GROUNDWATER-SEAWATER INTERACTIONS7944/...TRITA-LWR PhD Thesis 1019 ISSN 1650-8602 ISRN KTH/LWR/PHD 1019 ISBN 91-7178-027-0 GROUNDWATER-SEAWATER INTERACTIONS: SEAWATER INTRUSION, SUBMARINE

TRITA-LWR PhD Thesis 1019

ISSN 1650-8602

ISRN KTH/LWR/PHD 1019

ISBN 91-7178-027-0

GROUNDWATER-SEAWATER INTERACTIONS:

SEAWATER INTRUSION, SUBMARINE GROUNDWATER DISCHARGE AND TEMPORAL

VARIABILITY AND RANDOMNESS EFFECTS

Carmen Prieto

April 2005

Carmen Prieto TRITA-LWR PhD 1019

ii

Groundwater-Seawater Interactions

iii

“We are what we repeatedly do. Excellence then, is not an act, but a habit.”

Aristotle


iv


v

ACKNOWLEDGEMENTS

I gratefully acknowledge financial support from the Swedish Research Council (VR) and the EU project WASSER (Utilization of groundwater desalination and wastewater reuse in the water supply of seasonally-stressed regions) by the European Commission's Directorate General for Research (DGXII) under contract ENV4-CT97-0459.

Moreover, I am deeply indebted to Georgia Destouni, my supervisor, for giving me the opportu-nity to work with very interesting research projects and for all her support, guidance and help throughout this work. I am very grateful to Vladimir Cvetkovic for accepting me as a graduate student at KTH. Many thanks also to Sten Berglund and Hrvoje Gotovac for valuable aid at the beginning of this work, and to all the WASSER partners and the international SCOR/LOICZ working group 112 on submarine groundwater discharge for facilitating fruitful discussions and collaboration.

Thanks to all colleagues at the Department of Land and Water Resources Engineering, especially those at the former Division of Water Resources Engineering for creating a nice and friendly atmosphere in the everyday life, and to Aira Saarelainen and Monika Löwen for assisting me with all kinds of practical matters.

I also want to express my gratitude to all friends that shared with me the good and bad of these years, for their fellowship and encouragement. Furthermore, the support and help of my family has been essential in completing this thesis. In particular, I wish to thank my mother, Oma Monika and my sister for coming to Sweden just to help us at home with childcare, to my chil-dren for being a source of motivation and inspiration, and to my wonderful husband Frank, for his generosity, unconditional support and valuable help.


vi


vii

TABLE OF CONTENT

List of appended papers .......................................................................................................... ix

Abstract...................................................................................................................................... 1

1. Introduction ........................................................................................................................... 1

2. General problem description................................................................................................ 4

3. Seawater intrusion simulations ............................................................................................ 8

3.1. Deterministic long-term trend and seasonal variability analysis ....................................................... 8

3.2. Stochastic variability analysis................................................................................................... 9

4. SGD simulations ................................................................................................................. 10

4.1. Steady-state analysis with constant sea level boundary condition .................................................... 10

4.2. Tidal oscillation analysis....................................................................................................... 10

5. Results ................................................................................................................................. 10

5.1. Seawater intrusion simulations............................................................................................... 10

5.1.1. Deterministic long-term trend and seasonal variability analysis ................................. 10

5.1.2. Stochastic variability analysis .................................................................................... 11

5.1.3. Coastal groundwater management investigations...................................................... 13

5.2. SGD simulations................................................................................................................ 14

5.2.1. Steady-state analysis with constant sea level boundary condition .............................. 14

5.2.2. Tidal oscillation analysis ........................................................................................... 15

5.2.3. Comparative SGD analysis ....................................................................................... 16

6. General discussion and conclusions.................................................................................. 18

7. References ........................................................................................................................... 19


viii


ix

LIST OF APPENDED PAPERS

This thesis comprises a summary and the following appended papers, referred to by Roman numeral in the text:

I. Prieto, C., Destouni, G. and Schwarz, J., (2001). Seawater intrusion in coastal aquifers: Effects of seasonal variations in extraction and recharge rates. In: Ouazar, D. and Cheng, A.H.-D. (Eds.), SWICA M3 Cyber Proceedings of First International Conference on Saltwater Intrusion and Coastal Aquifers - Monitoring, Modeling and Management, April 23-25, 2001, Essaouira, Morocco. Available online at http://www.olemiss.edu/sciencenet/saltnet-2003/page9.html

II. Prieto, C., Destouni, G. and Kotronarou, A. The influence of temporal hydrological random-ness on seawater intrusion in coastal aquifers (J. Hydrol., 2005, in review).

III. Koussis, A.D., Kotronarou, A., Destouni, G. and Prieto, C., (2003). Intensive groundwater development in coastal zones and small islands (Ch. 6). In: Llamas, R. and Custodio, E. (Eds.), Intensive Use of Groundwater: Challenges and Opportunities, Balkema, The Netherlands, 478 p.

IV. Destouni, G. and Prieto, C., (2003). On the possibility for generic modeling of submarine

groundwater discharge, Biogeochemistry, 66:171-186.∗

V. Prieto, C. and Destouni, G. Quantifying hydrological and tidal influences on groundwater discharges into coastal waters (in revision for publication in Water Resour. Res., 2005).

VI. Destouni, G., Hannerz, F., Jarsjö, J., Prieto, C. and Shibuo, Y. Resolving the diverse pathways of freshwater and pollutant inputs to coastal waters (to be submitted to an international journal, 2005).

For Papers I, II and V, I had the lead and made the major contributions in developing the model-ling platform and setting up the simulations, producing and interpreting results and writing the papers. Georgia Destouni made contributions to these papers with regard to the simulation de-sign, results interpretation and text formulation. Moreover, Joshua Schwarz for Paper I and A-nastasia Kotronarou for Paper II provided site-specific data and help with the conceptualizations of the case studies.

My contribution to Paper III was to produce the necessary simulation results and uncertainty quantification with regard to the core problem of groundwater dynamics; the lead author Antonis D. Koussis had the idea and main responsibility for writing the paper with different contributions from Anastasia Kotronarou, Georgia Destouni and me.

Regarding Paper IV, Georgia Destouni had the idea of investigating submarine groundwater discharge (SGD) and the lead responsibility for writing the paper. I made the groundwater dy-namics simulations and graphic presentation of results, and we together designed the simulations and analyzed the results.

The last paper VI represents a synthesis of different contributions, which are combined here for resolving some open SGD issues. The authors are listed alphabetically in order to reflect that the weight of their different contributions is approximately equal. My main contribution to this paper is the quantitative framework for hydrological SGD estimates, mainly reported in Papers IV and

∗ Reproduced with kind permission of Springer Science and Business Media.


x

V, and together with Georgia Destouni the comparative analysis of SGD estimates in different cases over the world.


1

ABSTRACT

Fresh groundwater quality and availability in coastal areas is affected by seawater intrusion into coastal aquifers, and coastal water quality and ecosystem status may be significantly affected by groundwater pollutants that are transported into coastal waters by submarine groundwater dis-charge (SGD). This thesis uses an overall regional perspective for investigating: i) seawater intru-sion and its possible control in sustainable coastal groundwater management; ii) SGD and its relevant quantification as one interacting part among the diverse main regional pathways of freshwater and tracer/pollutant inputs from land to sea; and iii) the integrated system functioning of both i) and ii) as main components of the same coastal groundwater system.

Results show that intensive pumping rates may be maintained for a long time before major re-gional seawater intrusion problems are recognized by too high salinities in pumped groundwater. After such late recognition, pumping wells are no longer useful and a common strategy of mov-ing groundwater pumping further upstream from the coast only increases the extent of the salt-water intrusion zone into the aquifer. An alternative strategy may be to control seawater intrusion through artificial groundwater recharge, for instance by sufficiently treated wastewater, which may considerably reduce long-term trends of salinity increase in pumped groundwater, even for small artificial recharge rates compared to pumping rates. In general, account for natural spatial-temporal variability and randomness may be essential for relevant prediction of groundwater dynamics for management purposes. Spatial and temporal randomness effects, however, may not be additive, but rather largely overlapping, with either spatial or temporal randomness being the dominating part that must be accounted for in predictive groundwater dynamics calculations. Aquifer depth is identified as an important control parameter in this context, yielding much greater temporal randomness effects in shallow than in deep aquifers.

