Contaminant source identification in an aquifer using a Bayesian framework with arbitrary polynomial chaos expansion

Stochastic methods are widely used for the identification of contaminant source information. However, these methods suffer from low computational efficiency. To address this issue, surrogate models can be effectively utilized. In this paper, we propose a Bayesian framework with arbitrary polynomial chaos expansion (BaPC) to simultaneously identify the contaminant source information including contaminant location and release mass-loading rate. We test the applicability of the BaPC for simultaneous identification in a synthetic confined aquifer by the concentration observations from all-time steps multiple times. Our results demonstrate that this approach can efficiently and accurately identify the source information of the contaminant. In addition, the evolution of the contaminant plume can be successfully predicted by employing the estimated contaminant information. It is of crucial importance for the environmental protection and management of groundwater.