Groundwater Simulations and Uncertainty Analysis Using MODFLOW and Geostatistical Approach with Conditioning Multi-Aquifer Spatial Covariance

摘要: This study presents an approach for obtaining limited sets of realizations hydraulic conductivity (K) multiple aquifers using simulated annealing (SA) simulation and spatial correlations among to simulate heads quantify their uncertainty in the Pingtung Plain, Taiwan. The proposed used SA algorithm generate large natural logarithm (ln(K)) each aquifer based on aquifers. Moreover, small ln(K) were obtained from by ranking differences cross-variograms derived measured between pair Aquifer 1 2 (hereafter referred as Aquifers 1–2) 3 2–3), respectively. Additionally, conductivities honored horizontal variability distributions model groundwater precisely. analysis 100 combinations was successfully conducted with generalized likelihood estimation (GLUE). GLUE results indicated that could minimize iterations uncertainty, achieve behavioral simulations when reduced calibration evaluation runs, be effectively applied evaluate hydrogeological properties modeling, particularly those cases which lack three-dimensional data yet have high heterogeneity vertical conductivities.