Faced with geophysical non-uniqueness, data sets are usually combined with geologic data from other sources to calibrate interpretation results. However, the value of this approach is often limited due to lack of deep drilling cores, difficulties in obtaining access to lithology catalogs, and scarcity of existing wells in remote areas. Can hydraulic characteristics of the water level and water quality information contained in the geological body be used to assist the geophysical inversion? This study examines a method for interpreting geophysical data based on groundwater numerical modeling. By studying the known geology and generating multiple geological models interpreted from electromagnetic data, a set of groundwater numerical models was built to test the various geological models. Numerical simulation of groundwater flow and solute transport were then conducted and the optimal geological model was obtained by finding the optimal groundwater model. A sensitivity analysis was also conducted to verify the method.