Estimation of groundwater storage from seismic data using deep learning

We investigate the feasibility of employing convolutional neural networks to estimate the amount of groundwater stored in an aquifer and delineate water-table level from active-source seismic data. The seismic data to train and test the neural networks are obtained by solving wave propagation in a coupled poroviscoelastic-elastic media. A discontinuous Galerkin method is used to model wave propagation whereas a deep convolutional neural network is used for the parameter estimation problem. In the numerical experiment, the primary unknowns, the amount of stored groundwater and water-table level, are estimated, while the remaining parameters, assumed to be of less of interest, are successfully marginalized in the convolutional neural networks-based solution. This study, through synthetic data, illustrates the potential of deep learning methods to extract additional aquifer information from seismic data, which otherwise would be impossible based on a set of reflection seismic sections or velocity tomograms.