Identification of distributed parameter systems using plural pumping tests and prediction of groundwater flow

We propose an identification procedure of distributed parameter systems using the data of plural pumping tests. This procedure is based on the extended Kalman filter algorithm (Bayesian estimation). After the observational data of a pumping test are used to identify the spatially distributed hydraulic conductivities, the estimated values and the covariance matrix are updated iteratively with the observational data of the next pumping tests. The proposed procedure is verified with the simulated data in a model of a groundwater system, and the results are found to be reasonably accurate. Finally, we also demonstrate that the groundwater flow can be predicted with the estimated values and the covariance matrix.