Modelling Freshwater-Seawater Interactions in Coastal Aquifers : Long-term Trends and Temporal Variability Effects

Sammanfattning: Over-exploitation of coastal aquifers causes seawaterintrusion and eventually contamination of groundwater wells inmany parts of the world. Desalination of brackish water andartificial recharge of treated wastewater may then be asustainable strategy for managing groundwater supply inwater-stressed regions. A common characteristic of many coastalareas in Southern Europe is that they are seasonallywater-stressed due to that most of the annual water demand isconcentrated to a few summer months, whereas naturalgroundwater recharge takes place primarily during the wintermonths. This thesis investigates long-term salinity trends inpumped groundwater, and the effects of temporal variations inpumping, as well as in artificial and natural recharge ratesand boundary conditions on the dynamics of saltwater intrusionin two Mediterranean coastal aquifers, located in Israel andthe island of Rhodes, Greece. Furthermore, the thesis presentsa rational methodology for stochastic modelling and uncertaintyanalysis of seawater intrusion, which explicitly accounts forspatial aquifer heterogeneity and temporal randomness innatural recharge and boundary conditions. The methodology isspecifically applied to the Israel and Rhodes case studies, aswell as to an additional coastal aquifer in Cyprus.Numerical simulation of freshwater-seawater interactionsassuming constant sinks and sources may introduce deviations ofless than 10% in long-term predictions of salinity in pumpedgroundwater, relative to the mean salinity resulting fromseasonally variable management scenarios. In the short-term, upto 25 years after start of pumping and recharge activities,however, the assumption of constant management scenarios maylead to considerable (50-60%) underestima tion of maximumsalinity values under seasonally variable conditions. Suchdeviations may affect the optimal design, function andefficiency of brackish groundwater desalination plants duringthe common planning period of 25 years. Therefore, seasonalvariability must be accounted for in predictive modelling ofgroundwater salinity, if the selected management strategy is tosupply the water demand by desalinating brackish groundwater.However, possible effects of seasonal variability should alsobe assessed in comparison with possible inter-annualvariability, which may have even greater effects.The uncertainty analysis applied to three case studiesshowed that predicted salinity from deterministic simulationscan not be expected to accurately reproduce correspondingexpected concentrations of stochastic simulations.Accounting for temporal randomness, in addition to randomspatial heterogeneity, increases the difference betweendeterministic salinity results and expected salinity fromstochastic simulations. In contrast, the salinity standarddeviationdoes not increase very much by temporal randomness inaddition to spatial one. The resulting salinity coefficient ofvariation is site specific, but is also generally not very muchincreased by temporal randomness in addition to spatialrandomness.Keywords:seawater intrusion, seasonal variability,artificial recharge, spatial and temporal randomness