Combined simulation results suggest a simple, approximately linear regional relationship between total SGD and its hydrologically determined freshwater component. Tidal oscillation may signifi-cantly affect such linear dependence of steady-state SGD, but primarily for low SGD conditions. High SGD appears to depend mainly on a dominant freshwater component, which effectively counteracts density-driven flow of seawater into the aquifer and thus decreases also effects of sea-level oscillation on the seawater component of total SGD. Comparative analysis between different SGD estimation methods in different reported high-SGD regions of the world indicates possible anomalously large regional SGD estimation from tracer concentrations in coastal waters, by confusing different main pathways of groundwater flow and pollutant inputs to the sea.

Key words: Seawater intrusion; Submarine groundwater discharge; Coastal groundwater man-agement; Coastal zone management; Temporal variability; Temporal randomness.

1. INTRODUCTION

Freshwater quality and availability is one of the most critical environmental and sustain-ability issues of the twenty-first century (UNEP, 2002). Of all sources of freshwater on the Earth, groundwater constitutes over 90% of the world's readily available freshwa-ter resources (Boswinkel, 2000) with remain-ing 10% in lakes, reservoirs, rivers and wet-lands. Groundwater is also widely used, for instance, for drinking water supply and irriga-tion in food production (Zekster and Everett, 2004). However, groundwater is not only a valuable resource for water supply, but also a

vital component of the global water cycle and the environment. As such, groundwater provides water to rivers, lakes, ponds and wetlands helping to maintain water levels and sustain dependent ecosystems. Moreover, some field investigations indicate groundwa-ter as a surprisingly important source of water and solute input to coastal waters (Lewis, 1987; Moore, 1996; Kim et al., 2003). According to Church (1996), these scientific findings challenge our understanding of coastal and oceanic chemical mass balance and ecosystem functioning.


2

On the one hand, the seaward flow of fresh groundwater to coastal waters may carry land-generated pollutants, which constitute a serious threat to coastal ecosystems, in addi-tion to limit the available fresh groundwater resources. On the other hand, the quality and availability of these fresh groundwater re-sources in coastal areas are also threatened by seawater intrusion from the seaside (Bear et al., 1999; Paper III). Seawater intrusion is a natural process, by which seawater displaces and mixes with the fresh groundwater in coastal aquifers due to the density difference existing between waters of different salinities. As a consequence, a salinity transition zone is formed with salinity ranging from that of seawater at the land-sea interface to that of freshwater at some distance into the aquifer (Fig. 1). Within the transition zone, at least some of the intruding seawater re-circulates back into the sea, following flow patterns that are determined by the freshwater flow, density differences, thermal convection, tidal oscillations and wave set-up.

The shape and position of the transition zone depend on many factors, but at steady-state the transition zone is stationary indicating a dynamic equilibrium between the natural fresh groundwater flow towards the sea and the re-circulating seawater inflow. Any dis-turbance of this water flow balance in the aquifer will change the position and shape of the transition zone. In heavily exploited coastal aquifers, where groundwater pumping consistently exceeds recharge, the groundwa-ter table falls and seawater intrusion becomes a major concern. Eventually, the limits for salinity in drinking water (established at 500 ppm of total dissolved solids, TDS, by the American Environmental Protection Agency) as well as for agricultural uses may be ex-ceeded in the pumped groundwater, thus making it unsuitable for human uses.

A common management approach when a pumping well becomes contaminated by saltwater is to relocate pumping further inland (Barlow, 2003) and, if no additional fresh water sources are available to satisfy demand, the high groundwater extraction rate needs to be maintained or increased at the new location thus putting also groundwater further inland at risk of saltwater intrusion. However, a possible alternative, innovative and potentially more sustainable approach for groundwater supply in water-stressed coastal areas was proposed by the European research project WASSER (Utilization of groundwater desalination and wastewater reuse in the water supply of seasonally-stressed regions), within which part of this thesis work was carried out. Specifically, the WASSER project assessed the economic viability of desalinating brackish groundwater and controlling seawater intrusion through artificial recharge of treated wastewater in three seasonally water-stressed, semi-arid coastal areas of the Mediterranean Sea (Koussis, 2001; Paper III).

Whereas seawater intrusion is a relatively well understood phenomenon (Bear et al., 1999), the seaward flow of mixed fresh groundwater and re-circulated seawater that discharges into coastal waters has only recently become subject of closer investigation under the common term of submarine groundwater discharge (SGD, Burnett at al., 2003). The hydrological and hydraulic possibility of large SGD rates has been a topic of debate in the scientific literature (Moore, 1996; Younger, 1996; Moore and Church, 1996; Li et al., 1999) and is also further discussed and inves-tigated in this thesis (Papers IV-VI), coupled to and interacting with seawater intrusion as main components of the same coastal groundwater-seawater flow balance.


3

Temporal variability in SGD and associated seawater intrusion may occur on a variety of time scales by processes occurring on both the sea and the land sides. For instance, diur-nal or semi-diurnal temporal variations in SGD and seawater intrusion may occur due to tidal sea-level oscillations. Additionally, annual and seasonal temporal variations in SGD and seawater intrusion may result as a consequence of temporally variable hydro-logical conditions on the landside, i.e., tem-poral variability in natural groundwater re-charge rate and/or groundwater management operations, such as pumping and artificial recharge. Furthermore, temporal variability in natural groundwater recharge and groundwa-ter inflow to the coastal zone from upstream areas in a catchment is at least to some de-gree inherently random, for instance reflect-ing the temporally random nature of atmos-pheric boundary conditions. A rational way of handling such natural variability in model-ling of these groundwater recharge and flow processes is to use stochastic approaches

(see, e.g., Foussereau et al., 2000, 2001), which allow for estimation of both expected results and uncertainty around them.

Whereas seawater intrusion has been mod-elled for many coastal areas and islands over the world (Ghassemi et al., 1990, 1996; Oberdorfer et al., 1990; Underwood et al., 1992; Nishikawa, 1997; Oki et al., 1998; Yakirevich et al., 1998; Bear et al., 1999; Paniconi et al., 2001), modelling of both the fresh and seawater components of total SGD is more rarely and recently reported in the literature (Robinson and Gallaguer, 1999; Uchiyama et al., 2000; Kaleris et al., 2002; Langevin, 2003; Smith and Zawadzki, 2003; Smith, 2004; Wilson, 2005; Papers IV and V in this thesis). Even though the same models as for seawater intrusion assessment can also be used for SGD quantification, there has traditionally been less practical interest in the latter, with particularly the study by Moore (1996), however, drastically changing this interest. Furthermore, in the majority of seawater intrusion modelling studies, tempo-

Fig. 1 Fresh groundwater and seawater flow patterns and salinity transition zone in a generic cross-section of a phreatic coastal aquifer subject to pumping and artificial recharge. Natural recharge (NR), pumping (QP) and artificial recharge (QR) rates and a freshwater inflow rate (FWI) at the inland boundary are the landside-determined flow components considered in the simulations of the groundwater-seawater system with the aim of quantifying resulting subma-rine groundwater discharge (SGD), seawater inflow (SWI) and associated salinity transition zone. SGD is defined as the total mixture of seawater and fresh groundwater flowing out from the aquifer, into the coastal water, through the underlying sediments.


4

ral variability of groundwater management operations (pumping and possible artificial recharge rates) and deterministic or random temporal variability of groundwater recharge from rainfall or upstream formations have often been neglected, considering only con-stant average recharge and pumping rates (Ghassemi et al., 1990, 1996; Sherif and Hamza, 2001). Some recent studies, however, have for instance considered changing long-term trends in terms of linearly increasing pumping rates throughout the simulation period in order to account for expected in-creased freshwater demands (Yakirevich et al., 1998; Paniconi et al., 2001).

In order to identify rational and effective groundwater management strategies in coastal aquifers, however, we need to under-stand and quantify the dynamics of freshwa-ter-seawater interactions under different natural and operational conditions of tempo-ral variability.

