Constraining 2D Hydrogeological Inversion With 3D Lithological Data

Abstract In hydrogeology, piezometric data are the key constraints for model calibration. Their limits for three‐dimensional (3D) parameterization have often been pointed out, with water heads commonly regarded as having too little sensitivity to vertical heterogeneity. However, transmissivity encapsulates vertical heterogeneity by definition. Knowledge about an aquifer's vertical structure critically helps better characterize the transmissivity distribution over a hydrogeological domain, even in a 2D framework. With this in mind, we propose an original 2D inversion methodology constrained by borehole logs in addition to piezometric series. To incorporate vertical heterogeneity as hard data, the optimization algorithm handles not the mere transmissivity, but its decomposition into a limited number of lithofacies or hydrofacies, each of which is associated with its own hydraulic conductivity and storativity. The method was tested on a synthetic aquifer composed of four hydrofacies, with hydraulic conductivities spanning from 5 × 10 −6 to 5 × 10 −3 m s −1 . Even with few lithological or piezometric data, the process succeeded in reproducing the distribution of the hydrofacies and most of their hydrodynamic parameters––a favorable signal for its use to further constrain a 3D model. Also, a comparison was made with a more conventional approach in which the transmissivity was directly optimized. Our new method shows improved predictive potential, resulting in a more accurate and precise estimation of overall transmissivity in the domain, and greatly limiting parameter and piezometric discrepancy in locations without data.