosity occupied by the porous medium and the saturated water content, respectively. For further details on the governing equations, the reader is referred to Therrien et al. (2010) Model domain and grid HGS is a fully-integrated surface-subsurface numerical model. In this context, parameter values for the subsurface domain (porous media) and surface domain (river and floodplain) and ET are assigned separately. Hence, the river-floodplain system was conceptualized as a zone of porous media (consisting of the three soil formations), overlain by surface flow cells (consisting of the river, the floodplain (and the highland)). The surface flow cells allow the river to flow up its banks, so that the river width increases during high flows. The domains were linked with a dual node arrangement, with a coupling length of 0.01m. Previous research has showed little sensitivity to coupling length for values less than this (Doble et al., 2012). As illustrated in Figure 4.4a, the geometry grid covered 92 ha of Clark s Floodplain and a 1,550 m length of the river bank. According to the available drill log data (DES, 2013) three soil types were included in the model, namely Coonambidgal Clay, Monoman Sand and Upper Loxton Sand. The geometry grid of the study site was built using available GIS and LiDAR data. For this study, a digital elevation model of the study site was generated at a 10 m grid resolution using LiDAR data. A total of 15 non-uniform sub-layers were built ranging from 0.1 m up to 2.0 m in thickness including finer grids in the upper layers to model more efficiently the flow and solute exchange processes occurring between the surface and sub-surface domains (Figure 4.4b). The final grid geometry consisted of 146,800 nodes that formed 269,925 elements. Figure 4.4 shows the model domain and the geometry grid of the study site. 114

136 Domain properties The properties of the subsurface domain (porous media) of the model and unsaturated van Genuchten function parameters (van Genuchten, 1980) were adopted from Jolly et al. (1993) and Doble et al. (2006) who adjusted and proposed van Genuchten parameters for the Lower Murray River soil types including semi-confining heavy Coonambidgal Clay, Monoman Sands and Upper Loxton Sands. An ET function was applied to the surface domain. Two different ET function parameter values were considered according to the vegetation cover of the study site. As shown in Figure 4.4c, the study site is covered by Eucalyptus trees and grass. The ET parameter values for Eucalyptus and grass were adopted from Doody et al. (2009), Banks et al. (2011) and Verstrepen (2011). Under natural conditions, surface hydraulic parameters such as friction (x and y- planes), rill and obstruction storage height and coupling length have significant differences. So, the surface domain was divided into the main channel (the river) and the floodplain (including the highland). Moreover, riverbed hydraulic conductivity can vary spatially and temporally over several orders of magnitude (Calver, 2001; Wang et al., 2014). This can be due to erosion/deposition events (Hatch et al., 2010), biological activities (Treese et al., 2009) and temperature dependent material properties (Engeler et al., 2011). However, in numerical modelling studies the riverbed is often conceptualized as a homogeneous geologic structure with properties obtained through model calibration because quantifying riverbed heterogeneity in the field is often challenging. So, the riverbed complexity is replaced with homogeneous equivalents (Irvine et al., 2012). Irvine et al. (2012) showed that understanding the state of connection is the key to quantifying potential errors produced by numerical models when homogeneous equivalents are obtained by means of model calibration. They concluded that in connected and disconnected flow regimes, as defined by Brunner et al. (2009a) and Brunner et al. (2009b), homogeneous equivalence is an efficient assumption. However, in a transitional regime a homogeneous equivalent will result in errors for simulations of both a rising and falling water table (Irvine et al., 2012). In this case, the river-floodplain is a connected flow regime during the study period. The surface flow domain properties were adopted from Doble et al. (2012), Banks et 115

137 al. (2011), McCallum et al. (2010). All the model parameters values are summarized in Table 4.1. (a) (b) (c) Figure 4.4 a: 3D illustration of the study site (Z magnification: 5). Observation wells and SIS production wells are shown in red and blue circles, respectively. The green line indicates the extent of the floodplain depression. b: Conceptual configuration of the surface domain including the river (shown in red), the floodplain and the highland (shown in green). c: Soil layers and conceptual vegetation cover along Transect 2. Red columns represent the observation wells. 116

138 Table 4.1 Selected model parameters Model Parameters Value Units Maximum time step 1 days Minimum time step 0.1 days Monoman Sand hydraulic conductivity 20 m d -1 Coonambidgal Clay hydraulic conductivity 0.1 m d -1 Upper Loxton Sand hydraulic conductivity 10 m d -1 Monoman Sand porosity 0.35 Coonambidgal Clay porosity 0.6 Upper Loxton Sand porosity 0.4 Specific storage Monoman Sand 1.6 x 10-4 Specific storage Coonambidgal Clay 2.0 x 10-3 Specific storage Upper Loxton Sand 1.0 x 10-4 Residual water content Monoman Sand 0.04 Residual water content Coonambidgal Clay 0.04 Residual water content Upper Loxton Sand 0.04 Monoman Sand thickness Up to 12 m Coonambidgal Clay thickness Up to 5 m Upper Loxton Sand thickness Up to 40 m van Genuchten alpha for Monoman Sand 1.69 m -1 van Genuchten alpha for Coonambidgal Clay 0.28 m -1 van Genuchten alpha for Upper Loxton Sand 0.8 m -1 van Genuchten beta for Monoman Sand 8.25 van Genuchten beta for Coonambidgal Clay 2.52 van Genuchten beta for Upper Loxton Sand 3.6 Longitudinal dispersivity for Monoman Sand 5 m Longitudinal dispersivity for Coonambidgal Clay 5 m Longitudinal dispersivity for Upper Loxton Sand 5 m Transverse dispersivity for Monoman Sand 0.5 m Transverse dispersivity for Coonambidgal Clay 0.5 m Transverse dispersivity for Upper Loxton Sand 0.5 m Evaporation extinction depth defined by a quadratic decay evaporation distribution function (EDF) 1 m Evaporation limiting saturation for Eucalyptus & Grass (min) 0.05 & 0.25 Evaporation limiting saturation for Eucalyptus & Grass (max) 0.9 & 0.9 Transpiration extinction depth defined by a quadratic decay root distribution function (RDF) for Eucalyptus & Grass 5 & 0.5 m Leaf area index (LAI) for Eucalyptus & Grass 0.5 & 0.5 m 2 m -2 Tree canopy evaporation for Eucalyptus & Grass 4.5 x 10-4 & 4.0 x 10-4 m Transpiration fitting parameter (C1) for Eucalyptus & Grass 0.6 & 0.5 Transpiration fitting parameter (C2) for Eucalyptus & Grass 0 & 0 Transpiration fitting parameter (C3) for Eucalyptus & Grass 1 & 1 Transpiration limiting saturation (wilting point) for Eucalyptus & Grass 0.05 & 0.07 Transpiration limiting saturation (field capacity) for Eucalyptus & Grass 0.1 &

139 Transpiration limiting saturation (oxic limit) for Eucalyptus & Grass 0.8 & 0.8 Transpiration limiting saturation (anoxic limit) for Eucalyptus & Grass 0.95 & 0.95 Rill storage height for river and floodplain & 0.01 m Obstruction storage height for river and floodplain 0 & m Friction (x-plane) for river and floodplain (Manning coefficient) & 0.05 T L -1/3 Friction (y-plane) for river and floodplain (Manning coefficient) & 0.05 T L -1/3 Coupling length for river and floodplain 0.01 & 0.01 m Boundary conditions The recorded river level was represented using a time-varying first-type (Dirichlet) condition at the river nodes located at the northern and southern parts of the model boundary (Figure 4.5). Also, the recorded rainfall and potential evaporation were applied to the surface domain using a time-varying second-type (Neumann) condition. For ET modelling, two types of vegetation were assigned to the floodplain surface nodes including Eucalyptus (deep root depth) and grass (shallow root depth). In the sub-surface (porous media) domain, a constant first type (Dirichlet) boundary condition was specified at the north-eastern part of the domain. A 12 m AHD constant head was adopted from potentiometric contours (AWE, 2013). Zero lateral flux conditions were specified on all other model boundaries. The two SIS production wells were represented in the model according to the recorded pumping rates and durations and were obtained from Berens et al. (2009) and White et al. (2009). Furthermore, artificial flooding was included in the model using a time-varying first-type (Dirichlet) condition at the floodplain depression for up to 1 m head (12 m AHD). To represent the solute boundary conditions, a first-type (Dirichlet) or constant concentration boundary condition was assigned. Observed groundwater concentrations at the observation wells in the floodplain and river ranged from 300 to 50,000 µs/cm. Hence, constant values were applied at the porous media boundary (representing the regional saline aquifer) and the river nodes. 118

140 Figure 4.5 Configuration of the boundary conditions of the model domain Initial conditions The initial condition model was generated from a transient simulation of conditions that represented the 50,000 days prior to the first day of the study period (01/07/2005). Outputs from the initial model compared favourably with data from the observation wells on Clark s Floodplain recorded in June 2005 prior to the artificial inundation trials and also EM31 survey in June 2005 before the trials. These were obtained from White et al. (2009). For instance, Figure 4.6a shows the simulated pre-trial groundwater salinity distribution which is in a good correlation with EM31 survey in June 2005 (Figure 4.14a). This is backed up with the results shown in Figure 4.6b which demonstrates the correlation between the observed and simulated groundwater EC at the location of the 13 observation wells. Considering the observed and simulated results it is apparent that the generated initial condition appropriately represented the status of the river-floodplain system at the beginning of the artificial flooding applications. 119

141 (a) (b) Figure 4.6 a: Simulated pre-trial (June 2005) salinity distribution; b: Correlation between the simulated and observed groundwater salinity in June 2005 at the location of the observation wells Scenarios Four main manipulation scenarios were modelled which differ by management actions. The first of these was a combination of artificial flooding and groundwater lowering (hereafter referred to as A.Flood+SIS ). In fact, this is the calibration model that showed the observed river-floodplain behaviour during the study period. The second scenario included neither artificial flooding nor groundwater lowering (hereafter referred to as No manipulation ). The third scenario only included artificial flooding (hereafter referred to as Only A.Flood ). Finally, the fourth scenario investigated the impact of groundwater 120

142 lowering only (hereafter referred to as Only SIS ). In all cases, natural processes were simulated, including evapotranspiration, the change in river level over time and lateral groundwater flows in and out of the floodplain Model Calibration In order to calibrate the model, observed data from 12 observation wells were used. The calibration model time frame was from 1/7/2005 to 1/7/2008 (1,095 days). This period covers both the artificial flooding trials that took place in August 2005 and September 2006 as well as the groundwater lowering that took place via the two SIS production wells (30F and 32F). Ideally the model performance measures should include a combination of quantitative and nonquantitative measures (Barnett et al., 2012). Brunner et al. (2008) suggested a method to map the spatial distribution of phreatic evaporation (direct groundwater evaporation through capillary rise) using remote sensing data.. Later, Li et al. (2009) used this method and concluded that in case availability of observed groundwater head data is limited, the spatial distribution of phreatic evaporation can replace the groundwater head data in the model calibration process. However, in this case well-documented observed groundwater head data with a reasonable distribution was available. In terms of flow dynamic, the qualitative goodness-of-fit of the calibration was assessed through both visual comparison between the observed and the simulated time series of hydraulic heads and quantitative evaluation using the goodness-offit statistics shown in Table 4.2. For Root Mean Squared Error (RMSE) and Mean Sum of Error (MSE), the ideal values would be zero. Also, the closer the Coefficient of Determination (r 2 ) and the Nash-Sutcliffe Model Efficiency (NSE) are to 1, the more accurate is the model. Figure 4.7 presents the simulated and observed groundwater heads at the observations wells along with the recorded river levels. Moreover, ET flux was calculated as mm/yr for the study period. This is consistent with Doble et al. (2006) who suggested mm/yr for eucalyptus trees and 0-40 mm/yr for grasslands in this area. Field data can be used to develop and calibrate an accurate salinity map. However, accurate field data with sufficient spatial resolution are often not available 121

143 (Brunner et al., 2007). Brunner et al. (2007) showed that a combination of field data and remotely sensed data can be used to regionalize point data such as electrical conductivity measured in the field. Here, the combination of the surveyed EM31 data, reported data (Berens et al., 2009; Holland et al., 2009) and recorded electrical conductivity at the observation wells were used to calibrate the modelled salinity patterns in the floodplain aquifer (See Figures 4.6a and 4.14a). From Table 4.2 and Figure 4.7, it is clear that the calibration model set-up adequately represents the interaction between the river and the floodplain. A detailed description of the evaluation metrics used to assess the numerical model performance is given in Alaghmand et al. (2014b), Bennett et al. (2013), Straatsma et al. (2013), Matthews et al. (2011) and Jakeman et al. (2006). Figure 4.7 Simulated and observed groundwater heads at the observation wells. River levels and simulated and observed groundwater heads are shown as blue lines, black lines and red dots, respectively. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively. 122

144 Table 4.2 Model performance evaluation metrics Observation Well r 2 NSE MSE (m) RMSE (m) A A A A A A A A A A A A Sensitivity and uncertainty analysis For a given model, sensitivity analysis identifies whether changes in model inputs lead to large changes in modelled outputs (Yan et al., 2012). The sensitivity analysis involves changing a model parameter by a small amount to establish how model predictions are affected by that change. Manual sensitivity analysis requires changing a single model parameter, re-running the model to obtain a new set of predicted heads and fluxes and observing the effect of the change, either by eye or numerically by differencing (Barnett et al., 2012). In fact, a focus on sensitive parameters can lead to a better understanding and better estimated values and thus reduced uncertainty (Lenhart et al., 2002). Uncertainty analysis is a broader term, encompassing the estimation of uncertainty in model results due to poorly known parameter distributions, observation errors and simplified model assumptions such as omitted processes (Yan et al., 2012). The parameters investigated for the sensitivity analysis are those where there is a degree of uncertainty of their value and where their importance to model results is not immediately clear. Other model inputs are important, but their values are more easily and reliably observed (e.g. SIS pump rates, river levels, rainfall, evaporation and vegetation cover) or are expected to be less heterogeneous and therefore robustly interpolated from observations (e.g. groundwater salinity and potentiometric heads along model boundaries). Also, surface domain parameters are not included in the sensitivity analysis because in the current study all simulations are conducted in non-flooding (non-overbank) conditions and 123

145 sensitivity of the model outputs to the surface properties are unlikely to be significant. Table 4.3 shows the sensitivity analysis parameters and their range of values. The baseline simulation is the calibrated model. In each sensitivity analysis simulation, a single input parameter is changed. The parameter values were varied within a reasonable range around the calibrated parameter values. The sensitivity analysis were conducted on a 2D slice along transect 2 (Figure 4.4c). As the model is calibrated to groundwater head and salinity observations, the sensitivity results are presented in terms of groundwater head and salinity at the location of observation well A7 (Figure 4.8). Table 4.3 Sensitivity analysis parameters and their values Parameter Low Calibrated High Unit value value value Monoman Sand hydraulic conductivity m d - Coonambidgal Clay hydraulic conductivity m d - Upper Loxton Sand hydraulic conductivity m d - Monoman Sand porosity Coonambidgal Clay porosity Upper Loxton Sand porosity van Genuchten alpha for Monoman Sand m -1 van Genuchten alpha for Coonambidgal m -1 Clay van Genuchten alpha for Upper Loxton Sand m -1 van Genuchten beta for Monoman Sand van Genuchten beta for Coonambidgal Clay van Genuchten beta for Upper Loxton Sand Evaporation extinction depth m Transpiration extinction depth (Eucalyptus) m Leaf area index

146 (a) Sensitivity of the groundwater head and salinity to Monoman Sand porosity (b) Sensitivity of the groundwater head and salinity to Monoman Sand hydraulic conductivity (c) Sensitivity of the groundwater head and salinity to van Genuchten alpha for Monoman Sand (d) Sensitivity of the groundwater head and salinity to transpiration extinction depth Figure 4.8 Results of sensitivity of groundwater head and salinity to the tested parameters 125

147 Fifteen properties which are difficult to measure in the field have been used in sensitivity tests and varied within reasonable ranges in the model to determine the impact on model calibration. Among the tested parameters which are listed in Table 4.3, Monoman Sand hydraulic conductivity and porosity, van Genuchten alpha for Monoman Sand and Transpiration extinction depth (Eucalyptus) show relatively significant impact on the groundwater head and salinity. There are represented in Figure 4.8. Considering results of the sensitivity analysis of the model, groundwater head is most sensitive to Monoman Sand hydraulic conductivity and porosity. This is because the groundwater flows in the Monoman Sand formation, so, the hydraulic conductivity produces more sensitivity in compare with Coonambidgal Clay or Upper Loxton Sand. However, they show opposite behaviour on the groundwater hydraulic heads. As higher values of Monoman Sand hydraulic conductivity leads to higher groundwater heads. While increase in the porosity creates lower groundwater heads. On the other hand, the groundwater salinity dynamic is significantly sensitive to all the four parameters. In fact, higher values for Monoman Sand hydraulic conductivity and porosity and Transpiration extinction depth (Eucalyptus) may lead to underestimation of the modelled groundwater salinity. While, applying higher value for van Genuchten alpha for Monoman Sand causes over estimation of the modelled groundwater salinity 4.3 Results Figure 4.9 shows the rate of water storage changes (in-out) in each time step in the floodplain aquifer during the study period. The rate value is positive when the recharge rate is more than total discharge rate in each time step and vice versa. Total recharge may include river bank recharges, highland recharge, rainfall and recharge from the inundated floodplain depression. The total discharge refers to the total water removed from the floodplain aquifer via groundwater lowering, ET and discharges to the river and/or highland aquifer. Figure 4.10 shows the total water flux from the surface domain (including rainfall, river and artificial flooding) to the floodplain aquifer. Figures 4.10a, 4.10b and 4.10c compare the A.Flood+SIS, Only A.Flood and Only SIS scenarios with the No manipulation scenario, respectively. 126

148 Total solute mass stored in the floodplain aquifer is one of the useful outputs of the model and can be used to investigate the solute dynamic during the study period. Figure 4.11 shows the simulated total solute mass stored in the floodplain aquifer during the study period for the defined scenarios. Figure 4.11a shows the trend of solute mass accumulation for the A.Flood+SIS and No manipulation scenarios. The stored solute mass in each time step for the Only A.Flood and Only SIS scenarios are shown in Figures 4.11b and 4.11c. (a) (b) (c) Figure 4.9 Change in water storage in the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively. 127

149 (a) (b) (c) Figure 4.10 Total water flux from the surface domain to the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively. (a) (b) 128

150 (c) Figure 4.11 Total solute mass stored in the floodplain aquifer, a: No manipulation and A.Flood+SIS scenarios, b: No manipulation and Only A.Flood scenarios, c: No manipulation and Only SIS scenarios. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively. 4.4 Discussion Combination of artificial flooding and groundwater lowering Considering Figure 4.7, two types of groundwater dynamic regimes in the floodplain can be recognised. The areas relatively close to the river bank are responsive to river fluctuations. This can be noted at observation wells A1, A3, A4, A7, A8 and A11. The response magnitude decreases with distance from the river bank. Hence, those located a long distance from the river banks, such as A2, A5, A6, A9, A10 and A12 are more influenced by the groundwater lowering. Furthermore, artificial flooding does not appear to have a significant impact on the groundwater dynamics of the whole floodplain. Overall, it appears that in this case groundwater lowering has a stronger influence on the floodplain groundwater level dynamic since the artificial flooding and river fluctuation impacts are limited to the inundated area and river bank, respectively. Under normal conditions ( No manipulation scenario), changes in water storage are correlated with river level fluctuations. But, application of groundwater lowering alters this trend. For instance, changes in water storage in the two periods are positive (recharge rate is higher than discharge rate) which can be attributed to the artificial flooding trials (Figure 4.9a). But the second spike in Figure 4.9 is higher than the first one. To some extent this is due to an increased 129

