Hydrogeophysical inversions using ensembled smoother with multiple data assimilation

Environmental pressure on groundwater systems has intensified during the past century because of the massive development of industrial and agricultural activities. The effective and safe management of the groundwater environment represents a significant challenge to modern society, requiring a detailed understanding of the systems involved. In hydrogeology, direct measurements of subsurface geology are often limited. Recently, more attention has been paid to inverse hydrogeophysics modelling for the spatial prediction of hydrogeological subsurface properties. In this work, electrical resistivity tomography (ERT) data and pollutant con centrations measured sparsely at borehole locations were jointly used to predict the hydraulic conductivity field using the Ensemble Smoother with Multiple Data Assimilation (ES-MDA). A synthetic case that simulates a heterogeneous aquifer was developed to assess the efficacy of the approach. The ES-MDA is an itera tive data assimilation approach that allows the estimation of unknown parameters using observed data and a forward model that relates model parameters and ob servations. One of the advantages of the ES-MDA is its capability to assimilate multiple data sources simultaneously. The hydraulic conductivity field is estimated using the ERT data and concentrations as observations in this case. The forward model is represented by laws that establish the relationship between observed data and parameters to be estimated. The method workflow begins with an initialization phase, in which an initial ensemble of parameter realizations is defined, followed by an iterative phase consisting of a forecast and update steps. During the forecast step, the forward model provides predictions corresponding to the available observa tions for each parameter realization. Then, the algorithm updates the ensemble of parameters based on the misfit between predictions and observations. In ES-MDA, all available observations are assimilated multiple times during the iterative process. The results demonstrate the potential of ES-MDA for hydrogeophysical inver sion using both ERT data and concentrations concurrently for subsurface charac terization while accounting for the uncertainty of the predictions. Furthermore, the ES-MDA assimilates multiple data sources, which can significantly improve the ac curacy of the estimated conductivity field. As a future development, it is planned to use data collected in a laboratory experiment under fully controlled conditions.