Using a Monte Carlo approach to evaluate seawater intrusion in the Oristano coastal aquifer: A case study from the AQUAGRID collaborative computing platform

Uncertainties in the physical parameters of a groundwater system, due to the lack of direct access to the subsurface, strongly affect the design of water management policies, so that the risk of mismanagement becomes a critical factor in complex ecological and economic analyses. Stochastic modeling may help provide uncertainty quantification and also add robustness to the analysis by means of probabilistic forecasts. In this study a stochastic approach has been employed to model hydraulic conductivity of a confining formation in a multi-layered coastal aquifer system, under conditions of uncertainty. A Monte Carlo simulation, based on a coupled flow and transport groundwater 3D model, has been carried out to propagate the hydraulic conductivity parameter uncertainty to groundwater model outputs, namely pressure head and salt concentration. The aim of the study is to assess the risk of seawater intrusion into the aquifer by means of probabilistic threshold analysis on the simulated groundwater concentrations for different aquifer exploitation schemes. Maximum difference for nodal concentrations, with reference to homogeneous aquitard configuration, was found equal to 95%, proving how important can be the impact of the spatial variability of the hydraulic conductivity of the confining layer on the simulated salt concentrations. Such analysis enables to take better decisions about the management of the groundwater resource and to make additional field investigations consistent with environmental protection. The application workflow, based on the integration of both in-house developed and public domain software tools with hydrogeological data, has been deployed on a problem solving Grid platform (http://grida3.crs4.it). Further developments will include the planning of cost-effective additional field data acquisition based on the outcome of the stochastic model.