151 volume of water pumped to the floodplain depression. But the SIS production wells stopped working mid-way through the second artificial flooding trial due to a fault in the disposal pipeline. Hence, the water accumulation rate in the floodplain aquifer that was increasing during the second artificial flooding trial was accelerated when the groundwater lowering stopped. This is because when the SIS production wells stopped working, the groundwater level began to return back to its normal level. Therefore, the floodplain aquifer began to recharge from the river and highland aquifer (recharge from rainfall was not plausible) and these added to the recharge from the inundated floodplain depression. On the other hand, Figure 4.9a shows that during groundwater lowering the rate at which water leaves the system via the production wells and ET is higher than the floodplain aquifer recharge via the river and inundated floodplain depression (or rainfall). Particularly, at the beginning of the groundwater lowering in August 2005 (after the first artificial flooding trial) and the resuming of this in May 2007, the change in water storage has a high absolute value. Gradually the volume approaches to zero in both of the groundwater lowering periods. This reflects the formation of the pumping drawdown cone. In the second period of groundwater lowering (after resuming in May 2007) the change in water storage reaches zero much quicker compared to the first period (between July 2005 and November 2006). This is because during the second period, groundwater lowering was the main driver, but the first period occurred during a period of high river flow. It appears that the river and floodplain were in an equilibrium condition after August 2007 due to a combination of groundwater lowering and a stable river level. From Figure 4.10a, the water flux from the surface domain to the floodplain aquifer does not completely correlate with the river level fluctuation. For instance, two high river flows occurred during the study period, but neither of these produced a significant impact on the water flux exchange. The first one occurred in just after the first artificial flooding trial. In that period, enhanced groundwater levels beneath the inundated floodplain depression prevented a high rate of water flux from the river. Instead, the floodplain aquifer recharged from the highland aquifer. The second river high flow happened when the production wells were shut down. Consequently, during this period groundwater was rising up and approaching its normal level which led to a decline in the head gradient. This is 130

152 consistent with the change in water storage in Figure 4.9a. In fact, the minimum water flux from the river to the floodplain aquifer was calculated in April 2007 when the river level was at its lowest and the production wells were shutdown. This supports the zero change in water storage in Figure 4.9a. The water flux exchange between the surface domain and the floodplain aquifer domain suggests that during the artificial flooding and groundwater lowering, the floodplain aquifer recharges from the fresh river water (via river banks/bed and inundated floodplain depression). Hence, a relatively fresher (less saline) floodplain aquifer is expected during those periods. Figure 4.11a shows that during the operation of the salt management measures, relatively less solute mass is accumulated in the floodplain. On the other hand, as discussed earlier and shown in Figure 4.10a, despite operation of the SIS production wells, a decline in water flux from the surface domain to the floodplain aquifer is observed after the first artificial flooding trial for a short period of time. This implies that the floodplain aquifer is recharged more from the saline highland aquifer than from fresh river water during this period. This is consistent with the slight increase in stored solute mass shown in Figure 4.11a. Overall, it appears that application of salt interception measures (including artificial flooding and groundwater lowering) led to a decrease in the amount of stored solute mass in the floodplain aquifer by 4% (Figure 4.11a). The solute accumulation trend is decreasing except during the period that the SIS production wells were shut down and also after the first artificial flooding trial. In addition, there is no correlation between solute mass accumulation and river level fluctuations when both drivers are in force. Moreover, Figure 4.11 shows that an absence of salt management measures leads to a 6% increase in the accumulated solute mass in the floodplain aquifer (Figure 4.11a). In this case, the solute mass accumulation does respond slightly to river level fluctuations. As shown in Figure 4.11a, high river flows lead to a slight decrease in the solute mass accumulation in the No manipulation scenario Impacts of artificial flooding During the artificial flooding trials a 12 m AHD was formed in the floodplain depression. This was above the recorded river level at the same time which was 131

153 around 10.3 m. This led to floodplain aquifer recharge which created a temporary increase in the groundwater level below the inundated floodplain depression (See Figures 4.10a and 10b). This changed the head gradient towards both the river and the highland sides of the floodplain depression. Consequently, as shown in Figure 4.12b, there might be a potential risk of river salinity risk during the artificial flooding particularly when the groundwater is as saline as the study site. In this case, a maximum groundwater discharge 1500 m 3 /day with a salinity in the range of 5,000 and 10,000 mg/l (TDS) produces approximately 15 ton/day of salt during this period (November 2006). During the same period, 4016 ML/day flow was recorded at Lock 4 just upstream of the study site with a salinity of approximately 200 mg/l (TDS) (WaterConnect, 2013) which would produce approximately 800 ton/day of salt. Hence, in this case, the saline groundwater discharge cannot have more than a 2% impact on the river water quality. Considering Figure 4.11a, a slight increase in stored solute mass in the floodplain aquifer occurred after both of the artificial flooding trials. This is because the end of the artificial flooding application corresponds to the beginning of the absence of fresh surface water recharge. Consequently, an upward saline groundwater flow appears below and at the edge of the floodplain depression (See Figure 4.12c). Note that even after this slight increase, the groundwater aquifer remained less saline compared with the period before the artificial flooding trials. This is shown in Figures 4.14b and 4.14d which indicate higher salinities at the northern edge of the floodplain depression compared with those shown in Figures 4.14a and 4.14c, respectively. Depending on the scale of the floodplain and the flooded area, post artificial flooding salinity can be a salinity stress issue for a short period. These results indicate that the impacts of artificial flooding are limited spatially to the inundated areas and temporally to the duration of the trial. For instance, Figure 4.11b shows that the impacts of artificial flooding over the whole floodplain are not considerable. Therefore, in order to have a better understanding of the impact of artificial flooding on the solute dynamic of the flooded area, the simulated solute dynamic at the floodplain depression profile (37,000 m 2 area with around 1 m depth) is shown in Figure From Figure 4.13, it appears that after each of the artificial flooding trials, a significant reduction in stored solute mass may occur. The reduced solute mass during the first trial was only sustained for 4 to 6 132

154 months and then increased until the second trial commenced. This correlates with the time taken for the depression to naturally dry evaporation and percolation. Later, after 6 months of the second trial, the solute mass started accumulating in the profile. The general pattern is quite similar for both scenarios but a combination of groundwater lowering and artificial flooding can maintain a relatively less saline profile compared with artificial flooding alone. Despite an increased volume of water applied during the second trial (29.97 ML in the first trial compared to ML in the second), it seems that the first trial removed relatively larger amount of solute mass compared with the second trial, perhaps due to less solute mass accumulation between the first and second trials. Furthermore, the salinity after the second trial gradually increased due to a high ET rate and the absence of surface water recharge. Figure 4.12 Groundwater salinity along transect 2 at time steps 450, 520 and 600 days (Z magnification: 3). 133

155 Figure 4.13 Total solute mass stored in the floodplain depression profile for the defined scenarios during the simulated period. The blue and purple patterns represent the groundwater lowering and artificial flooding, respectively. The results of solute dynamic shown in Figures 4.11, 4.12 and 4.13 can be supported by the EM31 surveys conducted at the study site reported by White et al. (2009) (Figure 4.14). The February 2006 survey (Figure 4.14b) was carried out following dissipation of water from the first flooding trial. Figure 4.13b shows a noticeable reduction in the bulk conductivity within the extent of the inundation zone. In fact, vertical infiltration of fresh water from the soil surface resulted in a reduction in stored soil chloride at all of the flooded locations. The change is most evident over the eastern half of the inundated area, which is more depressed topographically and is the region where water pools for the longest time. The same pattern is observed after the second trial (Figure 4.14d). a. June 2005 b. Februry

156 c. September 2006 d. February 2007 e. September 2007 f. March 2008 Figure 4.14 Temporal distribution of electrical conductivity using Geonics EM31 at the study site (Holland et al., 2009) It was shown that artificial flooding is able to reduce root zone salinity beneath the flooded area and this can lead to improved vegetation health. However, these influences are generally limited spatially to the inundated zone and cannot permanently change the saline nature of the system. The model results are consistent with the findings of White et al. (2009) who investigated vegetation health response to artificial flooding using techniques such as EM31 conductivity surveys, visual tree condition assessments, juvenile river red gum heights and understory surveys. White et al. (2009) concluded that physical (water and soil) and biological (vegetation) processes respond favourably to the trials to improve vegetation condition. They recommended that artificial flooding as an intervention 135

157 technique should be considered when vegetation starts to decline. However, they also found that only trees with crown extents of 25-75% have a high chance of recovering during the intervention period Impacts of groundwater lowering From Figure 4.14, when the production wells are in operation they are able to form a lower floodplain water table. The response of each observation well (or floodplain zone) is proportional to its distance to the production wells and the river. For example, water levels in observation well A8, which is located close to the river bank closely follow the river fluctuations. In contrast, water levels in observation well A12 correspond significantly to the groundwater lowering rates. It seems that when the production wells are shut down, it takes 4 months for the groundwater to return to normal levels. The response of the bores such as A12 near to the production wells is rapid since when the SIS production wells were restarted in May 2007 an immediate drawdown can be observed. In terms of ET flux, the No manipulation scenario produces mm/day of ET while the Only SIS scenario produces mm/day. Perhaps this is due to the lowered groundwater table which keeps it below the evaporation depth for longer periods. In other words, the lower groundwater level maintained by groundwater lowering may decrease the groundwater discharge via ET. Moreover, Figures 4.9c and 10c imply that significant water flux exchange occurs during the operation of the SIS production wells. This is mainly due to the enhanced head gradient towards the floodplain aquifer due to groundwater lowering. For instance, Figure 4.10c shows that groundwater lowering creates an increased water flux from the surface domain to the floodplain aquifer compared to the Only A.Flood and No manipulation scenarios. In fact this water flux can be interpreted as river bank recharge (i.e. no rainfall or flood recharge). It appears that when groundwater lowering is in operation it can significantly increase the river bank recharge thereby creating a wider fresh water lens and this maintain a relatively less saline floodplain aquifer. 136

158 Figure 4.15 River level and simulated groundwater heads at the location of observation wells A8 and A12 for the defined scenarios. The blue pattern represents the groundwater lowering period. In terms of floodplain salinization, Figure 4.11a indicates an approximate 4% decrease in the solute mass stored in the floodplain aquifer during the study period which includes artificial flooding and groundwater lowering. It reveals that when the SIS production wells were in operation, less solute mass was accumulating. Figure 4.11c shows that the absence of the salt interception measures would lead to 6% more solute mass stored in the floodplain aquifer. In addition, it can be seen from Figure 4.11c that in the Only SIS scenario, the solute mass accumulation is strongly correlated with the SIS operation. In contrast, in the No manipulation scenario the solute mass accumulation corresponds to the river fluctuations with high river flows slightly decreasing the solute accumulation rate. It therefore appears that the 4% decrease in accumulated solute mass in the floodplain is a combined effect of solute mass removal and groundwater level lowering. Figure 4.16 explains this process. On one side (between the river and the production wells), the groundwater lowering can enhance the fresh water lens due to the boosted hydraulic gradient towards the floodplain aquifer. This is consistent with Figure 4.10a and 4.10c. But on the other side of the production wells (between the production wells and the floodplain highland), the floodplain recharges with saline groundwater from the highland aquifer, which is the source of the solute, making this zone more saline. For example, in this case the average accumulated solute masses on the river and highland sides of the production wells are approximately 7.3 kg.m -3 and 17.3 kg.m -3, respectively. Therefore, groundwater lowering is only able to keep the solute away from the river but does 137

159 not completely remove it from the floodplain. In fact, under normal conditions, a gaining river typically recharges from the groundwater aquifer. But, the groundwater lowering (here via SIS production wells) can prevent the occurrence of the saline groundwater recharge by solute mass removal and groundwater level lowering at the same time. It is interesting that the fresh zone at the top right side of Figures 4.12 and 4.16 represent the relatively fresh irrigation water from the irrigation practices on the adjacent highland which are consistent with the observed conductivity depth sections across Bookpurnong Floodplain reported by Berens et al. (2009). Therefore, it is clear that attention needs to be paid to the configuration of any groundwater lowering scheme if it is to be an effective measure to prevent saline groundwater discharge to the river and also to obtain a less saline floodplain aquifer. Indeed, if the production wells are located too close to the river banks they may attract more saline groundwater recharge from the highland aquifer and this may lead to a more saline floodplain aquifer. If the aim is to reduce regional groundwater flux into the floodplain, then the wells may be better positioned close to the highland. But, if the aim is to extend the freshwater lens, then the wells may be better located closer to the river, which would benefit riparian vegetation but may be detrimental to the health of vegetation on the highland side of the floodplain. Figure 4.16 Groundwater salinity along transect 2 for the No manipulation and Only SIS scenarios (Z magnification: 3). 138

160 4.5 Conclusion A combination of a reduction in the frequency, duration and magnitude of natural floods, rising saline water-tables in floodplains and excessive evapotranspiration have led to a decline in the condition of the dominant riparian tree species in the Lower Murray River. At Clark s Floodplain, the absence of overbank flooding during the study period resulted in severe stress and decline in the condition of long-lived vegetation on the floodplain. Groundwater lowering (via the SIS production wells) was installed to the study site as a salt interception measure in order to prevent saline groundwater discharge to the river. In addition, two artificial flooding trials have been conducted to improve the floodplain vegetation condition. A three-dimensional fully-integrated physically-based numerical model was developed. To simulate the site during the trials, well-documented observed data were used to calibrate the model. The aim was to explain the interaction between a river and a saline floodplain in a semi-arid area. According to the model results, groundwater lowering was able to lower the floodplain aquifer water-table and draw low salinity river water into the saline floodplain aquifer. The response of each observation well was proportional with its distance to the production wells and the river. When the production wells are in operation, groundwater lowering is the dominant driver of the river-floodplain interaction. Water storage rates in the floodplain aquifer are driven by both the artificial flooding and groundwater lowering, with positive rates of change during the artificial flooding trials and negative rates during the operation of the SIS production wells. In terms of salinity, it was shown that absence of the salt interception measures would lead to 6% more solute mass stored in the floodplain aquifer. However, a combination of artificial flooding and groundwater lowering resulted in a 4% decrease in accumulated solute mass in the floodplain aquifer. Of these measures, groundwater lowering had the dominant impact on the whole floodplain while artificial flooding was more significant in the localised flood area. It was shown that groundwater lowering maintains a less saline floodplain aquifer via two mechanisms. The groundwater lowering forms a wider fresh water lens on the river side and a more saline zone on the highland side. This prevents the discharge 139

161 of saline groundwater to the river. It was found that installation of the production wells close to the river banks may lead to a wider freshwater lens on the river side and more salt mass accumulation in the rest of the floodplain. After the artificial flooding has ceased, in the flooded area a slight increase in stored solute mass was observed due to the upward saline groundwater flow that can be formed beneath and at the edge of the floodplain depression. However, it was not able to increase the salinity of the whole floodplain in this case. Saline groundwater discharges to the river occurred during both of the artificial flooding trials. However, the saline groundwater discharges did not have a significant impact on the river water quality in this case, although in smaller river systems with lower flows, this could potentially be an issue. Furthermore, the washed-off solute mass from the soil profile beneath the floodplain depression due to the first artificial flood trial was only sustained for 4 to 6 months and then increased again until the second trial. Overall, it was shown that artificial flooding is an intervention technique that can be considered as a supplement for natural overbank flooding. Artificial flooding was able to decrease the soil and groundwater salinity and improve vegetation health. From an ecological point of view it can be concluded that artificial flooding delivers some of the same benefits as natural overbank floods, with increases in vegetation health condition, restoration of understory seed banks, decreases in soil and groundwater salinity and initiation of river red gum germination. However, its impacts are limited spatially to the flooded area and temporally to the flood duration and therefore artificial flooding cannot completely change the natural condition of the floodplain. Hence, artificial flooding may be considered as a short term management technique and which may need to be applied periodically if it is to be used as an effective for long term strategy. This study has been one of the first attempts to explain the river-floodplain interaction processes at this scale using a calibrated 3D fully-integrated numerical model. 140

162 References Alaghmand, S., Beecham, S., Hassanli, A., 2013a. Impacts of Groundwater Extraction on Salinization Risk in a Semi-Arid Floodplain. Nat. Hazards Earth Syst. Sci. 13(12) Alaghmand, S., Beecham, S., Hassanli, A., 2013b. A review of the numerical modelling of salt mobilization from groundwater-surface water interactions. Water Resources 40(3) Alaghmand, S., Beecham, S., Hassanli, A., 2014a. Impacts of Vegetation Cover on Surface-Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia. Environmental Processes 1(1) Alaghmand, S., Beecham, S., Jolly, I.D., Holland, K.L., Woods, J.A., Hassanli, A., 2014b. Modelling the impacts of river stage manipulation on a complex riverfloodplain system in a semi-arid region. Environmental Modelling & Software 59(0) AWE, Loxton Bookpurnong SIS Atlas. Australian Water Environment: Adelaide. Baldwin, D.S., Mitchell, A.M., The effects of drying and re-flooding on the sediment and soil nutrient dynamics of lowland river-floodplain systems: A synthesis. River Research and Applications 16(5) Banks, E.W., Brunner, P., Simmons, C.T., Vegetation controls on variably saturated processes between surface water and groundwater and their impact on the state of connection. Water Resources Research 47(11) W Barnett, B., Townley, L.R., Post, V., Evans, R.E., Hunt, R.J., Peeters, L., Richardson, S., Werner, A.D., Knapton, A., Boronkay, A., Australian groundwater modelling guidelines, In: report, W. (Ed.). National Water Commission: Canberra. Bear, J., Dynamics of fluids in porous media. American Elsevier, New York. Bennett, N.D., Croke, B.F.W., Guariso, G., Guillaume, J.H.A., Hamilton, S.H., Jakeman, A.J., Marsili-Libelli, S., Newham, L.T.H., Norton, J.P., Perrin, C., Pierce, S.A., Robson, B., Seppelt, R., Voinov, A.A., Fath, B.D., Andreassian, V., Characterising performance of environmental models. Environmental Modelling and Software Berens, V., White, M., Souter, N., Bookpurnong Living Murray Pilot Project: A trial of three floodplain water management techniques to improve vegetation condition. Department of Water, Land and Biodiversity Conservation: Adelaide. BOM, Bureau of Meteorology (BOM). 141

163 Brock, M.A., Nielsen, D.L., Shiel, R.J., Green, J.D., Langley, J.D., Drought and aquatic community resilience: The role of eggs and seeds in sediments of temporary wetlands. Freshwater Biology 48(7) Brown, D.L., An analysis of transient flow in upland watersheds: interactions between structure and process. University of California: Berkeley, p Brunner, P., Cook, P.G., Simmons, C.T., 2009a. Hydrogeologic controls on disconnection between surface water and groundwater. Water Resources Research 45(1) W Brunner, P., Li, H.T., Kinzelbach, W., Li, W.P., Generating soil electrical conductivity maps at regional level by integrating measurements on the ground and remote sensing data. International Journal of Remote Sensing 28(15) Brunner, P., Li, H.T., Kinzelbach, W., Li, W.P., Dong, X.G., Extracting phreatic evaporation from remotely sensed maps of evapotranspiration. Water Resources Research 44(8) W Brunner, P., Simmons, C.T., HydroGeoSphere: A Fully Integrated, Physically Based Hydrological Model. Ground Water 50(2) Brunner, P., Simmons, C.T., Cook, P.G., 2009b. Spatial and temporal aspects of the transition from connection to disconnection between rivers, lakes and groundwater. Journal of Hydrology 376(1 2) Brunner, P., Simmons, C.T., Cook, P.G., Therrien, R., Modeling surface water-groundwater interaction with MODFLOW: Some considerations. Ground Water 48(2) Bunn, S.E., Arthington, A.H., Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30(4) Busch, D.E., Smith, S.D., Mechanisms associated with decline of woody species in riparian ecosystems of the southwestern U.S. Ecological Monographs 65(3) Calver, A., Riverbed permeabilities: Information from pooled data. Ground Water 39(4) DES, Drill Enquiry System. SA Water: Adelaide. Doble, R., Brunner, P., McCallum, J., Cook, P.G., An Analysis of River Bank Slope and Unsaturated Flow Effects on Bank Storage. Ground Water 50(1) Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2)

