Dual State-Mechanism Adjustment Adaptive Filtering to Improve Accuracy of Subsurface Transport Models

Abstract

This study proposes a state-updating scheme and a mechanism-updating scheme for a subsurface contaminant model so as to compose a dual state-mechanism adjustment adaptive filter (DAAF) to improve the plume transport prediction. An ensemble Kalman filter (EnKF) is constructed as a state-updating scheme to incorporate the knowledge of uncertainties from both the model and the sparely scattered measurements. As an implementation of the proposed DAAF concept, a simulation with data assimilation and calibration (SDAC) system, which can adjust both the state and the mechanism, is then built to show the enhanced strength of a combined EnKF and forward data assimilation backward genetic algorithm (FDA-BGA) system. With several test cases examined, the results support the conclusion that DAAF is able to reduce both the systematic and random errors efficiently and that it provides predictions closer to the true field with lower root-mean-square errors.