Stochastic Approach In Groundwater Modeling: A Case Study Of The Buffalo Creek Watershed
Abstract
Simulation and prediction of groundwater flow and solute (contaminant) transport highly depends upon aquifer parameters and their spatial distribution. Since this variability in space is in fact random, solutions for groundwater flow and contaminants transport are better defined through a statistical approach. This study acknowledges and considers spatial variability of horizontal hydraulic conductivity values and compares calibrated steady-state condition groundwater flow both in deterministic and stochastic approach using MODFLOW model. Based upon the discretized model, for each model run 10, 495 different horizontal hydraulic conductivity values (set) were generated using Kriging statistical distribution method and results of groundwater depth was compared with measured depth, R2 value equal to 0.7471. Seven of the eight (87 %) sets of hydraulic conductivity values ranging from 10-3 m/second to 10-7 m/second generated less error than the deterministic approach. Similarly, using the calibrated parameter, contaminant plume path has also been defined using MT3D model, with five of the eight (62 %) sets of spatially varied hydraulic conductivity values generating less error than the deterministic value for solute mass balance. Potential groundwater paths were also determined and indicated using velocity vectors calculated by MODPATH model. Moreover, contaminant plume propagation in flat slope regions of the watershed showed little advance towards the predefined exits. Rather, higher concentration contours were observed in a limited area, indicating potentially polluted regions of the watershed in shallow aquifer zones that include South Buffalo wetland. Out of the total annual base flow, about 3 %, with expected rise during dry seasons, is contributed by impaired streams through groundwater-stream flow exchange.