164 Doody, T.M., Holland, K.L., Benyon, R.G., Jolly, I.D., Effect of groundwater freshening on riparian vegetation water balance. Hydrological Processes 23(24) Eng, K., Wolock, D.M., Carlisle, D.M., River flow changes related to land and water management practices across the conterminous United States. Science of the Total Environment Engeler, I., Hendricks Franssen, H.J., Müller, R., Stauffer, F., The importance of coupled modelling of variably saturated groundwater flow-heat transport for assessing river-aquifer interactions. Journal of Hydrology 397(3-4) Freeze, R.A., Cherry, J.A., Groundwater. Prentice-Hall Inc., New Jersey. George, A.K., Walker, K.F., Lewis, M.M., Population status of eucalypt trees on the River Murray floodplain, South Australia. River Research and Applications 21(2-3) Girard, P., Da Silva, C.J., Abdo, M., River - Groundwater interactions in the Brazilian Pantanal. The case of the Cuiabá River. Journal of Hydrology 283(1-4) Halford, K.J., Mayer, G.C., Problems associated with estimating ground water discharge and recharge from stream-discharge records. Ground Water 38(3) Hatch, C.E., Fisher, A.T., Ruehl, C.R., Stemler, G., Spatial and temporal variations in streambed hydraulic conductivity quantified with time-series thermal methods. Journal of Hydrology 389(3-4) Holland, K.L., Doody, T.M., McEwan, K.L., Jolly, I.D., White, M., Berens, V., Souter, N.J., Response of the River Murray floodplain to flooding and groundwater management: Field investigations, Water for a Healthy Country National Research Flagship. CSIRO: Adelaide, p. 65. Holland, K.L., Turnadge, C.J., Nicol, J.M., Gehrig, S.L., Strawbridge, A.D., Floodplain response and recovery: comparison between natural and artificial floods, Technical Report Series No. 13/4. Goyder Institute for Water Research: Adelaide. Irvine, D.J., Brunner, P., Franssen, H.J.H., Simmons, C.T., Heterogeneous or homogeneous? Implications of simplifying heterogeneous streambeds in models of losing streams. Journal of Hydrology Jakeman, A.J., Letcher, R.A., Norton, J.P., Ten iterative steps in development and evaluation of environmental models. Environmental Modelling and Software 21(5) Jarwal, S.D., Walker, G.R., Jolly, I.D.,, General site description; Salt and Water Movement in the Chowilla Floodplain, In: Walker, G.R., Jolly, I.D., Jarwal, S.D. (Ed.). CSIRO Division of Water Resources, pp

165 Jolly, I.D., McEwan, K.L., Holland, K.L., A review of groundwater-surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1(1) Jolly, I.D., Narayan, K.A., Armstrong, D., Walker, G.R., The impact of flooding on modelling salt transport processes to streams. Environmental Modelling & Software 13(1) Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4) Karim, F., Kinsey-Henderson, A., Wallace, J., Arthington, A.H., Pearson, R.G., Modelling wetland connectivity during overbank flooding in a tropical floodplain in north Queensland, Australia. Hydrological Processes 26(18) Kristensen, K.J., Jensen, S.E., A model for estimating actual evapotranspiration from potential evapotranspiration. Nordic Hydrology 6(3). Lamontagne, S., Leaney, F.W., Herczeg, A.L., Groundwater surface water interactions in a large semi-arid floodplain: implications for salinity management. Hydrological Processes Leblanc, M., Tweed, S., Van Dijk, A., Timbal, B., A review of historic and future hydrological changes in the Murray-Darling Basin. Global and Planetary Change Lenhart, T., Eckhardt, K., Fohrer, N., Frede, H.G., Comparison of two different approaches of sensitivity analysis. Physics and Chemistry of the Earth, Parts A/B/C 27(9 10) Li, H.T., Brunner, P., Kinzelbach, W., Li, W.P., Dong, X.G., Calibration of a groundwater model using pattern information from remote sensing data. Journal of Hydrology 377(1 2) Liggett, J.E., Werner, A.D., Simmons, C.T., Influence of the first-order exchange coefficient on simulation of coupled surface subsurface flow. Journal of Hydrology (0) Maheshwari, B.L., Walker, K.F., McMahon, T.A., Effects of regulation on the flow regime of the River Murray, Australia. Regulated Rivers: Research & Management 10(1) Matthews, K.B., Rivington, M., Blackstock, K., McCrum, G., Buchan, K., Miller, D.G., Raising the bar? - The challenges of evaluating the outcomes of environmental modelling and software. Environmental Modelling and Software 26(3) McCallum, J.L., Cook, P.G., Brunner, P., Berhane, D., Solute dynamics during bank storage flows and implications for chemical base flow separation. Water Resources Research 46(7). 144

166 Motta, D., Abad, J.D., Langendoen, E.J., García, M.H., The effects of floodplain soil heterogeneity on meander planform shape. Water Resources Research 48(9) W Munns, R., Physiological processes limiting plant growth in saline soils: some dogmas and hypotheses. Plant, Cell & Environment 16(1) Naiman, R.J., Décamps, H., The ecology of interfaces: Riparian zones. Annual Review of Ecology and Systematics Nouri, H., Beecham, S., Hassanli, A.M., Ingleton, G., Variability of drainage and solute leaching in heterogeneous urban vegetation environs. Hydrology and Earth System Sciences 17(11) Nouri, H., Beecham, S., Kazemi, F., Hassanli, A.M., A review of ET measurement techniques for estimating the water requirements of urban landscape vegetation. Urban Water Journal 10(4) O'Malley, C., Sheldon, F., Chowilla floodplain biological study. Nature Conservation Society of South Australia: Adelaide. Ohlmeyer, R.G., Investigation of the feasibility of manipulating water levels in the River Murray. South Australian Engineering and Water Supply Department: Adelaide, p. 47. Rassam, D.W., Peeters, L., Pickett, T., Jolly, I., Holz, L., Accounting for surface-groundwater interactions and their uncertainty in river and groundwater models: A case study in the Namoi River, Australia. Environmental Modelling and Software Rassam, D.W., Werner, A.D., Review of groundwater surfacewater interaction modelling approaches and their suitability for Australian conditions, ewater Technical Report. ewater Cooperative Research Centre: Canberra. Russo, T.A., Fisher, A.T., Roche, J.W., Improving riparian wetland conditions based on infiltration and drainage behavior during and after controlled flooding. Journal of Hydrology Schmid, W., Hanson, R.T., III, T.M.M., Leake., S.A., User s guide for the Farm process (FMP) for the U.S. Geological Survey s modular three-dimensional finite difference ground-water flow model, MODFLOW-2000, USGS Techniques and Methods 6-A17. USGS: Reston, Virginia. Slavich, P.G., Walker, G.R., Jolly, I.D., Hatton, T.J., Dawes, W.R., Dynamics of Eucalyptus largiflorens growth and water use in response to modified watertable and flooding regimes on a saline floodplain. Agricultural Water Management 39(2-3) Smith, F.M., Kenny, S.K., Floristic vegetation and tree health mapping, River Murray Floodplain, South Australia. Department for Environment and Heritage: Adelaide. 145

167 Sophocleous, M.S., Perkins, P., Methodology and application of combined watershed and ground-water models in Kansas. Journal of Hydrology 236(3-4) Straatsma, M.W., van der Perk, M., Schipper, A.M., de Nooij, R.J.W., Leuven, R.S.E.W., Huthoff, F., Middelkoop, H., Uncertainty in hydromorphological and ecological modelling of lowland river floodplains resulting from land cover classification errors. Environmental Modelling and Software Swain, E.D., Wexler, E.J., A coupled surface-water and ground-water flow model (MODBRANCH) for simulation of stream-aquifer interaction, Techniques of Water Resources Investigations of the United States Geological Survey. United States Geological Survey: Washington. Tagliaferro, M., Miserendino, M.L., Liberoff, A., Quiroga, A., Pascual, M., Dams in the last large free-flowing rivers of Patagonia, the Santa Cruz River, environmental features, and macroinvertebrate community. Limnologica 43(6) Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-Dimensional Numerical Model Describing Fully- Integrated Subsurface and Surface Flow and Solute Transport. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-dimensional Numerical Model Describing Fullyintegrated Subsurface and Surface Flow and Solute Transport, Code Documentation and User s Guide. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. Timbal, B., Arblaster, J., Braganza, K., Fernandez, E., Hendon, H., Murphy, B., Raupach, M., Rakich, C., Smith, I.K.W., Wheeler, M., Understanding the anthropogenic nature of the observed rainfall decline across South Eastern Australia, CAWCR technical report; 26: Melbourne, p Treese, S., Meixner, T., Hogan, J.F., Clogging of an effluent dominated semiarid river: A conceptual model of stream-aquifer interactions. Journal of the American Water Resources Association 45(4) van Genuchten, M.T., A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Science Society of America Journal 44(5) Verstrepen, L., Evaluating rainwater harvesting on watershed level in the semi-arid zone of Chile, Bioscience Engineering. Universiteit Gent: Gent, p Walker, K.F., Thoms, M.C., Environmental effects of flow regulation on the lower River Murray, Australia. Regulated Rivers: Research & Management 8(1-2) Wang, W., Li, J., Chen, X., Cheng, D., Jia, J., Estimating streambed parameters for a disconnected river. Hydrological Processes 28(10)

168 WaterConnect, River Murray Water Data Wolman, M.G., Leopold, L.B., River floodplains: some observations on their formation, USGS Professional Paper 282-C. US Geological Survey: Washington. Yan, W., Li, C., Woods, J.A., Waikerie to Morgan Numerical Groundwater Model 2012 Volume 1: Report and Figures. Government of South Australia through Department for Water: Adelaide, p Zilli, F.L., Paggi, A.C., Ecological responses to different degrees of hydrologic connectivity: Assessing patterns in the bionomy of benthic chironomids in a large river-floodplain system. Wetlands 33(5)

169 5 Impacts of groundwater extraction on floodplain salinization risk Overview Chapter 5 investigates the impacts of groundwater lowering, as a salt interception measure, on floodplain salinization in a semi-arid saline floodplain. The numerical model is built and calibrated based on observed data collected at a study site that has been part of a groundwater lowering salt interception scheme (SIS) since July To assess the impact of groundwater lowering on floodplain salinization risk, two scenarios are defined, namely with and without the SIS. The results demonstrate that groundwater lowering is an effective salt interception measure and may mitigate floodplain salinization through the following three mechanisms. First, the groundwater extraction reverses the hydraulic gradient towards the production wells in the floodplain. This leads to a larger amount of fresh river water flux (bank recharge) to the saline floodplain aquifer and a wider fresh water lens, which can be beneficial for riparian vegetation health. Meanwhile, a deeper water-table is observed as a result of groundwater extraction. Hence, less groundwater is discharged via ET in the with-sis scenario due to a shallower water table. This relatively alters the solute concentration mechanism. Second, the application of groundwater lowering reduces the flux from the regional aquifer to the floodplain and keeps the saline groundwater away from the river and the riparian vegetation zone. Hence, the saline groundwater discharge to the river is unlikely to happen during the operation. Third, some portion of the solute mass is extracted via the production wells. However, this is not significant in comparison with the total mass stored in the floodplain 148

170 aquifer. Furthermore, it is shown that this salt interception measure may remove some of the solute mass stored in the unsaturated zone which is spatially variable. Overall, it seems that the current management of the SIS operations at the study site is effective in maintaining the floodplain salinity at a stable level. The outcomes of this study were published in the journal Natural Hazards and Earth System Sciences (NHESS) (Paper 4). The manuscript was co-authored by my Principal Supervisor, Prof. Simon Beecham, and advisor, Dr. Ali Hassanli. The format of the paper has been changed to be consistent with the rest of this thesis. 149

171 Paper 4: Impacts of groundwater extraction on salinization risk in a semi-arid floodplain Published as: Alaghmand, S., Beecham, S. and Hassanli, A. (2013), Impacts of Groundwater Extraction on Salinization Risk in a Semi-Arid Floodplain, Natural Hazards and Earth System Sciences, European Geosciences Union, 13(12), pp Abstract: In the Lower River Murray in Australia, a combination of a reduction in the frequency, duration and magnitude of natural floods, rising saline water-tables in floodplains and excessive evapotranspiration have led to an irrigation-induced groundwater mound forcing the naturally saline groundwater onto the floodplain. It is during the attenuation phase of floods that these large salt accumulations are likely to be mobilised and will discharge into the river. This has been highlighted as the most significant risk in the Murray-Darling Basin and the South Australian Government and catchment management authorities have subsequently developed salt interception schemes (SIS). The aim of these is to reduce the hydraulic gradient that drives the regional saline groundwater towards the River Murray. This paper investigates the interactions between a river (River Murray in South Australia) and a saline semi-arid floodplain (Clark s Floodplain) that is significantly influenced by groundwater lowering due to a particular SIS. The results confirm that groundwater extraction maintains a lower water-table and more amount of fresh river water flux to the saline floodplain aquifer. In terms of salinity, this may lead to less solute stored in the floodplain aquifer. This occurs through three mechanisms, namely extraction of the solute mass from the system, reducing the saline groundwater flux from the highland to the floodplain and changing the floodplain groundwater regime from a losing to a gaining one. It is shown that groundwater extraction is able to remove some of the solute stored in the unsaturated zone and this can mitigate the floodplain salinity risk. A conceptual model of the impact of groundwater extraction on floodplain salinization has been developed. 150

172 5.1 Introduction As groundwater moves from a highland aquifer to the river, it needs to pass under the floodplain. Due to the high rate of evapotranspiration in arid and semi-arid regions such as the Lower River Murray in South Australia, part of the groundwater discharges to the floodplain and leaves salt in the floodplain soil (Figure 5.1). Overbank floods leach salt from the upper soil layers to the groundwater, wash salt off the soil profile and add fresh water to the floodplain soils. The highly variable nature of surface flow in arid/semi-arid regions has led to regulation of rivers by weirs and storage infrastructure (Jolly et al., 1996). This has affected surface-groundwater interactions in the floodplains. For example, the removal of salt by overbank floods occurs less frequently. A combination of reduction in the frequency, duration and magnitude of natural floods, rising saline water-tables in floodplains (due to river manipulations and irrigated agricultural land drainage) and excessive evapotranspiration (ET) have led to an irrigationinduced groundwater mound forcing the naturally saline groundwater onto the floodplain at a relatively high flow rate (Holland et al., 2009a; Jolly et al., 1993). This has caused reduction of leaching of salt from root zones and accumulation of salt in unsaturated zones causing dieback of environmentally important riparian vegetation such as red gum (Eucalyptus camaldulensis) and black box (Eucalytpuslargiflorens) and a decline in river water quality (Allison, 1990; Herczeg, 1993; Jolly et al., 1996; Jolly et al., 1993; Peck, 1973, 2003). Another example is Mona Park district along the Burdekin River in Northern Australia where widespread use and application of imported surface water resulted in rising water-table levels and the formation of a large groundwater mound during the wet season of 2000 (Petheram et al., 2008). 151

173 Figure 5.1 Conceptual model of groundwater inputs to the floodplain and potential groundwater discharge pathways within the floodplain in the Lower River Murray (Holland et al., 2009a). Prior to 2010, a high river flood event had not occurred for 13 years. However, salt accumulation has continued over this period. The Independent Audit Group for Salinity (IAG-Salinity) in their report (MDBA, 2010) mentioned the likelihood of severe salt accessions during flood recessions. This was articulated in their 1 st recommendation and in the previous audit reports. IAG-Salinity considers this as the most significant risk in the Murray-Darling Basin. As an effort to reduce the immediate risk of river salt accession induced by increased saline groundwater levels, due to field irrigation and excessive evaporation, the South Australian Government and catchment management authorities have developed salt interception schemes to pump the highly saline groundwater mixed with irrigation recharge from the floodplain to evaporation basins (DWR, 2001). Each bore yields 2-3 L s -1 to reduce the hydraulic gradient that drives the regional saline groundwater towards the River Murray and this has improved river water quality (Berens et al., 2009). The SIS bores have been in operation since August 2005 except for some periods of shut down (e.g. from November 2006 to May 2007). It is expected they will prevent about 200 tonnes of salt per day from entering the River Murray by 2040 (White et al., 2009). Before the SISs were operational, an irrigation-induced groundwater mound forcing the naturally saline groundwater onto the floodplain at a relatively high flow rate, thereby increasing 152

174 soil salinity in the root zone of the floodplain woodlands (Doble, 2004; Viezzoli et al., 2009) (Figure 5.1). For instance at Clark s Floodplain, field investigations have shown that significant salt accumulation and vegetation dieback has occurred. This is due to evapotranspiration from rising floodplain water-tables, altered flow regimes and increased irrigation in the surrounding highlands on this floodplain (Doble, 2004). Some of the most challenging aspects of water resources studies concern the interaction between surface and groundwater (Wheater et al., 2010). Rassam (2011) classified flow and solute exchange between a river and a floodplain aquifer into four categories: (1) natural exchange flux due to river stage fluctuations such as flooding (within-bank or overbank), base-flow discharge, reservoir regulations, etc. (Brutsaert and Lopez, 1998; Chen, 2003; Moench and Barlow, 2000; Squillace, 1996), (2) exchange flux induced by pumping wells in adjacent aquifers (Chen and Shu, 2006; Sophocleous et al., 1995; Sun and Zhan, 2007), (3) exchange flux due to changes in recharge rates; and (4) exchange flux due to changes in evapotranspiration. Groundwater extraction is an important process that affects the exchange flux between surface water and groundwater. Extraction-induced river depletion is defined as the reduction of river flow due to induced infiltration of stream water into the aquifer or the capture of aquifer discharge to the river (Rassam, 2011). The temporal and spatial scales at which these processes contribute to the exchange flux is variable. For instance, river depletion resulting from groundwater extraction is delayed by time lags that range from days to hundreds of years. Likewise, the extent of the groundwater extraction activity may vary along a river reach thus leading to gaining and losing sub-reaches. Because of the intensive spatial and temporal variability there is a need for dynamic modelling of their impacts on river flows. Near-river-aquifer systems are complex due to the difficulties associated with estimating flows and solute mass transfers into and out of the aquifer, the complicated nature of the groundwater-surface water interaction processes, and the uncertainty of aquifer properties (Sophocleous, 2010). Because of this complexity, computer models are often used to model groundwater systems and to estimate the exchange flux between surface water and ground-water. These models are computer-based numerical solutions to the boundary value problems 153

175 of concern (Wheater et al., 2010). In this regard, the need to accurately quantify and forecast surface and groundwater interactions has promoted the use of physically-based numerical modelling approaches in many studies (Beven, 2001, 2002, 2006; Beven and Binley, 1992; Ebel and Loague, 2006; Loague and VanderKwaak, 2004; Nasonova and Gusev, 2008). Physically-based models are generally founded on the blueprint for a physically-based mathematical model of a complete hydrological system developed by Freeze and Harlan (1969). Popular physically-based models include HydroGeoSphere (HGS) (Therrien et al., 2005), Integrated Hydrology Model (InHM) (VanderKwaak and Loague, 2001; VanderKwaak, 1999), MODular Hydrologic Modelling System (MODHMS) (HydroGeoLogic Inc, 2006), ParFlow (Kollet and Maxwell, 2006), MIKE SHE (Abbott et al., 1986), Modular Modelling System (MMS) (Leavesley et al., 1996), CATchmentHYdrology (CATHY) (Camporese et al., 2010), FIPR hydrologic model (FHM) (Ross et al., 1997), and Penn State Integrated Hydrologic Model (PIHM) (Qu and Duffy, 2007). Modelling of surface-groundwater interactions needs knowledge of groundwater modelling, but also a detailed understanding of the exchange processes that occur between the surface and sub-surface domains (Barnett et al., 2012). Surface-groundwater interactions have been investigated in several studies (Hoehn and Scholtis, 2011; Kollet and Maxwell, 2006; Krause et al., 2007; Lamontagne et al., 2005; Lenahan and Bristow, 2010; Liang et al., 2007; Meire et al., 2010; Panday and Huyakorn, 2004; Shlychkov, 2008; Sophocleous and Perkins, 2000; Winter, 1999), but floodplains in arid/semi-arid regions have received considerably less attention (Jolly et al., 2008). One of the major limitations in this regard is lack of high quality observed data (Pilgrim et al., 1988). This has resulted in application of experiences from humid regions to drier regions without knowledge of the consequences. At best, such results will be highly inaccurate while at worst, they can be adopted for inappropriate management solutions which disregards the key features of arid/semi-arid areas (Wheater et al., 2010). One issue can be the key role of salinity in arid and semiarid floodplains (Hart et al., 1991) and the role of the unsaturated zone as one of the main components of solute mass storage in the system. This paper investigates the interactions between a river (River Murray in South Australia) and a saline floodplain (Clark s Floodplain) in a semi-arid area 154