This thesis provides coupled quantification of SGD and seawater intrusion as two main components of the same coastal groundwa-ter-seawater flow balance on regional scales, and investigates in particular temporal vari-ability and randomness effects on this flow balance. For these quantification and investi-gation purposes, the thesis considers primar-ily as model case studies three different coastal aquifers by the Meditteranean Sea, which were conceptualized and investigated within the WASSER project (Koussis, 2001; Prieto 2001; Paper III). Based on these pri-mary model case studies, the thesis addresses the following specific objectives:

1) To investigate resulting trends, fluctuations and statistics of coastal groundwater dynam-ics and seawater intrusion under different temporal variability (Paper I) and random-ness (Paper II) conditions in regional groundwater recharge, inflow and manage-ment practices, and their practical implica-tions for effective coastal groundwater man-agement (Paper III).

2) To investigate main controls of expected SGD over regional coastline scales under steady-state (Paper IV) and tidal oscillation (Paper V) conditions, and under various case-

study conditions in addition to the primary simulation cases basis of the thesis (Paper VI).

2. GENERAL PROBLEM

DESCRIPTION

Fig. 1 shows schematically a cross-section of a phreatic coastal groundwater system, the groundwater flow patterns and salinity transi-tion zone in such a system, and the overall natural and managed boundary and sink/source groundwater flow rates that drive the resulting SGD and seawater intru-sion, which are the main focus of this thesis.

The terrestrial and marine components of the modeled groundwater flow system are (repre-sented by arrows in Fig. 1): the freshwater inflow rate through the landside boundary, FWI, specified directly, or determined by a specified freshwater pressure in the different simulations; the effective (infiltration from rainfall, minus evapotranspiration and surface runoff) natural groundwater recharge rate from the soil surface, NR; the pumping, or extraction rate of groundwater, QP; the po-tential artificial groundwater recharge (for instance of treated wastewater), QR; the sub-marine groundwater discharge, SGD, which is the total mixture of seaward fresh ground-water flow and re-circulating seawater flow that discharges into the sea; and the seawater inflow through the sea bed sediments into the aquifer, denoted hereafter SWI (but also sometimes called submarine groundwater recharge and denoted SGR), which is deter-mined in the simulations by either constant or temporally variable (to account for tidal oscillations) sea level boundary conditions.

In the different groundwater simulations performed in this thesis, initial flow and salinity conditions in the modeled aquifer system were disturbed by various assumed natural or man induced change scenarios. Maintaining then the same disturbing condi-tions for sufficient long simulation time, the coastal groundwater system evolved towards a new steady-state condition, in which aver-age flow rates and salinity transition zone do not change in time. When the groundwater flow system reaches its new steady-state


5

condition, the average water balance may be written as:

SGD+QP= FWI+NR+QR+SWI (1)

with the left hand side summarizing all the output flows from and the right hand side all the input flows to the aquifer part of the simulated cross-section. Under temporally variable conditions, equation (1) is not valid, but its different flow terms and their dynam-ics may still be quantified and compared to steady-state conditions by the numerical simulations.

The flow components SGD, SWI and FWI (the latter in case of specified pressure boundary condition on the land side), along with the resulting salinity in QP and SGD and spatial salinity distribution in the aquifer groundwater constitute then the main simula-tion-dependent output of the present thesis. In order to get this output, the flow compo-nents NR, QR, QP and FWI (the latter in case of specified flow at the landside boundary) were given input parameters that were esti-mated independently, either by off-line catchment-scale hydrological modelling or by use of site-specific available information on

Fig. 2 a) Map of the modelled Nitzanim area in the Coastal plain aquifer of Israel, and b) sche-matic representation of the site-specific simulated aquifer-sediment cross-section used in Pa-pers I and IV. The pumping and recharge well positions, P(XP, ZP) and R(XR, ZR), respectively, are given in meters, quantifying horizontal distance from shoreline for the X coordinate and vertical distance from mean sea level for the Z coordinate. SGD is defined as in Fig. 1.


6

groundwater recharge and management prac-tices, as outlined in Paper IV.

These input flow rate parameters were in the present simulations given as either constant (Papers I, IV and V), or temporally variable with deterministic and in some cases also random components (Papers I-III). Com-parisons between constant and temporally

variable simulation results show then the effects of temporal variability and random-ness on seawater intrusion. Moreover, com-parisons between simulations with constant and tidally oscillating seawater level boundary conditions show effects of tidal oscillation on SGD (Paper V).

All landside flows in equation (1) may be

Fig. 3 a) Map of the Tsairi basin in the island of Rhodes (Greece) showing the location of active wells and modeled sub-catchments areas: the so-called inland and coastal zones, and b) schematic representation of the site-specific simulated aquifer-sediment cross-section used in Paper IV. The pumping and recharge well positions, P(XP, ZP) and R(XR, ZR), respectively, are given in meters, quantifying horizontal distance from shoreline for the X coordinate and vertical distance from mean sea level for the Z coordinate. SGD is defined as in Fig. 1.


7

summed to form a key landside flow variable, QN= NR+QR+FWI-QP, which was first in-troduced, used and referred to as the net land-determined groundwater drainage in Paper IV, and which indicates the excess (positive QN) or deficit (negative QN) of fresh groundwater flow that is available for con-tributing to SGD into the coastal waters. In addition to this key landside flow component QN, discharge into the sea of the re-circulating part of seawater inflow SWI adds also to the total SGD that is a main investiga-tion focus of this thesis. The salinity dynam-ics that are also investigated here, finally, depend on the mixing of flows with different salt contents in equation (1) and on the solute transport processes of dispersion and mo-

lecular diffusion, which lead to further trans-port and mixing of salt between from saltwa-ter to freshwater in the coastal groundwater system.

The primary simulation case studies in this thesis were based on the WASSER-project conceptualization of considered representa-tive cross-sections for coastal groundwater dynamics conditions in the Coastal plain aquifer in Israel, the Tsairi basin aquifer on the island of Rhodes in Greece, and the Ak-rotiri aquifer in Cyprus (Koussis, 2001; Prieto, 2001). Fig. 2-4 show site-specific maps of each investigated coastal area and schematic representations of the associated conceptualized cross-sections that were used in the case-study simulations. Each cross-

Fig. 4 a) Map of the Akrotiri aquifer in the island of Cyprus showing the location of active wells and the modeled Zakaki area, and b) schematic representation of the site-specific simulated aquifer-sediment cross-section used in Paper IV. The pumping and recharge well positions, P(XP, ZP) and R(XR, ZR), respectively, are given in meters, quantifying horizontal distance from shoreline for the X coordinate and vertical distance from mean sea level for the Z coordinate. SGD is defined as in Fig. 1.


8

section illustrates the simulation domain, assigned boundary conditions, locations of the pumping and artificial recharge wells, and the natural and managed flow components of the simulated groundwater system. (Note that in Papers II, III, and V, the simulated cross-sections were slightly different than those illustrated in Fig. 2-4, as explained in the appended papers.)

The only two-dimensional representation of simulated cross-sections, with only one flow-weighted representative location of pumping and similarly of artificial recharge, is justified by the regional (rather than local) scale focus of the investigations, and by the groundwater pumping in the considered coastal areas taking place in wells that are more or less evenly distributed along and across the mean coastline direction (Fig. 2-4). Under such conditions, resulting flow lines may be con-siderably more normal to the coastline than radial into individual pumping and recharge wells.

The different groundwater dynamics simula-tions in the considered case-study cross-sections were carried out using SUTRA, a publicly available finite-element simulation model for saturated-unsaturated, fluid-density-dependent ground-water flow and solute transport (Voss, 1984). In general terms, four main sets of simulations were carried out for meeting the specific objectives of this thesis. With regard to seawater intru-

sion investigations (objective 1), one set of simulations addressed temporally determinis-tic long-term trends with and without sea-sonal variability fluctuations around these trends (Paper I), and another set determinis-tic versus stochastic groundwater flow condi-tions (Papers II and III).

