Abstract
A regional groundwater flow model of Muika-machi area in Nugata-ken, Japan, is built by an inverse analysis procedure so called the extended Bayesian method based on Akaike bayesian information criterion (ABIC). The groundwater observation data employed in the analysis is one at a steady state condition. It is illustrated through the analysis that ill-posedness which is very often encountered in this kind of inverse analysis can be overcome by introducing prior information whose relative weight to the observation data is appropriately adjusted by ABIC. The other important problem in the inverse analysis, which is model selection or parameterization problem, can also be simultaneously solved based on ABIC. The most important contribution of this paper is considered to demonstrate the effectiveness of the extended Bayesian method by analyzing actual large scale regional groundwater flow data.