176 significantly influenced by groundwater lowering due to the Bookpurnong SIS. Hence, the main objective of this research is to quantify the relative impacts of the groundwater lowering on the surface-groundwater interactions in a semi-arid saline floodplain to investigate the dynamics of both flow and solute. To this aim two numerical model scenarios are defined including one with SIS operation (with-sis) and another without SIS operation (without-sis). The question is what could be the water and solute dynamic at the study site if there was not any groundwater lowering. It was hypothesized that groundwater extraction via the SIS may lead to a lower water-table and a less saline floodplain aquifer. Moreover, the numerical model s capabilities to reproduce surface and groundwater flow and solute dynamics are also tested. In this regard, a physically based numerical model is developed and calibrated according to well-documented observed surface and groundwater data. This paper describes the development and calibration of a numerical model and the application of this model according to the defined scenarios. During evaluation of the scenarios, the calibrated model ( ) is used without further parameter changes. 5.2 Study site The study was the Clark s floodplain adjacent to the River Murray in the Bookpurnong Irrigation District of the Riverland region of South Australia (Figure 5.2). The area, which is located approximately 12 km upstream from the township of Loxton, has been the focus of trials to manage a marked decline in tree health that has been observed along the River Murray in South Australia. The study site is typically vegetated by a mixture of river red gum (Eucalyptus camaldulensis), black box (Eucalyptus largiflorens), river cooba (Acacia stenophylla) and lignum (Muehlenbeckia florenta). The study site is located within the semi-arid inland of Australia, with annual rainfall varying between 200 and 300 mm and annual potential evaporation of approximately 1800 mm. 155

177 Figure 5.2 Configuration of SIS production wells (in blue) and observation wells (in red) at the Clark s Floodplain. The inset map shows the location of the Bookpurnong floodplain in Australia. The geometry of the developed model in this study covers the upper 15 m of the floodplain aquifer that includes two soil types. The overlying Coonambidgal Clay ranges from 2 to 7 m thick, while the underlying Monoman Sand Formation is approximately 7 m thick in this area. The cliffs adjacent to the floodplains consist of a layer of Woorinen Sands over Blanchtown Clay, each approximately 2 m thick, overlying a layer of Loxton Sands up to 35 m in depth. The whole area is underlain by the Bookpurnong Beds, which act as an aquitard basement to the shallow aquifer that encompasses the Monoman Formation and Loxton Sands (Doble et al., 2006). Saline groundwater lies beneath the floodplain, within the Monoman Formation, with the depth to the water-table ranging from 2 to 4 m below the surface. The majority of the floodplain groundwater has an approximate electrical conductivity of 50,000 (μs cm -1 ). It is worth noting that the physiological limit for water uptake in this environment is 30,000 (μs cm -1 ) by river red gums and 55,000 (μs cm -1 ) by black box trees (Overton and Jolly, 2004). The selected parameters for the porous media are summarised in Table 5.1. A more detailed description of the study site is discussed by Brown and Stephenson (1991), Jarwal (1996) and Doble et al. (2006). 156

178 Table 5.1 Porous media and van Genuchten function parameter values Soil type Parameter Coonambidgal Clay Monoman Sand Units k (m d -1 ) Specific storage (m -1 ) Transverse dispersivity (m) Longitudinal dispersivity 3 3 (m) Porosity a (m -1 ) n Residual saturation Numerical model The HydroGeoSphere (HGS) model is capable of simulating fully coupled surface/sub-surface flow and transport. The subsurface module is based on the University of Waterloo and Université Laval three-dimensional (3D) subsurface and transport code FRAC3DVS (Therrien, 1992). The surface module is based on the Surface Water Flow Package of the MODHMS simulator, which is itself an enhancement of the popular U.S. Geological Survey code MODFLOW (Brunner and Simmons, 2012). HGS requires pre- and post-processor tools in order to handle input preparation (complex topography and grids) and visualization of the outputs. In this study, Grid Builder (McLaren, 2005) and Groundwater Modelling system (GMS) (AquaVeo, 2011) were used as pre-processors to generate the input grid domain. Also, GMS was applied as a post-processor to visualize the model results. The governing equations of the HGS model are described in Therrien et al. (2010) Model set-up The River Murray 2008 stitched Digital Elevation Model (DEM) was one of several outputs delivered through the Imagery Baseline Data Program, completed in late 2008 by the Department for Water of the Government of South Australia. The DEM, completed by CSIRO, is a product of several smaller 'River Murray' DEMs, stitched together using GIS methods. The resolution of these DEMs ranges from 2 m to 50 m with the final stitched DEM having a resolution of 2 m. Where LiDAR has been used to acquire data, the vertical accuracy is 157

179 approximately +/ m. For this study, the Digital Elevation Model (DEM) of the study site was generated at a 10 m grid resolution using LiDAR data. A 10 m grid size was used for computational purposes and was adequate to model the processes in the floodplain. The vertical discretization was chosen to meet the balance between the required computational time and sufficient spatial representation of the two soil layers. Two types of soil layers were present according to the observed drill log data. Hence, a total of 20 sub-layers were considered including finer grids, with 15 sublayers for the top 5 m, and five relatively larger layers for the bottom 10 m. The top 5 sub-layers correspond to Coonambidgal Clay and the lower 15 sub-layers to Monoman Sand. The final geometry grid consisted of 78,624 nodes that form 143,500 elements. As illustrated in Figure 5.3, the geometry grid covers part of Clark s floodplain from the floodplain slope break to the River Murray main channel. This includes two SIS production wells (32FP and 34FP) and nine observation wells. In this case, the length of the river bank was 570m and the distance from the river bank to the SIS well varied between 480m and 650m (Figure 5.3). Figure 5.3 Configuration of boundary conditions for the river, floodplain and groundwater domains (Z magnification: 10). 158

180 The properties of the porous media (soil) of the model and unsaturated van Genuchten function parameters (van Genuchten, 1980) are adopted from Jolly et al. (Jolly et al., 1993) and Doble et al. (2006). They adjusted and proposed van Genuchten functions parameters for the Lower River Murray soil types including semi-confining heavy Coonambidgal Clay, Monoman Sand and two forms of transition layer (Table 5.1). In natural conditions, the hydraulic parameters of the surface domain (river bed and floodplain corridor) have significant differences and so the model was divided into the main channel (river) and the floodplain. Table 5.2 indicates the values of the surface properties of the numerical model (Therrien et al., 2005). During the time frame of the model no flow above the river bank occurred (i.e. only non-flooding conditions occurred) and so the model results are insensitive to the surface properties. Table 5.2 Surface properties values of the numerical model Surface type Parameter River Floodplain Unit x friction (Manning coefficient) T L -1/3 y friction (Manning coefficient) T L -1/3 Rill storage height (m) Obstruction storage height (m) Coupling length (m) Longitudinal dispersivity 1 1 (m) Transverse dispersivity 1 1 (m) ET is one of the main drivers of the hydrological processes occurring in an arid/semi-arid region such as the lower River Murray (Doble et al., 2006; Holland et al., 2009a). The two main vegetation types occurring at the study site (Eucalyptus tree and grass) have significantly different characteristics in terms of root depth, water demand and leaf area index. In order to obtain a better representation of the actual conditions, vegetation coverage of the floodplain was classified into two different categories. Normalized evaporation and root depth functions were mapped onto porous media elements above the maximum depths. Currently, four evaporation and root depth functions are available in HGS; constant, linear, quadratic and cubic. In this study, quadratic evaporation and root depth functions were applied. Table 5.3 shows the values of the ET components 159

181 for Eucalyptus and grass adopted from Hingston et al. (1997), Banks et al. (2011) and Verstrepen (2011). The boundary conditions for the numerical model of the study site included specified head boundaries in the porous domain which were implemented at the end of the floodplain. In this case, observed groundwater heads at the location of the 31FO, 33FO and 35FO were assigned to nodes along the model edge as shown in Figure 5.3. On the other hand, specified heads were used to lower the watertable at the location of the 32FP and 34FP consistent with their recorded pumping rates. Observed river levels for the surface domain were set at the river side of the model using specified heads. In this regard, the observed water levels downstream of Lock 4 were applied to the river nodes of the model. In addition, rainfall was simulated for the entire model surface domain beginning on day 1. ET was dynamically simulated as a combination of evaporation and transpiration processes by removing water from all model cells of the surface and subsurface flow domains within the defined zone of the evaporation and root extinction depths. The daily rainfall and potential evaporation values used in the model were based on recorded daily rainfall at the Loxton station. To represent the solute boundary conditions, first-type (Dirichlet) or constant concentration boundary conditions were assigned. Observed groundwater TDS concentrations at the observation wells in the floodplain and river ranged from 30,000 mg L -1 to 200 mg L -1. Hence, constant values were applied at the porous media boundary (representing the regional saline aquifer) and the river nodes accordingly. Figure 5.3 illustrates the configuration of all boundary conditions in the model. Table 5.3 ET component parameters values for the study site Eucalyptus Grass Coverage area (ha) Canopy storage parameter (m) Transpiration fitting C C C3 1 1 Transpiration limiting saturations wilting point field capacity oxic limit

182 anoxic limit Evaporation limiting saturations min max LAI Root depth (m) Evaporation depth (m) 1 1 Initial conditions refer to the head and solute concentration distributions throughput the model at the beginning of the simulation. In this context, fieldmeasured head values or solute concentrations do not represent the real initial condition as they are obtained at a time when the natural ground-water system is in equilibrium (Barnett et al., 2012). For instance, if the field-observed data values are used as initial conditions, the model response in the early time steps would reflect not only the model stress under study but also the adjustment of model head values to offset the lack of correspondence between model hydrologic inputs and parameters and the initial head values (Franke et al., 1987). Therefore, in a transient state problem, the initial conditions should be determined through a steady/dynamic steady-state solution to generate dynamic cyclic initial conditions such as evaporation and rainfall seasonal cycles (Anderson and Woessner, 1992). Barnett et al. (2012) suggested carrying out a simulation which begins long enough before the calibration period allowing for an initial model equilibration time. In this study, the stress period starts from 1/01/2006 and ends on 1/09/2010. So, the initial model covers a 30 year period to create the equilibrium initial condition for the stress period. The initial model was intended to show equilibrium behaviour while its last time steps should be equal to the first time steps of the stress model which are observed (Figure 5.4). Hence, simulated groundwater heads are compared with absolute observed values at observation wells (BO1: 10.4, BO2: 10.15, BO3: 10.01, BO4: 10.20, BO5: and BO6: mahd (Holland et al., 2009c)). Also, the status of the solute concentration distribution at the beginning of the study (stress) period was checked with the general solute distribution pattern at the floodplain which was observed in the field and in related reports. This can be considered as two zones; a relatively fresh groundwater zone within 50 m distance of the river banks (BO1: 6,500 µs cm -1 and BO4: 1,200 µs cm -1 ) and a saline zone (BO2: 53,000 µs cm -1, BO3: 54,

183 µs cm -1, BO5: 50,900 µs cm -1 and BO6: 52,000 µs cm -1 ) for the rest of the floodplain (Holland et al., 2009c). (a) (b) Figure 5.4 3D demonstration of simulated initial condition along transects B1 and B2: a. Porous media saturation, b. Solute concentration distribution. Observation wells are in black (Z magnification: 5) Coupled flow and transport calibration The observed hydraulic heads and groundwater solute concentrations at the observation wells are used as calibration criteria during coupled flow-andtransport calibration of the model (Barnett et al., 2012). This process aims to assess the ability of the surface-groundwater model to correctly distribute water and solute between the two domains (Li et al., 2008). Two different approaches were employed for the flow and solute dynamics calibrations. The flow dynamic was calibrated against the absolute observed groundwater levels at the observation wells. But for the solute dynamic, given the difficulty associated with the quantification of the solute transport model parameters, the solute was calibrated 162

184 to the observed general salinity patterns of the floodplain aquifer. This was because concentration patterns are much more sensitive to local-scale geological heterogeneity than are hydraulic heads, and models may have difficulty reproducing the concentrations or their temporal variability at single observation wells. The general floodplain aquifer solute distribution was obtained from the EM31 surveys adopted from Berens et al. (2009). Hence, in this case, because of significant salinity differences between 50 m distance to the river bank (BO1 and BO4: EC <5,000 µs cm -1 ) and the rest of the floodplain (BO2, BO3, BO5 and BO6: EC= 30,000-50,000 µs cm -1 ), an aggregate quantity like the plume mass is a more suitable calibration criterion, as recommended by Barnett et al. (2012). Calibration of the model was conducted manually with more consideration to the sensitive parameters including soil hydraulic conductivity, porosity and dispersivity. The model performance for both flow and solute transport was tested by visual comparison between observed and simulated series of hydraulic heads and solute concentrations at observation wells BO1, BO2, BO3, BO4, BO5 and BO6. Moreover, quantitative evaluation was undertaken using goodness-of-fit measures. Table 5.4 and Figure 5.5 demonstrate the performance of the calibrated model of the Clark s Floodplain. Seeking to optimise the goodness-of-fit by minimizing errors between the observed and simulated values, or to achieve a specific predefined value of goodness-of-fit, may be the best way to increase confidence in predictions (Barnett et al., 2012). The goodness-of-fit measures, including root r-square (R 2 ), Nash-Sutcliffe (Nr),mean sum of residuals (MSR) and root mean squared error (RMSE), are used to evaluate the simulated values against the observed data (Table 5.4). Moreover, the solute concentration distribution results show that the calibrated model was able to reproduce the surface-groundwater interaction processes in an acceptable manner as they present a good agreement. For instance, the EM31 survey in November 2007 (Figure 5.6a) showed a distinct zone of low conductivity along the eastern margin abutting the river channel. This shows the presence of freshwater within the floodplain aquifer (bank storage) and this was supported by groundwater salinity data collected at the riverbank piezometers at that time. 163

185 Table 5.4 Results of the calibrated model performance statistics Observation Wells R 2 N r MSR (m) RMSE (m) BO BO BO BO BO BO Figure 5.5 Simulated and observed groundwater heads at observation wells Figure 5.6 Simulated solute concentration distribution (a) and EM31 survey (Berens et al., 2009) (b) in November 2007 at the study site 5.4 Results and discussion Often the objectives of numerical modelling involve a quantitative assessment of the response of heads or solute concentrations to future stresses on the surface or sub-surface system. Predictive scenarios can be formulated to quantify groundwater behaviour in either absolute or relative terms. In the case of the latter, the particular modelling outcome is obtained by subtracting one model result from another (null scenario). A null scenario is a predictive model that has 164

186 no future changes in the stresses that are being investigated. Considering the prediction approach suggested in the Australian groundwater modelling guidelines (Barnett et al., 2012), even though it may be difficult to calibrate the surfacegroundwater interaction model, there may be reasonable confidence in a model to predict the right trends. In these situations, it is not common practice for one set of predictions to be made using the best possible model, and for further predictions to be presented in absolute terms. In this case, to investigate the surfacegroundwater interactions induced by groundwater lowering, the calibrated model ( ) was used as the null scenario without further parameter changes to investigate both the water balance and the solute mass balance. It should be noted that the results discussed here are from a calibrated numerical model based on available data that may include some uncertainties particularly in terms of solute dynamics. Figure 5.7 shows the groundwater heads at the boundary of the models (SIS wells) for the defined scenarios. In the without-sis scenario there are constant values equal to 10.1 m for 31FO, m for 33FO and m for 35FO (observed just before commencement of the SIS production wells). This assumes that no significant groundwater head changes occurred, while for the with-sis scenario the heads are influenced by the SIS production wells. In this paper, a losing floodplain regime corresponds to a movement of flow from the floodplain aquifer to the river and a gaining floodplain regime refers to flow movement from the river to the floodplain aquifer. Conversely, a losing river regime refers to movement of flow from the river to the floodplain aquifer and a gaining river shows flow movement from the floodplain aquifer to the river. Figure 5.7 Groundwater heads at the boundary of the models (SIS wells) for the defined scenarios 165

187 5.4.1 Water balance One of the main starting points for analysis of the flow dynamics in a surfacegroundwater system is accurate modelling of the water balance. In this case, three forms of water balance outputs are considered as indicators to compare the scenarios. These indicators include changes in water storage in the porous and/or overland domain, the amount of water movement between the two domains (flow flux) and the groundwater head profile along the observation transects. Hence, three outputs of the model are considered in the analysis of the system water balance including the change in water storage (In-Out) in the porous medium (m 3 /day), the water flux (m 3 ) from the river to the floodplain aquifer and the groundwater head profile along transect B1. The change in water storage in the floodplain aquifer is shown in Figure 5.8. The change in water storage for the without-sis scenario shows a relatively balanced trend during the study period. A correlation between the change in water storage and the river water level fluctuation are observed. This is because as the river water level increases, it increases as more water is stored in the floodplain aquifer. In contrast, a river water level decrease leads to negative value as less water enters the floodplain aquifer in compare with the water that leaves. Note that in this study a constant groundwater head is applied (assuming no significant changes in groundwater head) as the boundary condition for the without-sis scenario. This is why the accumulation rate corresponds significantly to the river water level fluctuations. For the with-sis scenario, a clear connection between the groundwater head fluctuations due to groundwater extraction and the change in water storage can be seen, with an increase in change in water storage corresponding to an increase in groundwater head. In this scenario, it seems that the groundwater lowering (due to the extraction) is the main driver rather than river water level changes. In other words, when the SIS production wells are in operation, groundwater heads decline due to extraction and this leads to negative value. But, when the SIS production wells stop working (no extraction), the groundwater heads increase as the floodplain aquifer is recharged by the river and the highland groundwater. This leads to a positive value for the change in water storage. Another explanation for this process can be a change of floodplain groundwater regime from losing (due to groundwater extraction through the SIS 166

188 production wells) to gaining (due to groundwater recharge). This shows that the river water level fluctuation is not the dominant driver in this situation; otherwise, an increase in accumulation rate would have occurred during the operation of the SIS production wells when the floodplain aquifer was in a losing regime. Figure 5.8 Changes in water storage in the porous media for the defined scenarios (light blue pattern refers to the period that the pumps were in operation). Figure 5.9 shows the amount of water that moved from the river to the floodplain aquifer during the study period for the defined scenarios, and this is clearly related to the river water level. In other words, the amount of water that moves from the river to the floodplain aquifer increases with increasing river water levels and vice versa. This shows that for the study site there is a good connection between the river and the floodplain aquifer through bank recharge. On the other hand, the general trend in both scenarios is almost the same, although the amount of flux from the river to the floodplain aquifer is relatively higher for the with-sis scenario. This is attributed to the operation of the SIS production wells that creates a groundwater gradient away from the river. In the with-sis case, fresh river water is drawn towards the SIS production wells which may result in a relatively fresher floodplain aquifer. It is worth noting that in a high river level condition, which occurred at the end of the study period, a smaller difference in the flux is observed. This means that in high flow situations the amount of flux is too high for groundwater extraction (at least at this scale) to make a significant difference. Also, when the SIS was shut down from November 2006 to May 2007, the flux from the river to the floodplain was the same. 167