Moreover, with regard to the different SGD investigations (objective 2), an extended set of deterministic long-term simulations for many different groundwater management scenarios were used for obtaining steady-state SGD results under constant sea level bound-ary conditions (Paper IV) and a correspond-ing set of simulations with tidal sea level boundary conditions was used for direct comparison with the former (Paper V). Paper VI, finally, contains no additional simulation results for the primary case studies (Fig. 2-4) of this thesis, but compares directly SGD results obtained from these primary case study simulations with alternative SGD esti-mation approaches and other SGD case study results reported in the scientific litera-ture.

3. SEAWATER INTRUSION

SIMULATIONS

3.1. Deterministic long-term trend and seasonal variability analysis

For the specific purpose of investigating deterministic long-term trends and seasonal

Scenario NR

(m3 yr

-1 m

-1)

QP (m

3 yr

-1 m

-1)

QR (m

3 yr

-1 m

-1)

Resulting QN (m

3 yr

-1 m

-1)

1 707 946 0 91

2 707 946 239 269

3 707 946 478 448

4 707+707sin(2πt) 946+946sin(2πt+π) 0 91+1653sin(2πt)

5 707+707sin(2πt) 946+946sin(2πt+π) 239+239sin(2πt+π) 269+1414sin(2πt)

6 707+707sin(2πt) 946+946sin(2πt+π) 478+478sin(2πt+π) 448+1175sin(2πt)

Table 1. Simulated temporally constant and seasonally variable pumping/recharge scenarios. QN is the total large-scale net groundwater recharge calculated as QN= NR+QR+FWI-QP, with NR, QP, QR and FWI defined as in Fig. 1.


9

variability effects on seawater intrusion dy-namics caused by seasonally variable pump-ing, and artificial and natural recharge rates (Paper I), the Israel aquifer case study (Fig. 2) was used and subjected to different constant or seasonally variable groundwater recharge and management scenarios (Table 1). In these scenarios, artificial recharge rates were arbitrarily selected to correspond to some percentage of the simulated pumping rate (Fig. 5).

In the seasonally variable simulation scenar-ios, a hypothetical sinusoidal function with a return period of 1 year was considered as a reasonable approximation of seasonal vari-ability in natural recharge and pumping rates (Tahal, 1999). The amplitudes of the periodic functions were chosen equal to the mean and the different artificial recharge rates were assumed to have the same seasonal depend-ence as the pumping rate, but with smaller magnitude. Both the pumping and the artifi-cial recharge rate seasonality, however, was delayed by six months relative to the natural recharge rate seasonality. All simulations were run from some steady-state initial condition (Fig. 3 in Paper I) until a new steady-state, or a quasi-periodic steady-state condition was reached after a step change in QP and possi-ble QR rates. The dynamics of saltwater intru-sion was investigated through analysis of resulting salinity evolution in the pumped groundwater. A similar analysis, with consis-tent results, has also been carried out and reported for the Rhodes aquifer case study by Prieto (2001).

3.2. Stochastic variability analysis

In order to assess effects of temporally ran-dom natural groundwater recharge on sea-water intrusion (Paper II) and quantify the uncertainty associated with randomness in groundwater flow and salinity dynamics (Pa-per III), comparative deterministic and sto-chastic simulations were conducted for all three case studies (Fig. 2-4).

Various stochastic simulation scenarios were investigated using the Monte Carlo approach with 100 different, equally probable realiza-tions of considered random hydraulic con-ductivity fields and groundwater recharge

time series. The stochastic simulation scenar-ios (Table 1 in Paper II) considered: only temporal hydrological randomness with re-gard to NR in all three aquifer cases and also to FWI in the Cyprus aquifer case (denoted by QINF in Papers II, IV and V); only spatial aquifer heterogeneity in terms of a random distribution of the hydraulic conductivity, K, field; and both such temporal and spatial randomness. Corresponding results from these stochastic simulations were then com-pared with each other and with relevant deterministic simulation results in order to assess saltwater intrusion (Paper II) and groundwater management (Paper III) effects of separate and combined spatial-temporal randomness.

In all simulation scenarios, the natural groundwater recharge NR (in all aquifer case studies), the inland boundary inflow FWI (in the Cyprus aquifer case), and the pumping and artificial recharge rates were given as monthly variables. The generation of site-specifically relevant deterministic and sto-chastic QP, QR, NR and FWI time series was carried out and provided by partners within the WASSER project, as described in Paper II for all three case studies and in Paper III for the Rhodes case study. The different case studies constitute quite different examples of temporal variability and randomness (Fig. 3 in Paper II). For instance, the average coeffi-cient of variation of the temporally random input field NR+FWI over the entire simula-tion period ranges from about 0.1 to 2.4 for the different aquifer cases. In contrast, due to lacking site-specific characterization of spatial heterogeneity, the same relatively small de-gree of spatial heterogeneity (variance of log K equal to 0.5) was considered for all aquifer case studies.

The simulations for the stochastic variability analysis were run for a period of 20 years based on the selected planning period for testing the application of the WASSER con-cept in the three case studies. The results of these simulations were analysed in terms of expected salinity evolutions in pumped groundwater and associated uncertainty as quantified by the salinity standard deviation and coefficient of variation.


10

4. SGD SIMULATIONS

4.1. Steady-state analysis with constant sea level boundary condition

Long-term simulations (simulation period of 192 yr) of many different groundwater man-agement scenarios in the three case studies (Table 2 of Paper IV) were carried out for SGD quantification under constant sea level boundary conditions. The different simulated scenarios correspond to different manage-ment alternatives with artificial recharge rates not only smaller than the pumping rate but also higher than that. Moreover, the consid-ered groundwater pumping and artificial recharge rates, as well as natural groundwater recharge and freshwater inflow rates were temporally constant over the simulation period. Table 2 in Paper IV lists the net land-determined groundwater drainage, QN, value resulting from each case and management scenario, which at steady-state is a represen-tative constant indicative of the aquifer flow balance conditions (see QN definition in section 2). All considered simulations were run from some steady-state or arbitrary initial conditions with an initial QN value until a new steady-state or quasi-steady state was reached with a new representative QN value. The resulting steady-state SGD rate, its sea-water and freshwater components, and asso-ciated SWI were combined for all cases and management scenarios and analysed as func-tions of the steady-state QN value.

4.2. Tidal oscillation analysis

In order to analyse the effects of tidal oscilla-tions relative to steady-state SGD results (Paper V), two different periodic tidal oscilla-tion cases were used as temporally variable sea-level boundary condition in correspond-ing groundwater management scenarios as for the steady-state SGD results (Table 1 of Paper V). The considered tidal oscillations were a relatively large diurnal tidal oscillation of 0.65 m amplitude (tidal period, T= 24 h; tidal amplitude, a= 0.65 m) and a relatively small semidiurnal tidal oscillation of 0.1 m amplitude (T= 12 h, a= 0.1 m). The steady-state flow conditions and associated salinity distributions resulting from the non-tidal

SGD simulations (section 4.1, Paper IV) of groundwater management scenarios were used as initial conditions for corresponding simulations with tidal boundary conditions (Paper V). The tidal simulations required a very small time step size to get accurate solu-tions and were run for a total period of about 2 years using a maximum time step size of 20 min.

5. RESULTS

5.1. Seawater intrusion simulations

5.1.1. Deterministic long-term trend and sea-sonal variability analysis

Fig. 5 shows the evolution of salinity in the pumped groundwater resulting from the simulation of the temporally constant pump-ing/recharge scenarios (1-3, Table 1) and corresponding seasonally variable pump-ing/recharge scenarios (4-6, Table 1) for the Israel aquifer case study (Fig. 2, Paper I). The seasonal variability in pumping and recharge rates causes the net groundwater drainage flow rate, QN, (Table 1) and the salinity in pumped groundwater (Fig. 5) to also vary seasonally around a mean value towards the new steady state condition. At this new steady-state, which is reached after the long simulation period of 192 yr in all scenarios, the amplitude of salinity fluctuations relative to the mean are about 20-25% for all scenar-ios.

