Quantitative characterization and calibration of salt water intrusion models with electrical resistivity tomography

Groundwater quality and coastal ecosystems in coastal areas are among the most vulnerable as they are threatened by excessive groundwater withdrawals, sea level rise and storm events potentially leading to salt water intrusions or infiltration into fresh water aquifers. The environmental protection and sustainable management of these groundwater resources often involves the development and calibration of a groundwater model subsequently used to forecast the total dissolved solid content (TDS). However, groundwater models are often built based on a limited number of sparse data due to borehole availability. Geophysical methods can provide spatially and temporally distributed data for hydrogeological modeling at relatively limited costs. In particular, electrical resistivity tomography (ERT) is very sensitive to the conductivity of pore water which is directly linked to the TDS content. The method is therefore well-suited for the monitoring of salt water intrusions. However, the inversion of ERT data involves a regularization process so that the resulting tomogram is only an estimate of the true resistivity distribution, suffering from smoothing and varying resolution. In many cases, the interpretation of ERT remains qualitative and skewed. In this contribution, we propose two different methods to improve the information content that can be extracted from ERT data. First, we show with a field example from Belgium how alternative regularization methods can be developed to integrate independent information into the inversion process of ERT. This enabled us to obtain a resistivity distribution much closer to the one observed in validation boreholes. Then, a site-specific petrophysical relationship is used to derive the TDS content of the aquifer from ERT tomograms. This can be directly used as input in the calibration process of a hydrogeological model. We also show how it is possible to counterbalance the effect of resolution loss with depth for surface ERT by filtering the results relative to their sensitivity. We show that this filtering is mandatory to use the ERT-derived information for calibrating a hydrogeological model. In a second example, we show how a fully coupled inversion approach can be used to directly invert geophysical data together with hydrogeological data for the calibration of hydrogeological models. At each iteration of the calibration, the simulated TDS content is transformed in a resistivity distribution using a parameterized petrophysical relationship and forward geophysical modeling yields the geophysical response. We show that this approach enables to better estimate the hydrogeological parameters of the simulated coastal aquifer than with an uncoupled approach if the conceptual model is sufficiently representative. With those two examples, we demonstrate the usefulness of ERT in the monitoring of salt water intrusions, both qualitatively to identify most vulnerable zones and quantitatively to estimate ERT-derived TDS contents or geophysical data and calibrate hydrogeological models. An innovative approach may consist in a conjunctive use of filtered geophysically-derived and geophysical data within the coupled hydrogeophysical inversion framework. Such an uncoupled-coupled approach based on a resolution threshold approach may offer a promising developing trend in hydrogeophysical inversion