189 Figure 5.9 Water flux from the river to the floodplain aquifer for the defined scenarios (light blue pattern refers to the period that the pumps were in operation). Following the SIS commencement in July 2005, a water-table gradient away from the river developed with the groundwater level at observation well BO3 being up to 0.5 m below the observed river level. From June to November 2006, under relatively stable river levels, observations indicate a groundwater gradient away from the river between BO1 (at the riverbank) and BO3 of 0.4 m over a distance of 130 m. During the SIS shutdown from November 2006 to May 2007, the groundwater levels across Transect B1 indicated a reduced gradient with BO1, BO2, and BO3 at similar elevations. Monthly means of the BO1 groundwater level hydrograph indicate groundwater elevations were greater than river levels during February, March and April 2007, indicating gaining stream conditions with the B1 and the SIS midpoint hydrographs above the recorded river level. Following the reinstatement of the SIS in May 2007, recorded levels in the Transect B1 wells indicate that a losing stream gradient was rapidly restored and maintained in the absence of further SIS stoppages. In term of the evaporation, Figure 5.10 shows that the cumulative amount of water which left the system via the evaporation in the without-sis scenario was around 18% larger than for the with-sis case. This may be due to the increased groundwater level in the without- SIS scenario which provides more water that is available to be evaporated. 168

190 Figure 5.10 Cumulative evaporation from the floodplain aquifer for the defined scenarios The dynamic of the floodplain groundwater as a hydrograph and as a longitudinal profile along transects B1 and B2 are shown in Figures 11 and 12, respectively. In Figure 5.11, the impact of groundwater lowering due to the SIS production wells is much more significant at the end of the floodplain (BO3) compared to at the river bank (BO1). The only times that the two defined scenarios show the same groundwater heads are when the SIS production wells stopped working (from November 2006 to May 2007). For instance, in March 2007 the groundwater head increased to its normal level (equal to the without-sis scenario). Given the river water level fluctuations and the groundwater responses, it can be seen that in the without-sis scenario, river water level change is the main driver of the surfacegroundwater processes. Hence, the floodplain aquifer near the river bank (BO1) is more sensitive to river water level changes compared to further away from the river bank (BO2 and BO3). In the with-sis scenario, it is groundwater lowering induced by the SIS production wells that has more influence on the system. Figure 5.12 shows three longitudinal profiles of the floodplain aquifer groundwater head. Again, areas further away from the river banks are more influenced by the SIS production wells and these influences become more significant during the SIS operation periods. 169

191 Figure 5.11 Groundwater head dynamics at the observation wells on transect B1 for the with-sis and the without-sis scenarios Figure 5.12 Groundwater head longitudinal profiles on 7/03/2007 (left), 16/12/2008 (middle) and 29/07/2009 (right) on transect B2 for the with-sis and the without-sis scenarios Solute mass balance Figure 5.13 shows the temporal trend of the total amount of solute mass stored in the system. The without-sis scenario leads to a more saline floodplain aquifer, and also the amount of solute mass stored in the floodplain aquifer increases with time. In contrast, salinity levels were reduced for the with-sis scenario with the exception of the period of time when the SIS was shut down. In fact, this was due to an increased flux of fresher river water induced by the SIS, in addition to the removal of saline groundwater and reduced saline groundwater flux to the floodplain from the highland. This is consistent with the field observations of Berens et al. (2009) and Holland et al. (2009b) at the same study site. According to these results, the total solute mass stored in the system in the with-sis scenario reduces by up to 4% (1680 tonnes) while the without-sis scenario shows a 2% (846 tonnes) increase. Depending on the scale of the model, these values can be considerable. It is worth noting that in the without-sis scenario, there is a relative decline in solute mass in the system at the end of the study period. This is due to the occurrence of high river flows and overbank flows that took place just after 170

192 the study period. Hence, in that short period, solute accumulation decreased and relatively less solute mass was stored in the system. The unsaturated zone may act as an essential component of the solute mass stored in the floodplain aquifer, particularly in an area such as the study site where salinity is driven by increased discharge of saline groundwater and reduced leaching of salts from the soils. A high rate of ET can accelerate this process. According to the results, at the last time step 7,120 tonnes solute mass was stored in the unsaturated zone of the without-sis model. The corresponding value for the with-sis model for the same time step was 5,562 tonnes. This proves that the groundwater extraction is able to remove a significant amount of solute stored in the unsaturated zone. It is worth noting that this model was up to 16 m in depth but shallower models might produce different proportions of unsaturated zone storage. Figure 5.14 illustrates the solute mass changes in the unsaturated zone for the defined scenarios. In Figure 5.14a the distribution of solute mass removed from the unsaturated zone is shown. Groundwater extraction via the SIS operation removed solute mainly from the middle part of the floodplain. Figure 5.14b shows the amount of solute mass that could be stored in the system if the SIS was not installed on the floodplain. In fact, without groundwater extraction more solute could have been stored in the floodplain aquifer. This is consistent with the results in Figure 5.13 which show that groundwater extraction may lead to a less saline floodplain as well as less solute mass storage in the unsaturated zone. Figure 5.13 Solute mass stored in the system in each time step for the defined scenarios. Cumulative pumped water is also shown in dark blue and light blue pattern refers to the period that the pumps were in operation. 171

193 Figure D visualization of solute mass changes in the unsaturated zone for the defined scenarios; a. amount of solute mass removed from the unsaturated zone during the with-sis scenario, b. amount of solute mass that could be stored in the system if SIS was not installed on the floodplain The dynamic of groundwater salinity is demonstrated in Figure It appears that at the relatively fresh buffer zone near the river bank, the groundwater salinity is almost the same in both scenarios. Away from the river bank, towards the SIS production wells, the influence of the SIS production wells can be clearly seen. The groundwater salinity slightly increases during the study period in the without- SIS scenario while in the other scenario the ability of the SIS operation to mitigate 172

194 the salinity is significant. Again, the influence is stronger in the floodplain than at distances closer to the river bank. However, in this case, groundwater extraction is not able to change the overall pattern of the salinity of the floodplain aquifer. Since even with the SIS operation there is a dramatic difference in salinity between the river bank (less than 5,000 μs cm -1 ) and the floodplain (above 40,000 μs cm -1 ) and the decrease in salinity due to the SIS is of the same order. Given the model results, a conceptual model of the impact of groundwater extraction on floodplain salinization is shown in Figure Figure 5.15 Solute dynamics at the observation wells BO1 (left), BO2 (middle) and BO3 (right) Figure 5.16 Conceptual model of the impacts of groundwater extraction on salinization risk in a semi-arid floodplain 173

195 5.5 Conclusion The relative impacts of saline groundwater extraction on the interactions between a river and its adjacent semi-arid floodplain have been investigated. A 3D fully integrated physically-based numerical model was used to simulate two defined scenarios, namely with and without SIS. The numerical model was first calibrated using observed data. The results showed a reasonable correlation between observed and simulated values. The model was able to effectively reproduce the surface-groundwater interactions. Then the calibrated model was used to simulate the defined without-sis scenario. Water balance analysis showed that groundwater extraction may change the floodplain aquifer regime from losing to gaining (or at least to reduce the losing rate). This happens by changing the head gradient towards the floodplain. This can lead to more amount of fresh river water flux to the saline floodplain aquifer and a wider fresh water lens along the riparian vegetation at the river bank. Also, a deeper water-table is observed as a result of groundwater extraction. This is more significant in the area around the production wells in the floodplain rather than closer to the river banks. In the without-sis scenario it is the river water fluctuations that dominate the surface-groundwater interactions while in the with- SIS scenario, the groundwater extraction is the main driver. Moreover, more amount of groundwater has been removed from the floodplain aquifer via evaporation in the without-sis scenario. In terms of the solute balance, the SIS results in a less saline floodplain aquifer, as evidenced by the reduced amount of solute stored in the with-sis scenario. Moreover, it was shown that groundwater extraction is able to remove significant proportions of the solute mass from the unsaturated zone. Overall, the saline groundwater extraction from the floodplain aquifer is shown to be an effective salt interception measure. This occurs through three mechanisms, namely extraction of the solute mass from the system, reducing the saline groundwater flux from the highland to the floodplain and changing the floodplain groundwater regime from a losing to a gaining one. The latter may result in more flux from the river to the floodplain aquifer. The current management of the SIS operation seems to be effective in maintaining the floodplain salinity at a stable level. 174

196 References Abbott, M.B., Bathurst, J.C., Cunge, J.A., O Connell, P.E., Rasmussen, J., An introduction to the European Hydrological System Syst`eme Hydrologique Europ een, SHE, 2: Structure of a physically-based distributed modeling system. Journal of Hydrology 87(1-2) Allison, G.B., Cook, P.G., Barnett, S.R., Walker, G.R., Jolly, I.D., Hughes, M.W., Land clearance and river salinization in the western Murray Basin, Australia. Journal of Hydrology Anderson, M.P., Woessner, W.W., Applied groundwater modelling: simulation of flow and advective transport. Academic Press, San Diego, USA. AquaVeo, GMS: Provo, UT. Banks, E.W., Brunner, P., Simmons, C.T., Vegetation controls on variably saturated processes between surface water and groundwater and their impact on the state of connection. Water Resources Research 47(11). Barnett, B., Townley, L.R., Post, V., Evans, R.E., Hunt, R.J., Peeters, L., Richardson, S., Werner, A.D., Knapton, A., Boronkay, A., Australian groundwater modelling guidelines, In: report, W. (Ed.). National Water Commission: Canberra. Berens, V., White, M., Souter, N., Bookpurnong Living Murray Pilot Project: A trial of three floodplain water management techniques to improve vegetation condition. Department of Water, Land and Biodiversity Conservation: Adelaide. Beven, K., On explanatory depth and predictive power. Hydrological Processes 15(15) Beven, K., Towards a coherent philosophy for modelling the environment. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 458(2026) Beven, K., A manifesto for the equifinality thesis. Journal of Hydrology 320(1-2) Beven, K., Binley, A., The future of distributed models: model calibration and uncertainty prediction. Hydrological Processes 6(3) Brown, C.M., Stephenson, A.E., Geology of the Murray Basin, Southeastern. BMR Bulletin 235. Brunner, P., Simmons, C.T., HydroGeoSphere: A Fully Integrated, Physically Based Hydrological Model. Ground Water 50(2) Brutsaert, W., Lopez, J.P., Basin-scale geohydrologic drought flow features of riparian aquifers in the southern Great Plains. Water Resources Research 34(2)

197 Camporese, M., Paniconi, C., Putti, M., Orlandini, S., Surface-subsurface flow modeling with path-based runoff routing, boundary condition-based coupling, and assimilation of multisource observation data. Water Resources Research 46(2). Chen, X., Stream water infiltration, bank storage, and storage zone changes due to stream-stage fluctuations. Journal of Hydrology 280(1-4) Chen, X., Shu, L., Groundwater evapotranspiration captured by seasonally pumped wells in river valleys. Journal of Hydrology 318(1-4) Doble, R., Quantifying spatial distributions of groundwater discharge and salt accumulation on a semi-arid floodplain to determine vegetation health response, School of Chemistry, Physics and Earth Sciences. Flinders University of South Australia: Adelaide, p Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2) DWR, South Australian River Murray Salinity Strategy Department for Water Resources, Government of South Australia: Adelaide. Ebel, B.A., Loague, K., Physics-based hydrologic-response simulation: Seeing through the fog of equifinality. Hydrological Processes 20(13) Franke, O.L., Reily, T.E., Bennett, G.D., Definition of boundary and initial conditions in the anaysis of saturated ground-water flow systems; an introduction, Techniques of Water Resources Investigations. USGS, p. 15. Freeze, R.A., Harlan, R.L., Blueprint for a physically-based, digitallysimulated hydrologic response model. Journal of Hydrology 9(3) Hart, B., Bailey, P., Edwards, R., Hortle, K., James, K., McMahon, A., Meredith, C., Swadling, K., A review of the salt sensitivity of the Australian freshwater biota. Hydrobiologia 210(1) Herczeg, A.L., Simpson, H.J., Mazor, E., Transport of soluble salts in a large semiarid basin: River Murray, Australia. Journal of Hydrology Hingston, F.J., Galbraith, J.H., Dimmock, G.M., Application of the processbased model BIOMASS to Eucalyptus globules subsp. Globules plantations on ex-farmland in south Western Australia: I. Water use by trees and assessing risk of losses due to drought. Forest Ecology and Management Hoehn, E., Scholtis, A., Exchange between a river and groundwater, assessed with hydrochemical data. Hydrology and Earth System Sciences 15(3) Holland, K.L., Jolly, I.D., Overton, I.C., Walker, G.R., 2009a. Analytical model of salinity risk from groundwater discharge in semi-arid, lowland floodplains. HYDROLOGICAL PROCESSES

198 Holland, K.L., M.T, D., McEwan, K.L., Jolly, I.D., White, M., Berens, V., Souter, N., 2009b. Response of the River Murray floodplain to flooding and groundwater management: Field investigations., Water for a Healthy Country National Research Flagship. CSIRO, p. 65. Holland, K.L.D., T.M., McEwan, K.L., Jolly, I.D., White, M., Berens, V., Souter, N.J., 2009c. Response of the River Murray floodplain to flooding and groundwater management: Field investigations, Water for a Healthy Country National Research Flagship. CSIRO: Adelaide, p. 65. HydroGeoLogic Inc, MODHMS: a comprehensive MODFLOW-based hydrologic modelling system, version 3.0. HydroGeoLogic Incorporated, Herndon: USA. Jarwal, S.D., Walker, G.R., Jolly, I.D.,, General site description; Salt and Water Movement in the Chowilla Floodplain, In: Walker, G.R., Jolly, I.D., Jarwal, S.D. (Ed.). CSIRO Division of Water Resources, pp Jolly, I.D., McEwan, K.L., Holland, K.L., A review of groundwater-surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1(1) Jolly, I.D., Walker, G.R., Hollingworth, I.D., Eldridge, S.R., Thorburn, P.J., McEwan, K.L., Hatton, T.J., The causes of decline in eucalypt communities and possible ameliorative approaches, In: walker, G.R., Jolly, I.D., Jarwal, S.D. (Eds.), Salt and Water Movement in the Chowilla Floodplain. CSIRO Division of Water Resources: Canberra, Australia. Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4) Kollet, S.J., Maxwell, R.M., Integrated surface-groundwater flow modeling: A free-surface overland flow boundary condition in a parallel groundwater flow model. Advances in Water Resources 29(7) Krause, S., Bronstert, A., Zehe, E., Groundwater-surface water interactions in a North German lowland floodplain - Implications for the river discharge dynamics and riparian water balance. Journal of Hydrology 347(3-4) Lamontagne, S., Leaney, F.W., Herczeg, A.L., Groundwater surface water interactions in a large semi-arid floodplain: implications for salinity management. Hydrological Processes Leavesley, G.H., Restrepo, P.J., Markstrom, S.L., Dixon, M., Stannard, L.G., The modular modeling system (MMS): User s manual, USGS Open-File Report USGS: Reston, Virginia. Lenahan, M.J., Bristow, K.L., Understanding sub-surface solute distributions and salinization mechanisms in a tropical coastal floodplain groundwater system. Journal of Hydrology 390(3-4)

199 Li, Q., Unger, A.J.A., Sudicky, E.A., Kassenaar, D., Wexler, E.J., Shikaze, S., Simulating the multi-seasonal response of a large-scale watershed with a 3D physically-based hydrologic model. Journal of Hydrology 357(3-4) Liang, D., Falconer, R., Lin, B., Coupling surface and subsurface flow in a depth averaged flood wave model. Journal of Hydrology Loague, K., VanderKwaak, J.E., Physics-based hydrologic response simulation: platinum bridge, 1958 Edsel, or useful tool? Hydrol. Proc. 16(5) McLaren, R.G., Grid Builder: A pre-processor for 2-D, triangular element, finite-element programs. Groundwater Simulations Group, University of Waterloo: Waterloo, Ontario. MDBA, Report of the Independent Audit Group for Salinity Murray-Darling Basin Authority (MDBA): Canberra, Australia. Meire, D., De Doncker, L., Declercq, F., Buis, K., Troch, P., Verhoeven, R., Modelling river-floodplain interaction during flood propagation. Natural Hazards 55(1) Moench, A.F., Barlow, P.M., Aquifer response to stream-stage and recharge variations. I. Analytical step-response functions. Journal of Hydrology 230(3-4) Nasonova, O., Gusev, E., Investigating the ability of a land surface model to reproduce river runoff with the accuracy of hydrological models. Water Resources 35(5) Overton, I., Jolly, I., Integrated studies of floodplain vegetation health, saline groundwater and flooding on the Chowilla floodplain, South Australia. Integrated Studies of Floodplain Vegetation Health, Saline Groundwater and Flooding on the Chowilla Floodplain South Australia. Panday, S., Huyakorn, P., A fully coupled physically-based spatiallydistributed model for evaluating surface/subsurface flow. Advances in Water Resources Peck, A.J., Hatton, T., Salinity and the discharge of salts from catchments in Australia. Journal of Hydrology Peck, A.J., Hurle, D.H., Chloride balance of some farmed and forested catchments in Southwestern Australia. Water Resources Research Petheram, C., Bristow, K.L., Nelson, P.N., Understanding and managing groundwater and salinity in a tropical conjunctive water use irrigation district. Agricultural Water Management 95(10) Pilgrim, D.H., Chapman, T.G., Doran, D.G., Problems of rainfall runoff modelling in arid and semi-arid regions. Hydrological Sciences Journal 33(4)

200 Qu, Y., Duffy, C.J., A semidiscrete finite volume formulation for multiprocess watershed simulation. Water Resources Research 43(8). Rassam, D.W., A conceptual framework for incorporating surfacegroundwater interactions into a river operation-planning model. Environmental Modelling and Software 26(12) Ross, M.A., Tara, P.D., Geurink, J.S., Stewart, M.T., FIPR hydrologic model users manual and technical documentation, CMHAS Water Resources Report FIPR University of South Florida: Tampa. Shlychkov, V., Numerical modeling of river flows with account for vortex generation at the channel-floodplain boundary. Water Resources 35(5) Sophocleous, M., Review: Groundwater management practices, challenges, and innovations in the High Plains aquifer, USA-lessons and recommended actions. Revue critique: Pratiques, défis et innovations dans le domaine des de la gestion des eaux souterraines de l'aquifère des Grandes Plaines (High Plains), aux Etats Unis d'amérique - Leçons et recommandations 18(3) Sophocleous, M., Koussis, A., Martin, J.L., Perkins, S.P., Evaluation of simplified stream-aquifer depletion models for water rights administration. Ground Water 33(4) Sophocleous, M.S., Perkins, P., Methodology and application of combined watershed and ground-water models in Kansas. Journal of Hydrology 236(3-4) Squillace, P.J., Observed and simulated movement of bank-storage water. Ground Water 34(1) Sun, D., Zhan, H., Pumping induced depletion from two streams. Advances in Water Resources 30(4) Therrien, R., Three-dimensional analysis of variablysaturated flow and solute transport in discretely-fractured porous media. University of Waterloo: Waterloo. Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-Dimensional Numerical Model Describing Fully- Integrated Subsurface and Surface Flow and Solute Transport. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. Therrien, R., McLaren, R.G., sudicky, E.A., Panday, S.M., HydroGeoSphere; A Three-dimensional Numerical Model Describing Fullyintegrated Subsurface and Surface Flow and Solute Transport: User Manual. Groundwater Simulations Group, University of Waterloo: Waterloo, Ontario, Canada. van Genuchten, M.T., A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Science Society of America Journal 44(5)

201 VanderKwaak, J., Loague, K., Hydrologic-response simulations for the R-5 catchment with a comprehensive physics-based model. Water Resources Research 37(4) VanderKwaak, J.E., Numerical simulation of flow and chemical transport in integrated surface-subsurface hydrologic systems. University of Waterloo: Waterloo, Canada. Verstrepen, L., Evaluating rainwater harvesting on watershed level in the semi-arid zone of Chile, Bioscience Engineering. Universiteit Gent: Gent, p Viezzoli, A., Auken, E., Munday, T., Spatially constrained inversion for quasi 3D modelling of airborne electromagnetic data - an application for environmental assessment in the Lower Murray Region of South Australia. Exploration Geophysics 40. Wheater, H.S., Mathias, S.A., Li, X., Groundwater Modelling in Arid and Semi-Arid Areas. Cambridge University Press. White, M.G., Berens, V., Souter, N.J., Bookpurnong Living Murray Pilot Project: Artificial inundation of Eucalyptus camaldulensis on a floodplain to improve vegetation condition. Science, Monitoring and Information Division, Department of Water, Land and Biodiversity Conservation. Winter, T.C., Relation of streams, lakes, and wetlands to groundwater flow systems. Hydrogeoly Journal

