Model Calibration of a Groundwater Flow Analysis for an Underground Structure Using Data Assimilation Technique

Abstract:

Model calibration of groundwater flow analysis is a difficult task, especially in the complicated hydrogeological condition, because available information about hydrogeological properties is very limited. This often causes non-negligible differences between predicted results and real observations.
We applied the Ensemble Kalman Filter (EnKF), which is a type of data assimilation technique, to groundwater flow simulation in order to obtain a valid model that can reproduce accurately the observations. Unlike conventional manual calibration, this scheme not only makes the calibration work efficient but also provides an objective approach not depending on the skills of engineers.
In this study, we focused on estimating hydraulic conductivities of bedrocks and fracture zones around an underground fuel storage facility. Two different kinds of groundwater monitoring data were sequentially assimilated into the unsteady groundwater flow model via the EnKF.
Synthetic test results showed that estimated hydraulic conductivities matched their true values and our method works well in groundwater flow analysis. Further, influences of each observation in the state updating process were quantified through sensitivity analysis.
To assess the feasibility under practical conditions, the assimilation experiments using real field measurements were performed. The results showed that the identified model was able to approximately simulate the behavior of groundwater flow. On the other hand, it was difficult to reproduce the observation data correctly in a specific local area. This suggests that inaccurate area is included in the assumed hydrogeological conceptual model of this site, and could be useful information for the model validation.