Joint inversion of physical and geochemical parameters in groundwater models by sequential ensemble-based optimal design

Abstract

Joint inversion of physical and geochemical parameters in groundwater reactive transport models is still a great challenge due to the intrinsic heterogeneities of natural porous media and the scarcity of observation data. In this study, we make use of a sequential ensemble-based optimal design (SEOD) method to jointly estimate physical and geochemical parameters of groundwater models. The effectiveness and efficiency of the SEOD method are illustrated by the comparison between the sequential optimization strategy and the conventional strategy (using fixed sampling locations) for two synthetic cases. Since the SEOD method is an optimization method based on the ensemble Kalman filter (EnKF), it invokes the time-consuming genetic algorithm at every assimilation step of the EnKF to obtain the optimal sampling locations. To enhance its computational efficiency, we improve the SEOD method by replacing the EnKF with the ensemble smoother with multiple data assimilation. Furthermore, the influence factors of the original and improved SEOD method are also discussed. Our results show that the SEOD method provides an effective designed sampling strategy to accurately estimate heterogeneous distribution of physical and geochemical parameters. Moreover, the improved SEOD method is more advantageous than the original one in computational efficiency, making this SEOD framework more promising for future application.