202 6 Injection of fresh river water into a saline floodplain aquifer Overview Chapter 6 examines whether injection of fresh river water into a floodplain aquifer can effectively displace saline groundwater and provide water for stressed riparian and floodplain plants and trees. The results of the numerical modelling are compared with the reported outcomes of an actual application of this salt interception measure at the study site. The river water injection field trial was conducted in September 2006 at Clark s Floodplain in the Lower Murray River in South Australia. In this study, a physically-based, fully integrated numerical model is developed using the observed data from the field trial. The aim is to further explore the impacts of fresh water injection on floodplain salinity. A number of scenarios are defined to examine four factors, namely injection rate, injected volume, injection screen depth and injection wells configuration. The results confirm that the volume of injected water is the dominant factor to achieve an effective result. Increased injected volumes can form a more extensive freshwater lens and can also mobilise more solute mass from the unsaturated zone. Moreover, it is discussed how the number and configuration of injection wells also plays an important role. It may also be seen that for given injection volumes, injecting at a lower rate and for a longer duration is a more sustainable and efficient strategy. Also, choosing an appropriate injection screen depth is vital for a successful injection outcome. The results show that injecting freshwater beneath a thick layer of saline groundwater may lead to an upward movement of this saline groundwater, which is unfavourable in terms of vegetation health. 181

203 Overall, it is found that fresh river water injection is a plausible shortterm intervention technique that can create a thin freshwater lens at the saturated/unsaturated interface and that this may enhance water availability for stressed trees. Also, freshwater injection, as a salt management measure, is able to mobilize some of the solute mass stored in the unsaturated zone. However, for long-term strategies, the injection needs to be periodically repeated unless there is overbank flooding. This chapter is accepted (in-press) in the journal Ecological Engineering (Paper 5). The manuscript was co-authored by my Principal Supervisor, Prof. Simon Beecham, and advisor, Dr. Ali Hassanli. Also, Dr. Juliette Woods (research fellow at Flinders University and DEWNR) contributed to this paper. The format of the paper has been changed to be consistent with the rest of this thesis. 182

204 Paper 5: Injection of fresh river water into a saline floodplain aquifer as a salt interception measure in a semiarid environment Published as: Alaghmand, S., Beecham, S., Woods, J.A., Jolly, I.D., Holland K.L., and Hassanli, A., Injection of fresh river water into a saline floodplain aquifer as a salt interception measure in a semi-arid environment, Ecological Engineering (Accepted for publication). Abstract: Floodplains in arid and semi-arid environments are hydrologically and ecologically essential components of the landscape. However, floodplain salinization has been highlighted as a significant risk for riparian tree health and river water quality. While various salt management measures have been developed, some of these have limited application in arid and semi-arid regions because of insufficient infrastructure and limited availability of supplemental environmental flows. Fresh river water injection into a saline floodplain aquifer can lead to environmental improvement using a relatively small amount of water and without the need for water disposal infrastructure. To explore the impacts of fresh river water injection on floodplain salinity, a physically-based, fully integrated numerical model was developed and calibrated against the observed data from a trial conducted in September 2006 at Clark s Floodplain in the Lower Murray in South Australia. It is shown that injection of an increased volume of river water leads to a larger extent of the subsequent freshwater lens. In addition, it is shown that for a given injection volume, it is more efficient to inject at a lower injection rate and for a longer duration. Also, the interface of the saturated/unsaturated zone appears to be the most effective injection screen depth. Moreover, in this case, a linear configuration of the injection wells was more effective compared to a rectangular configuration. Overall, the fresh river water injection is only able to maintain a temporary and spatially limited local freshwater lens. For long-term salt management, the river water injection needs to be periodically repeated in the absence of overbank flooding in the meantime. 183

205 6.1 Introduction Floodplains are an integral part of a river basin s landscape, supporting agriculture, recreation and industry (Berens et al., 2009a; Holland et al., 2009b; Vugteveen et al., 2006), particularly, in arid and semi-arid environments where water resources are limited. Thus, arid and semi-arid floodplains and their adjacent wetlands often have higher biodiversity value than surrounding areas, providing habitat for many aquatic and riparian species, including vegetation, birds, mammals and fish species (Doble et al., 2006). But, riverine floodplains have been gradually isolated from their parent rivers and are now amongst the world s most vulnerable and impacted ecosystems (Berens et al., 2009b; Mussared, 1997; Tockner and Stanford, 2002). In a gaining river-floodplain system, groundwater moves from a higher gradient (the adjacent highland) to a lower gradient (the river) and it moves through the floodplain before discharging to the river. If the water table in the floodplain becomes shallow then the evapotranspiration (ET) rate becomes significant and this leads to evapoconcentration (Alaghmand et al., 2014; Doble et al., 2006; Jolly et al., 1996). For example, in the Murray River in South Australia, the installation of a series of weirs, constructed to regulate the river flow, along with irrigation practices on the highlands adjacent to the floodplains, has increased recharge by a three orders of magnitude. This has caused saline groundwater in the floodplain to rise (Jolly et al., 2002). River regulation and diversions for consumptive use have altered the extent, duration and timing of overbank flooding (Holland et al., 2013). Floodplains that were once flooded three or four years out of five are often now only inundated once in 10 or 12 years, whilst some have not received any flooding for over two decades (Mussared, 1997). These factors have increased the rates of soil salinization and this has degraded environmentally significant riparian vegetation health such as river red gum (Eucalytpus camaldulensis) and black box (Eucalytpus largiflorens) (Berens et al., 2009a; Cunningham et al., 2007; Holland et al., 2009a; Jolly et al., 1996). Delivering environmental flows has been recommended as a floodplain management strategy to enhance the affected floodplains (Arthington and Pusey, 2003; Hughes and Rood, 2003). These environmental flows are provided via 184

206 releases of water from large storages (Richter and Thomas, 2007). However, this is often of limited feasibility in arid and semi-arid environments due to the large volume of water required. In such regions, recharge from rainfall is unlikely to occur, therefore the river and groundwater are the most likely available sources for water supply. If the groundwater aquifer is naturally saline, the river often has to supply the water demands including irrigation, potable water and the generation of hydroelectric power (Tockner and Stanford, 2002). Hence, the amount of water allocated for the environment is often limited. For instance, in the Murray River in South Australia only 1% of the total flow was explicitly available for environmental flows in (Berens et al., 2009a). Even if enough water is available, delivery of environmental flows is usually associated with some technical difficulties. The target area in the floodplain may be at a higher elevation than the river level and the amount available may be insufficient to fill the channel, let alone the floodplain. The distance of the floodplain from the storage is also important as transmission losses increase with distance, and may render the environmental flows ineffective by the time they reaches the stressed floodplain. For example, one of the most significant storages on the Murray River is the Hume Dam which is 1570 km upstream of the South Australian border (Berens et al., 2009a). Managed aquifer recharge (MAR) is a method to balance water supply with demands. The goal is to generate environmental improvement using small amounts of water relative to surface flooding and water extraction techniques, and without the need for water disposal infrastructure. It consists of injecting water into an aquifer during a period of excess, in order to keep it stored until required (Antoniou et al., 2013; Pyne, 1995). In MAR, excess surface water is injected into a subsurface aquifer for subsequent recovery. This can be undertaken in order to create a reservoir of fresh water by displacing the natural saline groundwater in the capillary fringe (Eastwood and Stanfield, 2001). In such situations, injection of fresh river water into the floodplain shallow saline aquifer can be implemented as a novel engineering technique for environmental flow delivery (Berens et al., 2009a). This is particularly beneficial for trees that are able to access fresh water from both surface infiltration and the groundwater such as E. camaldulensis (Mensforth et al., 1994; Thorburn et al., 1994). It would also reduce the water 185

207 losses due to evaporation (Berens et al., 2009a). Berens et al. (2009a) and Berens et al. (2009b) reported the application of river water injection into a floodplain saline aquifer as an alternative methodology for managing the health decline of riparian vegetation. However, due to technical issues including problems of aquifer clogging and breaching, and the installation and operation of the injection bores was a costly and intrusive operation. Therefore, they ceased the operation after 57 days. Thus, the small volume of injected water (4.9 ML of the allocated 10 ML) only slightly improved soil water availability in the capillary fringe. The aim of this paper is to explore the hypothesis: is fresh river water injection to a saline floodplain aquifer a suitable floodplain management option? We further explore how this can improve water availability for riparian trees. In this context we study the following questions: Is fresh river water injection a plausible mechanism to create a thin freshwater lens at the saturated/unsaturated interface zone where tree roots uptake available soil water? Can fresh river water injection mobilize some of the solute mass stored in the unsaturated zone? The impacts of factors such as injection rate, allocated water volume, pump screen depth and injection well configuration are also examined to determine how these variables influence the state of saline groundwater displacement. This displacement is studied in terms of extent, depth and duration. For this, a fully integrated, physically-based numerical model is developed and calibrated based on observed data collected during a field trial conducted at a study site. 6.2 Material and methods Field trial setup The study site, known as Site E, is in the southern part of Clark s Floodplain ( E, S), which is located downstream of Lock 4 on the Lower Murray River in South Australia and covers an area of approximately 5 km 2 (Figure 6.1). It has a semi-arid environment with a mean annual rainfall of 264 mm and potential annual evaporation of approximately mm (BOM, 186

208 2013). The 1500 ha Bookpurnong irrigation district, developed since 1964, is located adjacent to Clark s Floodplain (Telfer, 1999). Excess irrigation drainage beneath this district induces groundwater mounding, which hydraulically displaces regional saline groundwater into the floodplain alluvial aquifer and has caused seepage of saline groundwater at the edge of the river valley (Holland et al., 2013). For further details on the study site the reader is referred to Doble et al. (2006), Alaghmand et al. (2013) and Holland et al. (2013). Figure 6.1 Location of Site E on Clark s Floodplain on the Lower Murray River in South Australia (Purple dotted line shows the floodplain perimeter). 187

209 A total of five injection wells, each 200 mm in diameter, and seven observation wells, each 80 mm in diameter, were installed at Site E in April 2006 using a rotary mud method, as suggested for unconsolidated aquifers by Segalen et al. (2005). The environmental flow volume allocated for this trial was 10 ML of Murray River water. Data-loggers were fitted in all seven observation and five injection wells to record groundwater levels. Observation wells EO1, EO2, EO3 and EO5 were also monitored for conductivity and temperature. The injection trial began on 19 September 2006 and ceased on 8 November The injection rate varied between 0.8 and 1.25 l.s -1. Over the injection period a total of 4.9 ML of water was injected into the aquifer. The injection ceased on 8 November 2006 due to injection well and aquifer clogging with biological and particulate matter, resulting in the breaching of the confining clay layer approximately m away from the injection wells (Figure 6.3). For further details on the trial conducted in September 2006 at Clark s Floodplain the reader is referred to Berens et al. (2009a). Figure 6.2 shows the configuration of the injection and observation wells at Site E, while Table 6.1 summarises the specification details. Table 6.1 Specifications details of the injection and observation wells Ground Screen Drilled Reference Diameter Total Well name Purpose elevation depth date elevation (m) depth (m AHD) (m) EI1 Injection 12-Apr EI2 Injection 11-Apr EI3 Injection 11-Apr EI4 Injection 12-Apr EI5 Injection 12-Apr EO1 Observation 7-Apr EO2 Observation 8-Apr EO3 Observation 8-Apr EO4 Observation 8-Apr EO5 Observation 7-Apr EO6 Observation 9-Apr EO7 Observation 10-Apr

210 Figure 6.2 Configuration of injection (blue dots) and observation (red dots) wells. Rectangle in yellow represents the perimeter of the injection zone. Figure 6.3 Photo of the aquifer breach on 8 November 2006 next to injection well EI4 at Site E (Berens et al., 2009a). 189

211 6.2.2 Base case model A numerical model was developed which simulated the observed behaviour of a saline floodplain aquifer induced by fresh river water injection at the study site. The calibration model (hereafter referred to as the base case model) included the design specifications for the injection trial. The base case model was run from 1/09/2006 to 1/12/2006. This included injection of 10 ML water via five-point rectangular configuration scenarios (see Figure 6.2). The base case model and all the simulated scenario were based on the assumption that an aquifer breach would not occur Scenarios To explore the impacts of fresh river water injection on the saline floodplain aquifer, the base case model was modified for various different scenarios (Table 6.2). These included seven scenarios with different injection rates, volumes and screen depths. Also, two alternative injection wells configurations were tested, namely single-point and five-point linear. The study time frame was from 1/09/2006 to 1/09/2007 in 365 daily time-steps. The saline groundwater displacement induced by the defined injection scenarios were analysed in terms of extent and duration. Table 6.2 Specifications of the defined scenarios Scenario name Injection rate (l.s -1 ) Volume (ML) Screen depth (m) Pump configuration Time steps (days) A Five-point rectangular B Five-point rectangular C Five-point rectangular D Five-point rectangular E Five-point rectangular F Single-point G Five-point linear Numerical model development HydroGeoSphere Surface water-groundwater interactions were simulated using the groundwater flow model HydroGeoSphere (HGS) (Therrien et al., 2006). HGS is a physically based numerical model describing fully integrated surface and unsaturated and 190

212 saturated flows in the subsurface. HGS models flow in the unsaturated zone using the Richards equation. This is a significant advantage over models that do not explicitly consider the unsaturated zone (Brunner and Simmons, 2012). For further details on the code and a recent software review, the reader is referred to Therrien et al. (2006) and Brunner and Simmons (2012). HGS requires pre- and post-processor tools in order to handle input preparation (complex topography and grid) and visualization of the outputs. In this study, Grid Builder (McLaren, 2005) and Groundwater Modelling System (GMS) (AquaVeo, 2011) were used to generate the model grid. GMS was also used to visualize and interpret the model outputs. In this study, HGS used the control volume finite element approach to solve surface and subsurface flow and transport. Here an initial time step of 0.1 days, a maximum time step of 1 day and a maximum time step multiplier of 1.25 were used. The model solves non-linear equations for variably-saturated subsurface flow, surface flow and solute transport. To solve the non-linear equations, HGS uses the Newton-Raphson linearization method. Newton iteration parameters include Newton maximum iterations (25), Jacobian epsilon (10.0 d -5 ), Newton absolute convergence criteria (1.0 d -5 ), Newton residual convergence criteria (1.0 d -3 ) and flow solver maximum iterations (1.0 d 5 ) Geometry grid The geometric grid was based on a LiDAR Digital Elevation Model of the study site with a 10 m grid resolution and 15 non-uniform sub-layers. Figure 6.4 shows the resulting grid of Clark s Floodplain covering 238 ha from the floodplain break of slope to the Lower Murray River main channel. In this case, the length of the river bank was 5950 m and the distance from the river bank to the floodplain break of slope was between 170 m and 1850 m. The final grid consisted of 226,629 nodes and 444,840 elements. The ground elevation ranges between 9.8 m Australian Height Datum (AHD) and 38.4 m AHD Model parameters The floodplain aquifer consisted of three soil types, namely a continuous 10 m thick layer of Monoman Formation sand, overlaid by a spatially variable, 2 to 5 m thick layer of semi-confining heavy Coonambidgal Clay, and Upper Loxton Sand in the adjacent highland (Figure 6.4). Soil properties and van Genuchten function (a and n) (van Genuchten, 1980) parameters were adopted from Doble et al. 191

213 (2006) and Jolly et al. (1993) (Table 6.3). Longitudinal and transverse solute dispersivity values were estimated through model calibration. The hydraulic properties of the surface domain (river bed and floodplain corridor) were divided in the model into two categories, namely main channel (river) and floodplain. Furthermore, the vegetation coverage of the floodplain was classified as Eucalyptus trees and grass. The evapotranspiration parameter values for both categories were adopted from Hingston et al. (1997), Banks et al. (2011), Doody et al. (2009) and Verstrepen (2011). For further details on the model parameters the reader is referred to Doble et al. (2006), Doody et al. (2009) and Alaghmand et al. (2013). Figure 6.4 3D visualization of the geometric grid of the study site including soil types (Z magnification = 10). Red line in the inset map shows the perimeter of the model geometry grid. Blue and brown lines in the inset map show the location of constant first type (Dirichlet) and time-varying first-type (Dirichlet) boundary conditions. 192

214 Table 6.3 Soil parameter values of the model for the study site Model parameter Monoman Sand Soil type Coonambidgal Clay Upper Loxton Sand Units Porosity % Hydraulic conductivity m d -1 Specific storage 1.6 x x x 10-4 m -1 Evaporation limiting saturation (min) Evaporation limiting saturation (max) Longitudinal dispersivity m Transverse dispersivity m Residual water content a m -1 n Boundary and initial conditions Two types of boundary conditions were used in the model including first-type (Dirichlet) boundaries of prescribed head/concentration and second-type (Neumann) boundaries of prescribed flow/solute flux (Therrien et al., 2006). In the subsurface (porous media) domain, a constant first type (Dirichlet) boundary condition of 12 m AHD constant head was specified at the north-eastern part of the domain to represent the potentiometric head in the regional Upper Loxton Sand aquifer at the edge of the floodplain (AWE, 2013). The observed river levels for the surface domain were assigned at the river side of the model using a timevarying first-type (Dirichlet) boundary condition. In this regard, the observed water levels downstream of Lock 4 were applied to the river nodes of the model (WaterConnect, 2013). The locations of the boundary conditions are displayed in the inset map in Figure 6.4. Moreover, the solute boundary conditions were represented using a first-type (Dirichlet) constant concentration boundary condition. Hence, constant values were applied at the subsurface outer boundary nodes (representing regional groundwater flow from the Loxton aquifer with salinity of 30,000 µs.cm -1 ) and the river nodes (200 µs.cm -1 ) (Holland et al., 2013). Potential evapotranspiration and rainfall were simulated using a timevarying second-type (Neumann) boundary condition according to recorded data (BOM, 2013). Also, the five injection wells were represented in the model using recorded injection rates and durations obtained from Berens et al. (2009a). Figure 193

215 6.5 shows the recorded river stage, rainfall and ET during the trial at Site E as well as total injection rate and injected volume of river water. The initial conditions for the base case model were obtained from a steady-state flow and transport model (constant river level and ET) that represented the condition of the river-floodplain system the prior to the study period. Therefore, hydraulic head and solute concentration outputs from the initial model compared favourably with recorded data from the observation wells and the EM31 survey results undertaken at the beginning of the study period. Figure 6.6 shows the simulated groundwater salinity distribution on the 1 st day of the trial (1/09/2006). The observed groundwater salinity levels at the observation wells show a good agreement with the simulated results. (a) (b) Figure 6.5 a: Recorded river stage, daily rainfall and daily ET during the trial; b: Total injection rate and injected volume of river water Model calibration Calibration was undertaken using an iterative trial-and-error method. The hydraulic conductivity, porosity, dispersivity (longitudinal and transverse) and leaf area index were adjusted within known ranges and reasonable limits in order to achieve an acceptable match to observed data. The water table and salinity were compared to the observed data at observation wells EO1, EO3 and EO4. Moreover, the EM31 data were available to check the patterns of groundwater salinity. 194

216 Figure 6.6 Simulated groundwater salinity at the beginning of the study period (pre-trial) on 1/09/2006. Values in brackets represent the observed groundwater salinity. 6.3 Results and discussion Base case model The calibrated model was developed with the aim to reproduce the observed behaviour of the flow and solute dynamics of the floodplain aquifer over the period 1/09/2006 to 1/12/2006. Thus, the model performance in terms of groundwater dynamic was tested by visual comparison between the observed and simulated series of groundwater levels at observation wells along transect 1 (EO1, EO3 and EO4). Figure 6.7 shows a good agreement between the simulated and observed data. During the trial, the injection rates were between 0.8 l/s -1 and 1.25 l/s -1 between 19/09/2006 and 8/11/2006. This produced a small but rapid increase in groundwater level at the observation wells closest to the injection wells. For instance, along transect 1, a 0.05 m head increase was recorded at EO1 and EO3, which were located less than 25 m from the injection wells. Also, a small but rapid reduction in the groundwater level was detected when the trial was stopped 195