All the scenarios lead also to significant in-crease of salinity in the pumped groundwater, due to the initial step decrease in the constant or mean QN value from 741 m3 yr-1 m-1 coast-line to the values shown in Table 1 for each simulated scenario. The resulting mean salin-ity in the seasonally variable scenarios is not exactly the same but close to the resulting salinity in the corresponding constant scenar-ios (Fig. 5). Initially, the mean salinity in the seasonally variable scenarios is greater than the corresponding salinity in constant scenar-ios, but may eventually develop to being smaller than latter, such as in scenarios 1 and 2. In the long-term, the relative deviation between the mean salinity and the corre-sponding, constant-scenario salinity is gener-ally quite low, i.e., less than 10% (Paper I).


11

Moreover, the maximum relative deviations between the seasonal and the corresponding constant-case salinity values are also relatively small (about 27%) in the long-term, but con-siderably higher deviations (about 55%) are found during the first 25 simulation years (Paper I). The scenarios with zero artificial recharge (1 and 4) have the lowest mean or constant QN values (Table 1) and cause the most adverse increase of salinity in the pumped groundwater of all considered sce-narios. Increasing artificial recharge and associated mean or constant QN value re-duces considerably the constant or mean (quasi-)steady-state long-term salinity value relative to the scenario without recharge. Artificial recharge also decreases the ampli-tude of fluctuations in QN (Table 1) and salinity (Fig. 5) around their respective mean values in the seasonally variable scenarios. Consistent results were also obtained in the analysis of long-term trends and seasonally variable effects for the Rhodes aquifer case study presented in Prieto (2001).

5.1.2. Stochastic variability analysis

Fig. 6 illustrates comparative stochastic and deterministic simulation results for the Rho-des case study, which is considered in both Papers II and III. Specifically, Fig. 6a shows the temporal evolution of expected salinity in pumped groundwater obtained by the cou-

pled spatial-temporal stochastic (denoted E[CST]), and corresponding spatial stochastic (denoted E[CS]), temporal stochastic (de-noted E[CT]), and deterministic (denoted CD) simulation scenarios. The deterministic solu-tion, CD, does not reproduce exactly but follows quite well the general trend of salinity decrease and patterns of temporal fluctua-tions in expected salinities E[CS], E[CT] and E[CST] from stochastic simulations. This result is consistent in the Israel and Cyprus aquifer case studies, although the overall salinity development and its temporal vari-ability are very much site-specific (e.g., Fig. 4 and 5 in Paper II; Koussis, 2001; Prieto, 2001). Effects of temporally random rainfall and upstream inflow on expected salinity in pumped groundwater is analysed by consid-ering the ratios E[CT]/CD and E[CST]/E[CS] for each aquifer case. Fig. 5 in Paper II shows that these effects are generally greater in a homogeneous than in a heterogeneous aqui-fer.

Moreover, the overall magnitude and tempo-ral evolution of the coefficient of variation, CV, of salinity in pumped groundwater re-sulting from the different stochastic simula-tion scenarios (CV[CST] from the coupled spatial-temporal stochastic, CV[CS] from the spatial stochastic and CV[CT] from the tem-poral stochastic simulations) are also site-specific and generally follow the same overall

Fig. 5 Evolution of salinity in the pumped groundwater under the temporally constant pump-ing/recharge scenarios 1-3 (defined in Table 1) and corresponding seasonally variable pump-ing/recharge scenarios 4-6 (Table 1); NR, QP and QR are defined as in Fig. 1.

0 25 50 75 100 125 150 175

Time (yr)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000T

DS

(pp

m)

atP

(110

0,-5

0)

Scenarios 4-6Q

P= 946 + 946 sin(2πt+π) m

3yr

-1m

-1

NR= 707 + 707 sin(2πt) m

3yr

-1m

-1

Scenarios 1-3Q

P= 946 m

3yr

-1m

-1

NR= 707 m

3yr

-1m

-1

Scenarios 1 & 4, QR= 0% Q

P

Scenarios 2 & 5, QR= 25.27% Q

P

Scenarios 3 & 6, QR= 50.55% Q

P


12

trend and fluctuation patterns for the differ-ent investigated stochastic simulation scenar-ios (Fig. 6b; see also Fig. 15 in Paper III for the Rhodes aquifer case study; and Koussis (2001) and Prieto (2001) for the Israel and Cyprus aquifer cases). The coupled spatial-temporal simulations yield generally over the whole simulation period greater coefficient of variation (CV[CST]) of salinity in the pumped groundwater than the only temporal (CV[CT]) and only spatial (CV[CS]) simulations.

Fig. 7a further shows in terms of the ratio CV[CT]/CV[CS] that for the Israel and Rho-des cases, the effect of temporal randomness on the relative salinity uncertainty is smaller

than the effect of spatial randomness, whereas the opposite applies to the Cyprus case. Fig. 7b also shows in terms of the ratio CV[CST]/CV[CS] that the additional effect of temporal randomness to that of only spatial randomness is small in the Israel and Rhodes cases, but much greater in the Cyprus case. The Cyprus case, however, was not the one with the highest coefficient of variation of temporally random fluctuations in the input field NR+FWI (Fig. 3 in Paper II). The greater effect of temporal randomness in this case relative to the Israel and Rhodes cases is better explained by case differences in aquifer depth (Fig. 7 in Paper II), implying that

Fig. 6 Comparison of the results from the spatial stochastic (solid line), the temporal stochastic (dash-dot line), the coupled spatial-temporal stochastic (dash-dot-dot line) and the determinis-tic (dashed line) simulation scenarios (defined in Table 1 of Paper II) for the Rhodes aquifer case study in terms of a) expected salinity in pumped groundwater and b) coefficient of varia-tion of salinity in pumped groundwater.

0 5 10 15 20

Time (yr)

6000

7000

8000

9000

10000

11000

12000

13000

14000

15000T

DS

(pp

m)

atP

(15

00

,-6

4)

E[CT]

NR

random variable

Deterministic solution (CD)

E[CS]

K random variable

E[CST

]K & N

Rrandom variables

a

0 5 10 15 20

Time (yr)

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

0.3

CV

[CX]atP

(15

00

,-6

4)

CV[CT]

NR

random variable

CV[CS]

K random variable

CV[CST

]K & N

Rrandom variables

b


13

deeper aquifers have greater possibility to attenuate effects of temporal randomness in hydrological boundary inflows (NR and FWI) than shallow aquifers.

5.1.3. Coastal groundwater management investigations

In addition to their own specific results and scientific contributions, the above-described model developments, investigations and results of groundwater and seawater intrusion dynamics formed also a necessary basis for the applied, practically relevant and much broader investigations of potentially viable groundwater management solutions for wa-ter-stressed coastal areas, as presented in

Paper III. By examples from around the world, Paper III shows the general world-wide vulnerability of coastal aquifers to sea-water intrusion resulting from intensive groundwater development. Given a continu-ing high demand pressure on coastal aquifers and the high risks and costs of coastal groundwater pollution by intruding seawater, it is an issue of world-wide importance to make coastal zone development sustainable.

A main contribution of Paper III in the spe-cific context of this thesis is to clarify the role and importance of the specific basic ground-water dynamics and uncertainty simulation investigations of Papers I and II within the much wider framework of an applied re-

0 5 10 15 20

Time (yr)

0

2

4

6

8

10

12

14C

V[C

T]/C

V[C

S]atP

(XP,Z

P)

a

Rhodes

Israel

Cyprus

0 5 10 15 20

Time (yr)

0

2

4

6

8

10

12

14

16

18

20

CV

[CS

T]/C

V[C

S]atP

(XP,Z

P)

RhodesIsrael

Cyprus

b

Fig. 7 Relative effects of temporal randomness on the coefficient of variation of salinity in pumped groundwater for the Israel, Rhodes and Cyprus aquifer case studies in terms of result-ing ratios: a) CV[CT]/CV[CS], which relates the pure temporal randomness effects (CV[CT]) to those from pure spatial randomness (CV[CS]), and b) CV[CST]/CV[CS], which relates the com-bined spatial-temporal randomness effects (CV[CST]) to CV[CS].


