Calibration and Reliability in Groundwater Modelling: Credibility of Modelling

(Proceedings of ModelCARE 2007 Conference, held in Denmark, September 2007). IAHS Publ. 320, 2008, 45-51.

The potential of data assimilation methods for modelling groundwater flow systems: review and synthetic examples

Harrie-Jan Hendricks Franssen

Institute of Environmental Engineering, ETH Zurich, Wolfgang Paulistrasse 15, CH-8093 Switzerland

hendricks@ifu.baug.ethz.ch

Abstract A brief literature review on Ensemble Kalman filtering methods is given. In particular, applications where internal parameters together with states were estimated, are discussed. These applications are especially found in the petroleum engineering literature, while in meteorology, surface water hydrology and models for land–atmosphere interaction, almost exclusively initial conditions are updated, without adapting internal parameters. A synthetic example illustrates the application of Ensemble Kalman filtering methods in groundwater hydrology. In cases where only the recharge rate was uncertain, no filter inbreeding was observed, although the number of realizations was small. In cases where the strongly heterogeneous transmissivity field was uncertain, up to 500 realisations were needed to avoid filter inbreeding, but in that case, states and parameters could be successfully jointly updated and improved. For the case with an unknown spatio-temporal variable recharge rate, a high monitoring frequency reduced the error more strongly than a dense network, while for an unknown spatial transmissivity the opposite was the case.

Key words data assimilation; Ensemble Kalman filter; parameter estimation; filter inbreeding