217 following the aquifer breach. This was not observed in the other observation wells further away. It appears recorded groundwater level at EO4 was due to river stage fluctuations rather than the injection trial. In fact, observation wells EO4 and EO7 located closer to the river banks are influenced with the river fluctuations rather than the injection trial. Figure 6.7 Simulated and observed groundwater heads at observation wells EO1, EO3 and EO4. Light green pattern represents the injection trial period. The groundwater salinity results show that the calibrated model is able to reproduce the observed dynamic in an acceptable manner (RMSE (m); EO1: 0.069, EO3: and EO4: 0.074). The simulated results are consistent with the EM31 survey which was conducted on 2 November 2006 before the aquifer breach. As a general pattern, two distinct zones were identified, namely a nearriver fresh zone (includes EO4 and EO7, with salinity less than 5,000 μs.cm -1 ) and a saline zone over the rest of the floodplain (Figure 6.8a). However, relatively fresh zones were formed in the immediate vicinity, approximately m from each injection well but this did not extend to the adjacent observation wells, with the exception of EO1. A reduction in the groundwater salinity (from 26,000 μs.cm -1 to 10,000 μs.cm -1 ) was only exhibited at EO1 which is situated less than 196

218 5 m from EI5 (Figure 6.8b). Soil water availability at the capillary fringe (depths greater than 2 m) displayed a similar behaviour (Berens et al., 2009a). (a) (b) Figure 6.8 a: Plan view of simulated groundwater salinity distribution at Site E on 1/12/2006 (time step = 92 days); b: Simulated and observed groundwater salinity at observation well EO1. Light green pattern represents the injection trial period Scenarios Seven scenarios are defined to explore different river water injection strategies (Table 6.2). All the scenarios run for 1 year from 1/09/2006 and 1/09/2007 (366 time steps). To be able to compare the numerical model results to the conducted field trial at the study site, the injection simulation commences on 19/09/2006 (time step 19) in each scenario. 197

219 Injection rate Scenarios A, B and C inject a total of 10 ML of fresh river water to the saline floodplain aquifer starting from day 19. They differ by the rate at which the water is injected. Scenario A injects over 92 days at 1.25 l.s -1, Scenario B injects over 57 days at 2 l.s -1 and Scenario C injects over 23 days at 5 l.s -1. Figure 6.9 displays the simulated groundwater head dynamics at observation wells EO1 and EO4 during scenarios A, B and C. The water table response appears to be proportional to the distance from the injection well. For instance, observation well EO1 which is located 5 m from injection well EI5 shows a rapid 0.9 m increase in scenario A, while in the same scenario the groundwater head at observation well EO4 (30m further) increases by less than 0.3 m. Due to the high hydraulic conductivity of the floodplain aquifer (Monoman Sand), the groundwater head fluctuation is almost instant. Figure 6.9a shows that higher injection rates lead to higher groundwater heads. Furthermore, it can be seen that when the injection ceases, the groundwater dynamic is a function of the river fluctuation. Thus, during the river recession from March 2007 to September 2007, a decrease in groundwater head can be observed. (a) (b) Figure 6.9 Simulated groundwater head for the injections with 1.25 l.s -1, 2 l.s -1 and 5 l.s -1 injection rates at Site E. Figure 6.10 shows the simulated groundwater salinity dynamics for scenarios A, B and C. Observation well EO1 responds strongly to the injection. However, a similar pattern is not apparent at observation well EO4 which is located further away and closer to the river. Moreover, the three scenarios can maintain a fresh water lens (at salinities less than 10,000 μs.cm -1 ) in the vicinity of the injection 198

220 wells (Figure 6.11). The formation of the fresh water lens takes 53, 34 and 18 days for scenarios A (lowest injection rate), B and C (highest injection rate), respectively. However, the three scenarios can displace the saline groundwater for the same duration. Both field and laboratory studies have found that groundwater salinity 15,000-20,000 μs.cm -1 causes river red gum death (Eamus et al., 2006). In this case, the three scenarios can maintain the groundwater salinity at less than 15,000 μs.cm -1 for approximately 165 days (Figure 6.10a). Moreover, the extent of the freshwater lens induced by the fresh river water injection can be interpreted according to Figure This shows the groundwater salinity along transect 1 at the last time step of the injection. It appears that the lateral magnitude of freshening is similar in all three scenarios, despite the different injection rates. Overall, it appears that for given fresh river water volumes, the three defined injection rates form the same freshwater lens extent. However, technically it should be considered that higher injection rates may increase the risk of aquifer breach due to increased pressure heads. Therefore, because of the same freshwater lens extents, lower injection rates seem to be more appropriate. Furthermore, scenarios A, B and C all show an increase in groundwater salinity between March 2006 and September This is attributed to the river stage recession which leads to floodplain recharge by the saline regional groundwater aquifer. (a) (b) Figure 6.10 Simulated groundwater salinity for the injection with 1.25 l.s -1, 2 l.s -1 and 5 l.s -1 injection rates at Site E. 199

221 (a) Scenario A, time step 111 days (b) Scenario B, time step 76 days (c) Scenario C, time step 42 days Figure 6.11 Simulated groundwater salinity distribution for scenarios A, B and C at time steps 111, 76 and 42 days, respectively (Z magnification = 3) Injected volume The results of scenario D are useful for exploring the impact of fresh river water injection on the saline floodplain aquifer. Figure 6.12 shows scenario D at time step 134 days which corresponds to the time by which a total of 20 ML of fresh river water has been injected into the floodplain aquifer. Comparing Figures 6.12b and 11b shows that, for a given injection rate (2 l.s -1 ), an increased volume of injected fresh river water considerably increases the extent of the freshwater lens. Moreover, Figure 6.13 shows an increase in groundwater head (up to 0.3 m) at observation well EO7 is induced by scenario D. This was not observed during scenarios A, B and C. Overall, it is shown that with a given injection rate, 200

222 delivering more injected water leads to more effective floodplain salinity mitigation. (a) Scenario D, time step 19 days (b) Scenario D, time step 134 days (c) Plan view of Site E for scenario D, the time step 134 days Figure 6.12 Simulated groundwater salinity distribution for scenario D at pre and post injection time steps (Z magnification = 3). 201

223 Figure 6.13 Simulated groundwater heads at observation well EO7 for scenarios B and D during the injections Injection screen depth To explore the impact of injection screen depth on the freshwater lens, scenarios B and E can be compared. In both of these scenarios, 10 ML of river water is injected at a rate of 2 l.s -1, but with different injection screen depths (B: 3-6 m depth and E: 3-10 m depth). Figure 6.14 shows the simulated floodplain aquifer salinity at the first (19 days) and last (74 days) time steps of the injection for scenario E. From Figures 6.14b and 6.11b, it can be seen that locating the injection screen at the interface of the saturated/unsaturated zone (in this case at a depth of 3.5 m) is more effective for forming a freshwater lens. Furthermore, it appears that deeper injection depths cannot create a freshwater lens at the interface of the saturated/unsaturated zone, and even it may expands the saline groundwater plume (See Figures 6.14a and 6.14b). In fact, injection of fresh water beneath the thick layer of saline groundwater fails to significantly mitigate the local aquifer salinity and create the freshwater lens. This configuration pushes the saline groundwater plume upward and increases the overall floodplain shallow aquifer salinity. It seems that selection of an appropriate injection screen depth is vital to obtain an efficient outcome. Otherwise, it may be detrimental for the health of the targeted vegetation. Consideration needs to be placed on the near surface (top of aquifer) groundwater displacement and the ability to create a freshwater lens to target the tree root zone, rather than the displacement of water at greater depths in the aquifer. 202

224 (a) Scenario E, time step 19 days (pre-injection) (b) Scenario E, time step 76 days Figure 6.14 Simulated groundwater salinity distribution for scenario E at pre and post injection time steps (Z magnification = 3) Injection pumps configuration In the base case model and scenarios A, B, C, D and E the five injection wells are configured as a rectangular pattern. Scenarios F and G are defined to examine the impact of the configuration of the injection wells on the outcome of the floodplain aquifer freshening. Scenario F injects the 10 ML of fresh river water via one injection well (single-point). Figures 6.15a, b and c show a plan view of the simulated groundwater salinity distribution for scenarios C, F and G at the end of the injection period. It can be seen that at the end of the injection period different patterns can be observed. Scenario F leads to a circular dispersion of the fresh water and groundwater freshening, while scenario G forms a larger freshwater lens along the straight line where the injection wells are located. Even scenario G seems to be more effective at freshening the floodplain aquifer compared to scenario C. This shows that for given injection volumes, the extent of induced saline groundwater displacement can be different depending on the injection well configuration. Depending on the topography of the floodplain and distribution of 203

225 the targeted stressed trees, different configurations can be considered to deliver the injected fresh water more efficiently. (a) Scenario F, time step 134 days (b) Scenario G, time step 42 days (c) Scenario C, time step 42 days Figure 6.15 Simulated groundwater salinity distributions for scenarios C, F and G Solute mass mobilization Due to an altered overbank flooding regime, natural saline groundwater and high rate of evapotranspiration, there is often a significant amount of solute mass stored in the unsaturated zone in arid and semi-arid floodplains. Therefore, one of the benefits or aims of any salt management measure is to mobilise some of the solute mass stored in the unsaturated zone. Figure 6.16 shows the distribution of the solute mass mobilization induced by scenarios B, C, D and G. The results of 204

226 the model show that up to 4 kg.m -3 of solute mass can be mobilised from the unsaturated zone. (a) Scenario B, time step 76 days (b) Scenario C, time step 42 days (c) Scenario D, time step 134 days (d) Scenario G, time step 42 days Figure 6.16 Simulated distribution of solute mass mobilization from the unsaturated zone for the defined scenarios It appears that scenarios B and C, despite different injection rates, mobilise almost the same amount of solute mass from the unsaturated zone, but that this is limited to the immediate vicinity of the injection wells. Moreover, injecting 20 ML of fresh river water in scenario D leads to the most effective outcome in term of solute mass mobilization from the unsaturated zone. After scenario D, scenario G shows the second most effective solute mass mobilization which is attributed to its injection well configuration. 205

227 6.4 Conclusion Riparian trees in the Lower Murray River have been declining. An aquifer injection strategy was used to deliver fresh river water to the interface of the saturated/unsaturated zone to increase water availability for stressed trees. The field trial data was used to develop and calibrate a numerical model to further explore the impacts of fresh water injection on floodplain salinity. Seven scenarios were defined to examine the effects of different factors on the efficiency of river water injection. Four factors were explored in this research, namely injection rate, injected volume, injection screen depth and injection well configuration. The results confirmed that the volume of injected water is the dominant factor to achieve an effective result. An increased injection volume can form a more extensive freshwater lens and can also mobilise more solute mass from the unsaturated zone. In addition, it was shown that for given injection volumes, injecting at lower injection rates and for longer durations produces a more sustainable and efficient strategy. Furthermore, the numerical model results showed that choosing a proper injection screen depth is vital for a successful outcome. It was shown that injecting the freshwater deep into the aquifer and beneath a thick layer of saline groundwater may lead to upward movement of saline groundwater which is unfavourable in terms of vegetation health. Moreover, it was demonstrated that the number and configuration of injection wells also plays an important role in the distribution of injected fresh water. Note that density-dependant simulation was not taken into account in this paper. This limitation for the model here should be considered in the next research. Overall, it was demonstrated that fresh river water injection is a plausible mechanism to create a thin freshwater lens at the saturated/unsaturated interface and this may improve plant and tree health. This salt management measure is able to mobilize some of the solute mass stored in the unsaturated zone but this mainly depends on the injection well configuration and total volume of injected water. However, fresh river water injection can only maintain a temporary and spatially limited local freshwater lens which can be applied as short-term management strategy for protecting the health of declining vegetation. For long-term strategies, the injection needs to be periodically repeated unless there is overbank flooding in the meantime. This research is one of the first attempts to incorporate observed data 206

228 into a physically-based, fully integrated model to explore such complex riverfloodplain interactions. 207

229 References Alaghmand, S., Beecham, S., Hassanli, A., A review of the numerical modelling of salt mobilization from groundwater-surface water interactions. Water Resources 40(3) Alaghmand, S., Beecham, S., Hassanli, A., Impacts of Vegetation Cover on Surface-Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia. Environmental Processes Accepted for publication: February Paper EWRA (in press). Antoniou, E.A., Stuyfzand, P.J., van Breukelen, B.M., Reactive transport modeling of an aquifer storage and recovery (ASR) pilot to assess long-term water quality improvements and potential solutions. Applied Geochemistry 35(0) AquaVeo, GMS: Provo, UT. Arthington, A.H., Pusey, B.J., Flow restoration and protection in Australian rivers. River Research and Applications 19(5-6) AWE, Loxton Bookpurnong SIS Atlas. Australian Water Environment: Adelaide. Banks, E.W., Brunner, P., Simmons, C.T., Vegetation controls on variably saturated processes between surface water and groundwater and their impact on the state of connection. Water Resources Research 47(11) W Berens, V., White, M., Souter, N., 2009a. Bookpurnong Living Murray Pilot Project: A trial of three floodplain water management techniques to improve vegetation condition. Department of Water, Land and Biodiversity Conservation: Adelaide. Berens, V., White, M.G., Souter, N.J., 2009b. Injection of fresh river water into a saline floodplain aquifer in an attempt to improve the condition of river red gum (Eucalyptus camaldulensis Dehnh.). Hydrological Processes BOM, Bureau of Meteorology (BOM). Brunner, P., Simmons, C.T., HydroGeoSphere: A Fully Integrated, Physically Based Hydrological Model. Ground Water 50(2) Cunningham, S.C., Read, J., Baker, P.J., Mac Nally, R., Quantitative assessment of stand condition and its relationship to physiological stress in stands of Eucalyptus camaldulensis (Myrtaceae). Australian Journal of Botany 55(7) Doble, R., Simmons, C., Jolly, I., Walker, G., Spatial relationships between vegetation cover and irrigation-induced groundwater discharge on a semi-arid floodplain, Australia. Journal of Hydrology 329(1-2)

230 Doody, T.M., Holland, K.L., Benyon, R.G., Jolly, I.D., Effect of groundwater freshening on riparian vegetation water balance. Hydrological Processes 23(24) Eamus, D., Hattin, T., Cook, P., Colvin, C., Ecohydrology: Vegetation Function, Water and Resource Management. CSIRO Publishing, Victoria. Eastwood, J.C., Stanfield, P.J., Key success factors in an ASR scheme. Quarterly Journal of Engineering Geology and Hydrogeology 34(4) Hingston, F.J., Galbraith, J.H., Dimmock, G.M., Application of the processbased model BIOMASS to Eucalyptus globules subsp. Globules plantations on ex-farmland in south Western Australia: I. Water use by trees and assessing risk of losses due to drought. Forest Ecology and Management Holland, K.L., Charles, A.H., Jolly, I.D., Overton, I.C., Gehrig, S., Simmons, C.T., 2009a. Effectiveness of artificial watering of a semi-arid saline wetland for managing riparian vegetation health. Hydrological Processes Holland, K.L., Jolly, I.D., Overton, I.C., Walker, G.R., 2009b. Analytical model of salinity risk from groundwater discharge in semi-arid, lowland floodplains. Hydrological Processes Holland, K.L., Turnadge, C.J., Nicol, J.M., Gehrig, S.L., Strawbridge, A.D., Floodplain response and recovery: comparison between natural and artificial floods, Technical Report Series No. 13/4. Goyder Institute for Water Research: Adelaide. Hughes, F.M.R., Rood, S.B., Allocation of River Flows for Restoration of Floodplain Forest Ecosystems: A Review of Approaches and Their Applicability in Europe. Environmental Management 32(1) Jolly, I.D., McEwan, K.L., Cox, J., Walker, G.R., Holland, K.L., Managing Groundwater and Surface Water for Native Terrestrial Vegetation Health in Saline Areas, CSIRO Land and Water Technical Report 23/02. CSIRO Land and Water: Canberra, Australia. Jolly, I.D., Walker, G.R., Hollingworth, I.D., Eldridge, S.R., Thorburn, P.J., McEwan, K.L., Hatton, T.J., The causes of decline in eucalypt communities and possible ameliorative approaches, In: walker, G.R., Jolly, I.D., Jarwal, S.D. (Eds.), Salt and Water Movement in the Chowilla Floodplain. CSIRO Division of Water Resources: Canberra, Australia. Jolly, I.D., Walker, G.R., Thorburn, P.J., Salt accumulation in semi-arid floodplain soils with implications for forest health. Journal of Hydrology 150(2-4) McLaren, R.G., Grid Builder: A pre-processor for 2-D, triangular element, finite-element programs. Groundwater Simulations Group, University of Waterloo: Waterloo, Ontario. Mensforth, L.J., Thorburn, P.J., Tyerman, S.D., Walker, G.R., Sources of water used by riparian Eucalyptus camaldulensis overlying highly saline groundwater. Oecologia 100(1-2)

231 Mussared, D., Living on floodplains. The Cooperative Research Centre for Freshwater Ecology and The Murray Darling Basin Commission: Canberra. Pyne, R.D.G., Groundwater Recharge and Wells. Lewis Publishers, Boca Raton. Richter, B.D., Thomas, G.A., Restoring environmental flows by modifying dam operations. Ecology and Society 12(1). Segalen, A.S., Pavelic, P., Dillon, P.J., Review of Drilling, Completion and Remediation Methods for ASR Wells in Unconsolidated Aquifers, Technical Report No. 04/05. CSIRO Land and Water. CSIRO: Canberra. Telfer, A., Overton, I.C.,, The impact of irrigation on floodplain vegetation health adjacent the River Murray, In: Rutherford, I.D., Bartley, R. (Ed.), The Second Australian Stream Management Conference: Adelaide. Therrien, R., McLaren, R.G., Sudicky, E.A., Panday, S.M., HydroGeoSphere: A Three-Dimensional Numerical Model Describing Fully- Integrated Subsurface and Surface Flow and Solute Transport. Groundwater Simulations Group, University of Waterloo: Waterloo, Canada. Thorburn, P.J., Mensforth, L.J., Walker, G.R., Reliance of creek-side river red gums on creek water. Australian Journal of Marine and Freshwater Research 45(8) Tockner, K., Stanford, J.A., Riverine flood plains: Present state and future trends. Environmental Conservation 29(3) van Genuchten, M.T., A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Science Society of America Journal 44(5) Verstrepen, L., Evaluating rainwater harvesting on watershed level in the semi-arid zone of Chile, Bioscience Engineering. Universiteit Gent: Gent, p Vugteveen, P., Leuven, R.S.E.W., Huijbregts, M.A.J., Lenders, H.J.R., Redefinition and elaboration of river ecosystem health: Perspective for river management. Hydrobiologia 565(1 SPEC. ISS.) WaterConnect, River Murray Water Data 210

232 7 Impacts of vegetation cover on SW-GW flows and solute interactions Overview Chapter 7 explores the relationship between floodplain vegetation cover and flow and solute dynamics in terms of SW-GW interactions in a semi-arid floodplain. It is well-understood that evapotranspiration (ET) is one of the main processes in shallow aquifers, particularly in arid and semi-arid regions where overbank flows and rainfall recharge are unlikely to occur. To quantify such impacts on SW-GW interaction, three scenarios are defined, namely, current vegetation cover, deeprooted vegetation cover and shallow-rooted vegetation cover which are differentiated in the numerical model by modifying vegetation distribution and ET properties for each vegetation type. The fully integrated, physically-based model that was developed earlier is used to evaluate the defined scenarios. The deep-rooted scenario has deeper root depths and a higher leaf area index (LAI) resulting in more water uptake, which maintains a deeper water table in comparison with the shallowrooted vegetation. This is consistent with the modelling results. Moreover, it appears that a major portion of the groundwater discharge via ET is due to the transpiration process. This is due to the groundwater level being maintained below the evaporation depth in the deep-rooted scenario. On the other hand, shallow-rooted vegetation cover leads to higher evaporation rates and less transpiration because a shallower water table makes more water available for evaporation from the soil. In terms of floodplain salinity, it is shown that shallow-rooted vegetation forms a relatively more saline floodplain aquifer. This is because the shallower water table amplifies the solute concentration processes and increases 211

