UNCERTAINTY REDUCTION OF SUBSURFACE FLOW BY CONDITIONING HYDROGEOLOGICAL DATA

Because of the heterogeneity of geological materials and scarcity of in-situ data, the geological model is usually uncertainty embedded. In this study, the statistical moment differential equation (ME) based on the small perturbation method is applied to assess predictive uncertainty. The models were conditional on geological data such as hydraulic conductivity, hydraulic head or/and lithofacies jointly or separately. The meshless generalized finite difference method (GFDM) is adopted to obtain the first and second moment solutions advantageously and conveniently by virtue of its arbitrarily-distributed computational nodes. The conditioning data were randomly sampled from a hypothetical field with spatially correlated data, which was generated by Sequential Gaussian simulation (SGSIM). This study quantifies how different types of measurements act jointly or separately to reduce the predictive uncertainty of conditional models. The results show that, conditioning different types of measurements yields improved estimates of head.