14

search methodology for investigating the economic viability and sustainability potential of a groundwater management strategy that uses desalination of brackish groundwater, coupled with control of sea intrusion via recharge of treated wastewater, for meeting increasing freshwater demands while control-ling seawater intrusion in coastal regions. Through the example simulation case study of Rhodes, Paper III shows that this can be an economically viable, sustainable ground-water management strategy in seasonally water-stressed, semi-arid coastal zones and islands. Paper III further identifies the uncer-tainty in groundwater and seawater intrusion dynamics that stems from natural spatial-temporal randomness as a critical main issue to continue to address in further scientific investigation of the viability of different groundwater management possibilities in such water-stressed coastal regions.

5.2. SGD simulations

5.2.1. Steady-state analysis with constant sea level boundary condition

Fig. 8 shows resulting steady-state SGD and corresponding SWI for all three aquifer case studies and scenarios as functions of the net land-determined groundwater drainage QN. Steady-state SGD and SWI are approximately linear on QN for the considered simulation conditions regardless of the existing differ-

ences in site-specific aquifer characteristics, initial conditions and investigated groundwa-ter management scenarios. The resulting steady-state SWI varies over a relatively smaller range (from about 200 to 1300 m3 yr-1 m-1 coastline) relative to SGD (from almost 0 to 11000 m3 yr-1 m-1 coastline) over the inves-tigated range of QN values.

The steady-state freshwater and seawater fractions of SGD were also evaluated over this QN range showing an overall behaviour (Fig. 4 in Paper IV) that is also quite inde-pendent of field case, simulated scenario and initial conditions. From all these results and the analysis of long-term salinity transition zones and associated streamlines for different examples of QN scenarios (Fig. 5 and 6 in Paper IV), a number of observations can be made. In case of deficit in fresh groundwater flow (QN< 0 m

3 yr-1 m-1 coastline), both freshwater and seawater inflow is pumped, in the pumping well, yielding small total SGD, consisting mostly of seawater. If QN increases to 0 m3 yr-1 m-1 coastline, the available fresh-water flow equal the pumped groundwater rate and the resulting total SGD of about 400 m3 yr-1 m-1 coastline (Fig. 8) contains some freshwater and some re-circulated seawater

(≥ 30% of total SGD, Fig. 4 in Paper IV) within a relatively wide salinity transition zone (Fig. 5 in Paper IV).

Fig. 8 Resulting SGD (filled symbols) and SWI (open symbols) as functions of the net land-determined groundwater drainage QN at steady-state for Israel (squares), Rhodes (triangles) and Cyprus (diamonds) with linear fit for SGD(QN) results obtained as SGD= 1.3QN+470 m3 yr-1

m-1 (R2= 0.98). SGD and SWI are defined as in Fig. 1, and QN is defined as in Table 1.


15

Increasing QN to positive values, implies an excess of available fresh groundwater flow that can be discharged as SGD. Seawater inflow also increases with QN (Fig. 8), but its increase is relatively small compared to the increase in QN, with most of it re-circulating and contributing with up to 10% of high total SGD values under our simulation conditions (Fig. 4 in Paper IV). The fresh groundwater component, QN, of total SGD is then also sufficiently high to efficiently prevent sea-water intrusion into the aquifer and yields a narrow salinity transition zone close to the shoreline (Fig. 9 and Fig. 6 in Paper IV).

5.2.2. Tidal oscillation analysis

The influence of a periodic tidal sea-boundary condition on SGD and SWI is that both flows oscillate around some mean val-ues, with the same period as but with a phase lag relative to the tidal elevation. Fig. 10 shows the resulting average SGD and SWI values (SGDaver and SWIaver) from the simula-tion of the relatively high tide (a= 0.65 m, T= 24 h) along with the corresponding steady-state SGD and SWI results from non-tidal simulations (SGDss and SWIss), as functions of the net land-determined groundwater drainage, QN.

For values of QN> 5000 m3 yr-1 m-1 coastline,

the 65 cm tide yields average total SGD (SGDaver) with the same values and approxi-mate linear QN dependence as the resulting steady-state SGD under constant sea level conditions (SGDss). Consequently, average SWI (SWIaver) also coincides with SWIss for QN> 5000 m

3 yr-1 m-1 coastline. For smaller QN values, however, i.e., areas with relatively small recharge of fresh groundwater and/or excessive groundwater extraction, the 65 cm tide increased, at times considerably, the resulting SGDaver and SWIaver values above the corresponding steady-state results, SGDss and SWIss, respectively. Furthermore, the SGD value in the tidal cycle is temporally increased to a maximum of 1.5 to 4 times SGDaver for all range of investigated QN sce-narios (Fig. 5a in Paper V).

The effect of the relatively small, semidiurnal tidal fluctuations of 10 cm amplitude on SGDss was small for all range of investigated QN values, with resulting SGDaver being close

to SGDss for QN≥ 0 m3 yr-1 m-1 coastline, and with some relatively higher deviations for QN< 0 m

3 yr-1 m-1 coastline. Moreover, the resulting range of temporal variations of SGDaver in a tidal cycle is considerably smaller

Fig. 9 Salt concentration isolines (in ppm TDS) resulting from the steady-state simulation (dashed lines) scenario QN= 5000 m3 yr-1 m-1 coastline (see scenario definition in Table 2 of Paper IV or Table 1 in Paper V), compared with corresponding results from the simulation of diurnal tidal oscillation with amplitude 0.65 m (solid lines; this configuration does not change

over a tidal cycle); QN is defined as in Table 1.


16

than that from the 65 cm tide (Fig. 5 in Pa-per V).

In contrast to SGD and SWI results from the tidal simulations, the salinity transition zones does not change over with the tidal cycle, which is consistent with the results of Ataie-Ashtiani et al. (1999). Moreover, the transi-tion zones from tidal simulations are not significantly different from corresponding steady-state transition zones (Fig. 7 in Paper V). In particular for QN around or greater than 5000 m3 yr-1 m-1 coastline (Fig. 9), the salinity transition zone resulting from the tidal simulation coincides with the corre-sponding steady-state transition zone. In this case, the fresh groundwater flow is generally sufficient for pushing and narrowing the salinity transition zone into a more or less sharp interface between fresh groundwater and seawater at or close to the shoreline.

5.2.3. Comparative SGD analysis

From direct seepage meters or mini-piezometers measurements, high SGD has for instance been reported from Barbados (Lewis, 1987) and Jeju Island (Kim et al., 2003). Among the highest SGD estimates is also that reported for the South Atlantic Bight region by Moore (1996), from an indi-rect estimation methodology based on excess

of 226Ra in regional coastal waters, which can also be compared to direct SGD flux meas-urements in this region (Simmons, 1992). Papers IV and V in this thesis further pro-pose an additional SGD estimation method-ology, in terms of the simulation-suggested approximately linear dependence of total SGD on its freshwater component QN, which appears particularly robust for high SGD and QN conditions.

In paper VI of the thesis, we have directly compared the different measurement-based (SGDm), hydrological/freshwater-based (SGDQ) and coastal 226Ra-based (SGDJn) estimation methodologies for regional SGD for the three reported high-SGD regional cases of Barbados, Jeju Island and South Atlantic Bight. Table 1 in Paper VI lists and Fig. 11 in this summary illustrates the results of this direct comparison. The comparison shows that SGDQ estimates based on the proposed approximately linear hydrologically-determined SGDQ=SGD(QN) relation of Papers IV and V are generally (order-of-magnitude) consistent with the measurement-based estimate SGDm, which is the mid-range value of direct local SGD-flux measurements that is expected to represent a relevant order-of-magnitude estimate of average regional SGD. The coastal 226Ra-based SGD estimate

Fig. 10 Resulting temporal average SGD (SGDaver, open square symbol) and SWI (SWIaver, open diamond symbol) for the diurnal tidal oscillation with amplitude 0.65 m compared with previ-ous corresponding steady-state results, SGDss (filled square symbol with dashed best fitted line) and SWIss (filled diamond symbol), as functions of the net land-determined groundwater drain-age, QN. SGD and SWI are defined as in Fig. 1, and QN is defined as in Table 1.


17

(SGDJn) for the South Atlantic Bight region, however, is almost two orders of magnitude greater than the mutually consistent SGDQ and SGDm estimates.