233 flux from the regional saline groundwater. Overall, it is shown that ET can influence both the SW-GW interaction and floodplain salinity levels. The outcome of this work was published in the journal Environmental Processes (Paper 6). The manuscript was co-authored by my Principal Supervisor, Prof. Simon Beecham, and advisor, Dr. Ali Hassanli. The format of the paper has been changed to be consistent with the rest of this thesis. 212

234 Paper 6: Impacts of Vegetation Cover on Surface- Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia Published as: Alaghmand, S., Beecham, S. and Hassanli, A. (2014), Impacts of Vegetation Cover on Surface-Groundwater Flows and Solute Interactions in a Semi-Arid Saline Floodplain: A Case Study of the Lower Murray River, Australia, Environmental Processes, Springer, 1, pp Abstract: Despite many studies on floodplain vegetation, there is limited quantitative understanding of the role of vegetation in surface water (SW) and groundwater (GW) interactions through processes such as evapotranspiration. Moreover, most of the investigations that have been undertaken on SW-GW interactions consider 1D or 2D model set-ups. In addition, most of the modelling studies in this research area have only included water but not solute transport. This paper presents the results of a study on the potential impacts of vegetation cover on the interaction of a river and a saline semi-arid floodplain aquifer using a 3D physically-based fully integrated numerical model. In this regard, three scenarios were defined, which included: current vegetation cover (calibration model), deep-rooted vegetation cover and shallow-rooted vegetation cover. Clark s Floodplain, located on the Lower Murray River in South Australia was selected as the study site. The results show that deep-rooted vegetation cover may maintain relatively deeper groundwater levels and a less saline floodplain aquifer. Also, it is shown that in the shallow-rooted scenario, most of the ET component belongs to the evaporation process due to shallower groundwater. On the other hand, the deep-rooted model includes groundwater uptake largely via a transpiration process and consequently keeps the groundwater levels below the evaporation depth. Overall, in semi-arid areas, vegetation cover type can have significant impacts on the flow and solute interaction dynamics of a river and a floodplain aquifer due to the influence of ET as a dominant hydrological driver. 213

235 7.1 Introduction The interaction between surface water (SW) and groundwater (GW) is strongly controlled by the relative heads and these can vary significantly over a period of time (Rosenberry and Winter, 1997). For instance, changes in SW-GW interactions will occur when there are modifications to the native vegetation due to dry-land agriculture, irrigation, forestry, urban development (Allison et al., 1990; Doble et al., 2006). For instance, the processes leading to floodplain salinization after the clearance of native vegetation for agricultural practices is shown in Figure 7.1. Trees, particularly those with deep roots, behave like groundwater pumps, and play a key role in the catchment water balance (Banks et al., 2011; Butler et al., 2007; Loheide et al., 2005). On the other hand, evapotranspiration (ET) may create an unsaturated storage zone for salt in some areas of the floodplain, particularly where deep rooted vegetation types exist, or for certain times or seasons of the year. During overbank flow and/or extreme rainfall events, these unsaturated zones containing the stored salt can become saturated which may cause salt leaching and groundwater salinization. This shows the importance of ET on the dynamics of flow and solute in a river-floodplain interaction in arid/semi-arid areas. Evans (2011) concluded that groundwater under a floodplain is often more saline than the regional (input) groundwater. This infers that there is a salt concentration process operating under the floodplain. Holland et al. (2009) showed that salinization associated with groundwater discharge via ET is the principal process influencing floodplain vegetation health, particularly in arid and semi-arid regions. For instance, in the Lower Murray River in South Australia, there are natural inflows of saline regional groundwater to the floodplains. The raised groundwater level beneath the floodplain has led to increased rates of groundwater evapotranspiration. Because the groundwater is naturally saline, the increased ET results in floodplain salinization, which consequently affects the health of floodplain vegetation. In fact, groundwater flow into the floodplain is discharged mainly as ET when the water table is within the evapotranspiration extinction depth (Holland et al., 2009). Doble et al. (2006) demonstrated that long-term patterns of net groundwater discharge are dependent on vegetation distribution, elevation, soil type and river geometry. Bornman et al. (2004) showed that the distribution and health of vegetation in a floodplain 214

236 depends on the depth to the water table and the salinity of the groundwater. Also, anthropogenic changes to flooding regimes in highly variable arid catchments have a critical effect on floodplain vegetation (Alexander and Dunton, 2006; Capon, 2005; Mensforth and Walker, 1996). Indeed, the total exchange flux between a river and the adjacent floodplain aquifer includes the following components: (1) natural exchange flux due to river stage fluctuations; (2) exchange flux due to groundwater extraction/injection; (3) exchange flux due to a change in recharge rates (e.g., change in land use); and (4) exchange flux due to changes in ET patterns. It seems that ET is a significant mechanism in shallow aquifers, particularly in arid and semi-arid regions where overbank flows and rainfall recharge are unlikely to occur (Rassam, 2002; Rassam, 2011). Figure 7.1 Schematic diagram of processes leading to floodplain salinization after the clearance of native vegetation for agricultural practices (adopted from Leblanc et al.,(2012)) Despite long-term studies on floodplain vegetation, there is limited quantitative understanding of the role of vegetation (i.e., ET) on the SW-GW interactions 215

237 (Alaghmand et al., 2013b). It is also unclear how land clearance or revegetation affects the dynamic of flow and solute (Banks et al., 2011). Moreover, most of the investigations in the context of SW-GW interactions consider 1D or 2D model set-ups. To establish a more realistic representation of the natural environment, 3D modelling is important. Some examples are the spatial distribution of salt accumulated in a floodplain/wetland, the impact of variable vegetation cover on SW-GW interactions and ET distribution, and the state of SW and GW connections along a river induced by ET, pumping, flooding, and other factors (Banks et al., 2011). Banks et al. (2011) studied the impacts of floodplain vegetation cover on the state of connection of SW and GW. They suggested that in addition to the well-known influences of physical variables such as hydraulic conductivity and topography, the effects of vegetation need to be carefully considered when investigating SW-GW interactions. They recommended further work to be carried out in 3D to explore the effects of ET on river and floodplain interactions as function of vegetation cover. Most other modelling studies in this research area have only included water but not solute dynamics. This study aims to explore the following hypothesis: can vegetation cover significantly influence the dynamics of flow and solute in the context of a river and a semi-arid saline floodplain interaction? This is tested in this study through three scenarios, i.e.,: current vegetation (mix of deep rooted and shallow rooted vegetation); coverage by only deep rooted vegetation types such as Eucalyptus trees; and coverage by only shallow rooted vegetation types such as grass. The current vegetation scenario is considered as the base case scenario and is developed and calibrated using observed data. The other two scenarios are theoretically developed and compared with the base scenario. The impacts of vegetation cover on solute and water balances and the state of connection of SW- GW are investigated using a 3D fully-integrated numerical model. 7.2 Material and methods A total of three scenarios were defined to investigate the impacts of floodplain vegetation cover on a river and a saline semi-arid floodplain aquifer interaction. In fact, the defined scenarios are differentiated by modifying vegetation distribution and ET properties of each vegetation type (0.5m root depth value is used for grass 216

238 and 5m root depth for Eucalyptus coverage). It is worth noting that floodplain groundwater at the study site is influenced by groundwater extraction through a series of production wells that form part of the Bookpurnong Salt Interception Scheme (SIS). These wells aim to alter hydraulic gradients and intercept the movement of saline groundwater from the highlands to the alluvium and the river. Two of the SIS production wells (32F and 34F) are located in the study site and are included in the model, as shown in Figure Governing equations The HydroGeoSphere (HGS) model provides a rigorous simulation capability that combines fully-integrated hydrologic/water quality/subsurface and transport capabilities with a well-tested set of user interface tools (Therrien et al., 2010). HGS requires pre- and post-processor tools in order to handle input preparation and visualization of the outputs. In this study, the Groundwater Modelling System (GMS) (AquaVeo, 2011) was used as a pre-processor to generate the input grid domain and as a post-processor to visualize the model results. ET is calculated as a combination of transpiration and evaporation. Transpiration from vegetation occurs within the root zone of the subsurface and is a function of the leaf area index (LAI), nodal water (moisture) content (θ) and a root distribution function (RDF) over a prescribed extinction depth (Alaghmand et al., 2013a). Water content is simulated as saturation because it is more stable and always varies between 0 and 1, while in reality moisture content varies from 0 to a value equal to the porosity. The rate of transpiration (Tp) is estimated using the following relationships (Kristensen and Jensen, 1975): T p = f 1 (LAI) f 2 (θ) RDF [E p E can ] (1) where Ep is the reference potential evapotranspiration which may be derived from pan measurements or computed from vegetation and climatic factors such as temperature and humidity, and Ecan is the tree canopy evaporation. Ep can also be described as the amount of water that would be removed through ET if the water table was at the ground surface. The value and description of Ep has followed the notation and conceptualization of Therrien et al. (2010) and Kristensen and Jensen (1975). The vegetation function (f1) correlates the transpiration (Tp) with the leaf 217

239 area index (LAI) in a linear fashion and the moisture content (θ) function (f2) correlates Tp with the moisture state at the roots. The root zone distribution function (RDF) is defined by the relationship: RDF = c2 c1 rf(z)dz Lr 0 rf(z)dz (2) where C1 and C2 are dimensionless fitting parameters, Lr is the effective root length, z is the depth coordinate from the soil surface [L] and rf(z) is the root extraction function, which typically varies logarithmically with depth. Below the wilting point moisture content, transpiration is 0; transpiration then increases to a maximum at the field capacity moisture content. This maximum is maintained up to the oxic moisture content, beyond which the transpiration decreases to 0 at the anoxic moisture content. When available moisture is larger than the anoxic moisture content, the roots become inactive due to lack of aeration (Therrien et al., 2010). In HGS, evaporation from the soil surface and subsurface soil layers is a function of nodal water content and an evaporation distribution function (EDF) over a prescribed extinction depth. The model assumes that evaporation (Es) occurs along with transpiration, resulting from energy that penetrates the vegetation cover and is expressed as (Therrien et al., 2010): E s = α (E p E can ) [1 f 1 (LAI)] EDF (3) where α is a wetness factor which depends on the moisture content at the end of the energy-limiting stage and below which evaporation is 0. For further details on the code the reader is referred to Therrien et al. (2010) Study site Clark s Floodplain is located on the Lower Murray River in South Australia (34 21'S, 'E) (Figure 7.2). The climate in this region is semi-arid with mild winters and long hot summers. Annual potential evaporation (1900 mm) is over seven times the average annual rainfall (251 mm). Annual rainfall is highly variable, with Bureau of Meteorology records showing annual rainfall between 86.6 and mm since Annual rainfall was average or below average 218

240 over the study period (165.8 mm in 2006 and mm in 2007). The Lower Murray River floodplain is characterised by a flat, wide, meandering river within a deep river valley, excised during the Pleistocene period (Twidale, 1978). The hydrogeology of Clark s Floodplain is typical of the eastern part of the Lower Murray River (Jarwal, 1996). In terms of soils, Coonambidgal Clay, ranging from 2 to 7 m thick, covers a Monoman Formation (sand) on the floodplain. Also, Upper Loxton Sand exists on the adjacent highland. Groundwater salinity in the Loxton Sands and Monoman Formation is in excess of 30,000 mg L -1, while irrigation recharge salinity is typically 5,000 mg L -1 (Doble et al., 2006). Two SIS production wells are located in the study site. They pump the saline groundwater at a rate of 2-3 l/s (Figure 7.2). These were in operation during the study period except for the period from November 2006 until May 2007 due to a fault in the disposal pipeline. Figure 7.2 shows the configuration of the nine groundwater observation wells at the study site. Six of these are located along two transects dissecting the floodplain laterally; B1, B2 and B3 on transect A-A' and B4, B5 and B6 on transect B-B'. In addition, SIS observation wells (31F, 33F and 35F) are located at the mid-point between the SIS production wells. The observation wells are designed to monitor both groundwater level and salinity. In addition, observed water levels and flows of the Murray River at the study site were obtained from the Lock 4 water level station just upstream of the study site, which has continuous data from 1927 (ID: A ) (WaterConnect, 2012) Numerical model set-up Available LiDAR data was used to generate the 10 m resolution Digital Elevation Model (DEM) of the study site. The three dimensional geometry grid of the study site consisted of 15 sub-layers including finer grids at the top of the model. The final geometric grid contained 104,408 nodes that formed elements. Part of Clark s floodplain from the floodplain slope break to the Murray River main channel is included in the geometric grid. This included two SIS production wells (32F and 34F) and nine observation wells. In this case, the length of the river bank was 570 m and the distance from the river bank to the SIS well varied between 480 m and 650 m. The heterogeneous model domain consisted of three soil layers and was constructed according to drill log data. The 10 m thick Monoman 219

241 Formation Sand was overlaid by spatially variable semi-confining heavy Coonambidgal Clay and also Upper Loxton Sand at the highland (Figure 7.3b). The properties of the soil and unsaturated van Genuchten function parameters (van Genuchten, 1980) are adopted from Jolly et al. (1993), Doble et al. (2006) and Alaghmand et al. (2013a) (Table 7.1). The initial model was a transient model set up for 10,000 days to create the equilibrium initial conditions for the study period (1/1/2006 to 1/09/2010). It used an initial time step of 0.1 days, a maximum time step of 1 day and a maximum time step multiplier of The initial conditions of the model were determined numerically from a steady state model run under current vegetation cover. The generated initial model was verified using the recorded groundwater heads and salinity at the beginning of the study period (January 2006), which were adopted from Berens et al. (2009). Figure 7.2 Configuration of production wells (in red) and observation wells (in green) at the Clark s Floodplain. The inset map shows the location of the study site in Australia (red circle). Boundary conditions were defined for both the surface and sub-surface domains (Figure 7.3a). A constant first type (Dirichlet) boundary condition was specified at the north-eastern part of the floodplain to represent the 12 m AHD (Australian Height Datum) groundwater head which was adopted from AWE potentiometric contours (AWE, 2013). In addition, at the river boundary of the model, a time- 220

242 varying Dirichlet condition was specified which was based upon river level observations from downstream of Lock 4. Also, the observed groundwater concentration at the observation wells in the floodplain and river were characterized by the solute boundary conditions using first-type (Dirichlet) or constant concentration boundary conditions. The salinity for the floodplain groundwater was 30,000 mg L -1 (TDS) and river water 200 mg L -1 (TDS). Hence, constant values were applied at the porous media boundary (representing the regional saline aquifer) and at the river nodes. In addition, ET and rainfall were simulated for the entire model surface domain using the time-varying second-type (Neumann) boundary condition. In fact, ET was dynamically simulated as a combination of evaporation (equation 3) and transpiration (equation 1) processes by removing water from all model cells of the surface and subsurface flow domains within the defined zone of the evaporation and root extinction depths. To simulate different vegetation cover, the transpiration process was manipulated by changing the root extinction depth and LAI. The daily reference potential evapotranspiration (Ep) rate (in equations 1 and 3) and rainfall were based upon the recorded daily values at Loxton station (ID: ) (BOM, 2013). The parameter values for the ET components of the model are adopted from Doody et al. (2009), Hingston et al. (1997), Banks et al. (2011), Verstrepen (2011) and Alaghmand et al. (2013a) (Table 7.2). Table 7.1 Soil parameter values of the model for the study site Value Model parameter Upper Monoman Coonambidgal Units Loxton Sand Clay Sand Porosity % Hydraulic conductivity m d -1 Specific storage 1.6 x x x 10-3 m -1 Residual water content Evaporation limiting saturation (min) 0.25 Evaporation limiting saturation (max) 0.9 Longitudinal dispersivity m Transverse dispersivity m Van Genuchten alpha parameter m -1 Van Genuchten beta parameter

243 Table 7.2 ET parameter values of the model for the study site Model parameter Value Units Eucalyptus Grass Tree canopy evaporation 4.5 x x 10-4 m Evaporation extinction depth defined by quadratic decay Evaporation distribution function 1 1 m Transpiration extinction depth defined by quadratic decay Root distribution function m Leaf area index m 2 m -2 Transpiration fitting parameter c Transpiration fitting parameter c Transpiration fitting parameter c Transpiration limiting saturation (at wilting point) Transpiration limiting saturation (at field capacity) Transpiration limiting saturation (at oxic limit) Transpiration limiting saturation (at anoxic limit) Figure 7.3 a: Configuration of the model boundary condition (model perimeter is shown in red dotted line), b: Configuration of the vegetation and soil layers of Clark s Floodplain along transect B-B' (Z magnification= 3). Observation wells are shown in red columns. 222

244 7.2.4 Model calibration Observed groundwater levels and salinity at the six observation wells were used as calibration criteria during coupled flow-and-transport calibration of the model. Calibration of the model was conducted manually with more consideration given to sensitive parameters such as soil hydraulic conductivity, porosity and transverse and longitudinal dispersivity. Two different approaches were used to calibrate the model in terms of flow and solute dynamics. The model performance for flow dynamics was tested both quantitatively and qualitatively. The former was performed using goodness-of-fit parameters which produced averages of 0.88 and 0.08 (m) for R 2 and RMSE, respectively. Also, visual comparison between the observed and simulated series of groundwater levels at the observation wells showed that the calibrated model was able to reproduce the SW-GW interaction processes in an acceptable manner. On the other hand, due to the difficulty associated with the quantification of the solute transport model parameters and lack of accurate estimations of the thickness, hydraulic conductivity and porosity of the aquifer, the solute dynamic was calibrated based on the observed concentration patterns. Hence, the modelled solute concentration distributions were compared visually to electromagnetic survey results reported by Berens et al. (2009). For instance, the EM31 survey in November 2007 displays a distinct zone of low conductivity along the eastern margin abutting the river channel (Berens et al., 2009). A detailed description of the calibration process can be found in Alaghmand et al. (2013a). 7.3 Results and discussion To investigate the impacts of vegetation cover on the dynamics of flow and solute, three scenarios were defined and modelled. The dynamics of flow in the defined scenarios are discussed based on ET, evaporation, bank recharge (flux from the river to the floodplain aquifer) and GW heads. Figure 7.4 shows the total amount of water removed from the floodplain aquifer through ET and evaporation during the study period. The seasonal trends in both are obvious. However, among the simulated scenarios, the deep-rooted model shows the highest amount of ET. This is due to the deeper root depths and LAI values which were assigned for the deeprooted model. Clearly, vegetation with deeper roots and higher canopies use larger amounts of water. Doble et al. (2006) suggested mm/year for the 223

245 eucalyptus (red gum) and 0-40 mm/year for the grassland in this area. The results show that ET is more pronounced during the summer period compared to winter, as both transpiration and evaporation occurs with higher rates in summer. Figure 7.4b shows the amount of water removed from the system only through evaporation (with no transpiration). Shallow-rooted floodplain vegetation shows increased losses via evaporation. Considering the assigned evaporation depth (1 m) and transpiration depth (0.5 m), most of the ET component in this scenario belongs to the evaporation process, as the groundwater depth was unlikely to be less than 0.5 m. Furthermore, in the shallow-rooted vegetation cover, the water table is shallower compared to the deep-rooted vegetation. Hence, more groundwater is exposed to be evaporated. On the other hand, the deep-rooted model uptakes the groundwater via a transpiration process and consequently keeps groundwater levels below the evaporation depth (1 m). Therefore, Figure 7.4b shows the lowest evaporation for this scenario. (a) (b) Figure 7.4 ET (a) and evaporation only (b) during the study period for the defined scenarios Figure 7.5 illustrates the impacts of floodplain vegetation cover on the water flow exchange between the river and the floodplain aquifer. This shows the cumulative bank recharge during the study period for the defined scenarios. In fact, it represents the amount of water moved from the river (surface domain) to the floodplain aquifer (sub-surface domain). As expected, more water moves to the floodplain aquifer from the river in the deep-rooted model. In other words, the floodplain aquifer with deep-rooted vegetation cover consumes more water through ET compared with shallow-rooted vegetation. Also, when a floodplain is 224