The by now commonly recognized effect of spatial subsurface variability yielding highly variable local groundwater fluxes (e.g., Da-gan, 1989) implies that it should not be sur-prising to find locally measured SGD fluxes that are much greater than an average SGD estimate for an entire regional coastline (as is for instance found at the Jeju Island, with its high variability range of local SGD fluxes around the mid-range SGD value). Direct local flux measurements in seepage meters or mini-piezometers represent conditions locally at only a small fraction of a regional coastline and yield thus local SGD estimates that may largely vary between different locations along the coastline, whereas SGD determination from either average regional hydrological conditions or radium concentrations in coastal waters represent average regional SGD (Oberdorfer, 2003), for instance over coastline lengths of tens of kilometres. In the comparative analysis of Paper VI, it is there-fore the mid-range (or, if the necessary statis-tical information of spatial flux variability along a coastline would exist for calculating it, the statistical average) value SGDm, rather than any arbitrarily chosen local SGD-flux

value, that is the relevant measurement-based quantity to compare with the regional coast-line-average hydrological- and coastal ra-dium-based estimates SGDQ and SGDJn, respectively.

In the particular regional case of the South Atlantic Bight, the coastal water extent that was represented by the coastal radium-based SGDJn determination was 320 km of coast-line length and a 20 km off-shore extent perpendicular to the shoreline (Moore, 1996; Moore and Church, 1996). For such large-scale regional SGDJn representation, the order-of-magnitude difference between SGDJn and the mutually consistent regional SGDQ and SGDm estimates is remarkable.

This large difference between the estimation of large SGD from excessive coastal 226Ra loading and other mutually consistent SGD estimates is in Paper VI identified as a para-dox of general importance. Specifically, Paper VI proposes as a possible resolution to this paradox that significant freshwater inputs following the combined pathway of ground-water inputs to streams/rivers, and through the latter further to the sea, may be missed by current monitoring practices in even well-monitored parts of the world. With the inde-pendent example of the Baltic Sea drainage basin, Paper VI further shows that consider-able coastal data gaps may indeed exist in

Fig. 11 Radium-based SGD estimate (SGDJn) for South Atlantic Bight along with ranges of hydrological estimates of QN, the corresponding SGD ranges from the SGD(QN) relationship (SGDQ) and SGD ranges from direct measurements (SGDm) for South Atlantic Bight, Barbados and Jeju Island; QN is defined as in Table 1.


18

such a well-monitored regions of the world (Wulff and Rahm, 1990; HELCOM, 1998) and mislead our attribution of main coastal pathways to either major rivers or SGD. For the Baltic Sea coastal region, Paper VI also identifies a potentially scale-independent characteristic relative distribution between highly converging freshwater discharges through scale-relative major streams/rivers (80%) and diffuse freshwater discharges occurring through both much smaller coastal stream reaches and SGD (20%).

6. GENERAL DISCUSSION AND

CONCLUSIONS

A main methodological contribution of this thesis is its so far relatively unusual overall regional perspective and investigation of: i) seawater intrusion and its possible control in sustainable coastal groundwater management; ii) SGD and its relevant quantification as one interacting part among the diverse main regional pathways of freshwater and tracer/pollutant inputs from land to sea; and iii) the integrated system functioning of both i) and ii) as main components of the same coastal groundwater system.

Seawater intrusion and SGD issues have traditionally been addressed separately, with primarily hydrologists-hydrogeologists focus-ing on the seawater intrusion problem from the perspective of meeting more or less local freshwater supply demands, and mainly oceanographers starting to be concerned of and trying to measure from the seaside po-tential significant contributions of freshwater and associated pollutant inputs to coastal ecosystems. It is only relatively recently that researchers from both disciplines started to combine forces for jointly addressing the problems of SGD quantification, measure-ment and inter-calibration of different such quantification/measurement methodologies. The SGD part of the present thesis has been developed and benefited greatly from such inter/cross-disciplinary attempts to collabora-tive problem-understanding within the SCOR/LOICZ (Scientific Committee on Oceanic Research/Land-Ocean Interaction in the Coastal Zone) international working

group 112 on submarine groundwater dis-charge.

Furthermore, the applied framework of the collaborative WASSER project investigations of various possible coastal groundwater man-agement practices in three different regional case studies required realistic linking of re-gion-specific catchment-scale hydrological and water management conditions to detailed groundwater dynamics simulations of the relatively narrow groundwater-seawater inter-action zones perpendicular to the main coast-line direction. This linkage requirement led to the idea of investigating also regional SGD as function of catchment-scale hydrological and water management conditions, as quantified by the single key parameter of land-side de-termined net groundwater discharge QN. Combined simulation results for many differ-ent case-management scenarios suggested a general approximately linear regional rela-tionship SGD(QN). General consistency was found between region-specific results from this suggested simple approximate relation-ship and corresponding measurement-based estimates of average regional SGD. The comparatively anomalously large regional SGD estimate from coastal tracer concentra-tions in the South Atlantic Bight indicates possible confusion between different main pathways of groundwater flow and pollutant inputs to coastal waters, and additional meas-urement and monitoring needs for distin-guishing and quantifying input contributions of each main pathway even in well-monitored coastal regions of the world.

Other conclusions of possible general impor-tance from this thesis work include that in-tensive pumping rates may be maintained for long time before major seawater intrusion problems for regional coastal aquifers are recognized by too high salinities in the pumped groundwater itself. After such late recognition, however, pumping wells are no longer useful and no other strategy for meet-ing regional freshwater demands may have been developed than just moving groundwa-ter pumping further upstream from the coast, and thereby only increasing the inland extent of the saltwater intrusion zone into the aqui-fer. An alternative strategy for meeting fresh-


19

water demands by coastal groundwater in a sustainable way may be to control seawater intrusion through artificial groundwater re-charge, for instance by sufficiently treated wastewater. This may considerably reduce long-term trends of salinity increase in pumped groundwater, and fluctuations around such trends, even for small artificial recharge rates compared to pumping rates.

In general, when calculating conditions for coastal groundwater management practices, it is essential to account for natural spatial-temporal variability and randomness to the degree necessary for obtaining relevant groundwater dynamics results for making effective and sustainable long-term manage-ment decisions. This thesis work has shown that spatial and temporal randomness effects may not be additive, but rather largely over-lapping, with either spatial or temporal ran-domness being the dominating part that must be accounted for in predictive groundwater dynamics calculations. With regard to tempo-ral randomness effects, this work has further indicated aquifer depth as a main parameter for the resulting significance of such effects, with temporal randomness affecting uncer-

tainty in saltwater intrusion dynamics much more for shallow than for deep aquifers.

With regard to temporal variability effects on regional average SGD and in particular the simulation-suggested simple, approximately linear regional SGD(QN) relation, the addi-tional tidal oscillation simulations of this thesis showed that this type of tidal variability may significantly affect and change such a linear dependence of steady-state SGD, but only for low SGD conditions. High SGD conditions appear to depend primarily on a dominant freshwater component, which effectively counteracts density-driven flow of seawater into the aquifer and thus decreases the effects of sea-level oscillation on total SGD by limiting its recirculated seawater component. A general implication of this result is that hydrological temporal variability (seasonal and inter-annual, with deterministic and random components) may then be quite important for high SGD cases with dominant freshwater components. Such investigations were not part of the present thesis, but should be important to undertake in future SGD research.

7. REFERENCES

Ataie-Ashtiani, B., Volker, R. E. and Lockington, D. A., (1999). Tidal effects on seawater intru-sion in unconfined aquifers, J. Hydrol., 216:17-31.

Barlow, P.M., (2003). Ground Water in Freshwater-Saltwater Environments of the Atlantic Coast, U.S. Geol. Surv. Circular, 1262, 113 p.

Bear, J., Cheng, A.H.-D., Sorek, S., Ouazar, D. and Herrera, I. (Eds.), (1999). Seawater Intrusion in Coastal Aquifers - Concepts, Methods and Practices, Kluwer Academic Publishers, The Nether-lands, 625 p.

Boswinkel, J. A., (2000). Information Note, International Groundwater Resources Assessment Centre (IGRAC), Netherlands Institute of Applied Geoscience, Netherlands. In: UNEP (2002), Vital Water Graphics - An Overview of the State of the World's Fresh and Marine Waters, UNEP, Nairobi, Kenya.

Burnett, W.C., Bokuniewicz, H., Huettel, M., Moore, W.S. and Taniguchi, M., (2003). Groundwa-ter and pore water inputs to the coastal zone, Biogeochemistry, 66:3-33.

Church, T.M., (1996). An underground route for the water cycle, Nature, 380:579-580.

Dagan, G., (1989). Flow and transport in heterogeneous formations, Springer-Verlag, New York.

Foussereau, X., Graham, W.D. and Rao, P.S.C., (2000). Stochastic analysis of transient flow in unsaturated heterogeneous soils, Water Resour. Res., 36:891-910.


20

Foussereau, X., Graham, W.D., Akpoji, G.A., Destouni, G. and Rao, P.S.C., (2001). Solute trans-port through a heterogeneous coupled vadose-saturated zone system with temporally random rainfall, Water Resour. Res,. 37:1577-1588.

Ghassemi, F., Jakeman, A.J. and Jacobson, G., (1990). Mathematical modelling of seawater intru-sion, Nauru Island, Hydrological Processes, 4:269-281.

Ghassemi, F., Jakeman, A.J., Jacobson, G. and Howard, K.W.F., (1996). Simulation of seawater intrusion with 2D and 3D models: Nauru Island case study, Hydrogeology Journal, 4:4-22.

HELCOM, (1998). The third Baltic Sea pollution load compilation. In: Baltic Sea Environment Proceedings No 70, Helsinki Commission, Baltic Sea Environment Protection Commission, Helsinki, Finland.

Kaleris, V., Lagas, G., Marczinek, S. and Piotrowski, J.A., (2002). Modelling submarine ground-water discharge: an example from the western Baltic Sea, J. Hydrol., 265:76-99.

Kim, G., Lee, K.-K., Park, K.-S., Hwang, D.-W. and Yang, H.-S., (2003). Large submarine groundwater discharge (SGD) from a volcanic island, Geophys. Res. Lett., 30(21): 2098, doi:10.1029/2003GL018378.

Koussis, A.D. (Ed.), (2001). WASSER: Utilisation of Groundwater Desalination and Wastewater Reuse in the Water Supply of Seasonally-Stressed Regions, Final Project Report to the European Commis-sion DGXII-D, Environment and Climate Programme, Contract No ENV4-CT97-0459, National Observatory of Athens, Athens, Greece, 171 p.

Langevin, C.D., (2003). Simulation of Submarine Ground Water Discharge to a Marine Estuary: Biscayne Bay, Florida, Ground Water, 41(6):758-771.

Lewis, J. B., (1987). Measurements of groundwater seepage flux onto a coral reef: Spatial and temporal variations, Limnol. Oceanogr., 32:1165-1169.

Li, L., Barry, D. A., Stagnitti, F. and Parlange, J.-Y., (1999). Submarine groundwater discharge and associated chemical input to a coastal sea, Water Resour. Res., 35:3253-3259.

Moore, W. S., (1996). Large groundwater inputs to coastal waters revealed by 226Ra enrichments, Nature, 380:612-614.

Moore, W. S. and Church, T. M., (1996). Submarine groundwater discharge, reply to Younger, Nature, 382:122.

Nishikawa, T., (1997). Testing alternative conceptual models of seawater intrusion in a coastal aquifer using computer simulation, Southern California, USA, Hydrogeology Journal, 5:60-74.

Oberdorfer, J.A., Hogan, P.J. and Buddemeier, R.W., (1990). Atoll island hydrogeology: flow and freshwater ocurrence in a tidally dominated system, Journal of Hydrology, 120:327-340.

Oberdorfer, J.A., (2003). Hydrogeologic modeling of submarine groundwater discharge: compari-son to other quantitative methods, Biogeochemistry, 66:159-169.

Oki, D.S., Souza, W.R., Bolke, E.L. and Bauer, G.R., (1998). Numerical analysis of the hydro-geologic controls in a layered coastal aquifer system, Oahu, Hawaii, USA, Hydrogeology Journal, 6:243-263.

Paniconi, C., Khlaifi, I., Lecca, G., Giacomelli, A. and Tarhouni, J., (2001). Modeling and analysis of seawater intrusion in the coastal aquifer of Eastern Cap-Bon, Tunisia, Transport in Po-rous Media, 43:3-28.

Prieto, C., (2001). Modelling freshwater-seawater interactions in coastal aquifers: Long-term trends and temporal variability effects, Licentiate thesis, Water Resources Engineering, Royal Institute of Technology, Stockholm, Sweden, TRITA-AMI LIC 2067, 75 p.

Robinson, M. A. and Gallaguer, D. L., (1999). A model of groundwater discharge for an uncon-fined coastal aquifer, Ground Water , 37:80-87.


21

Sherif, M.M. and Hamza, K.I., (2001). Mitigation of seawater intrusion by pumping brackish water, Transport in Porous Media, 43:29-44.

Simmons, G. M. Jr., (1992). Importance of submarine groundwater discharge (SGWD) and sea-water cycling to material flux across sediment/water interfaces in marine environments, Mar. Ecol. Prog. Ser., 84:173-184.

Smith, L. and Zawadzki, W., (2003). A hydrogeologic model of submarine groundwater dis-charge: Florida intercomparison experiment, Biogeochemistry, 66:95-110.

Smith, A. J., (2004). Mixed convection and density-dependent seawater circulation in coastal aquifers, Water Resour. Res., 40:W08309, doi:10.1029/2003WR002977.

Tahal Consulting Engineers Ltd, (1999). Israel case study. Background data in 1st WASSER Pro-gress Report, Annex VII, Volume IV, National Observatory of Athens, Athens, Greece.

Uchiyama, Y., Nadaoka, K., Rölke, P., Adachi, K. and Yagi, H., (2000). Submarine groundwater discharge into the sea and associated nutrient transport in a sandy beach, Water Resour. Res., 30:1467-1479.

Underwood, M.R., Peterson, F.L. and Voss, C.I., (1992). Groundwater lens dynamics of atoll islands, Water Resour. Res., 28:2889-2902.

UNEP (United Nations Environment Programme), (2002). Global Environment Outlook (GEO-3), 416 p.

Voss, C. I., (1984). SUTRA-Saturated Unsaturated Transport. A Finite-Element Simulation Model for Saturated-Unsaturated, Fluid-Density-Dependent Ground-water Flow with Energy Transport or Chemically-Reactive Single-Species Solute Transport, U.S. Geol. Surv. Water-Resour. Invest. Rep. 84-4369, 409 p.

Wilson, A.M., (2005). Fresh and saline groundwater discharge to the ocean: A regional perspec-tive, Water Resour. Res., 41:W02016, doi:10.1029/2004WR003399.

Wulff, F. and Rahm, L., (1990). A database and its tools. In: Wulff, F. (Ed.), Large-scale environ-mental effects and ecological processes in the Baltic Sea. Research programme for the period 1990-1995 and background documents, Swedish Environmental Protection Agency, Stockholm, SNV REPORT 3849:217-225.

Yakirevich, A., Melloul, A., Sorek, S., Shaath, S. and Borisov, V., (1998). Simulation of seawater intrusion into the Khan Yunis area of the Gaza strip coastal aquifer, Hydrogeology Journal, 6:549-559.

Younger, P. L., (1996). Submarine groundwater discharge, Nature, 382:121-122.

Zekster, I.S. and Everett, L.G. (Eds.), (2004). Groundwater resources of the world and their use, IHP-VI, Series on Groundwater No. 6. UNESCO (United Nations Educational, Scientific and Cul-tural Organization), 342 p.

Documents

GROUNDWATER-SEAWATER INTERACTIONS7944/...TRITA-LWR PhD Thesis 1019 ISSN 1650-8602 ISRN KTH/LWR/PHD 1019 ISBN 91-7178-027-0 GROUNDWATER-SEAWATER INTERACTIONS: SEAWATER INTRUSION, SUBMARINE