An evaluation of data requirements for groundwater contaminant transport modeling

Groundwater flow and contaminant transport models have been widely used for planning and design purposes in the past decade. Two of the most significant limitations for application of these models are data availability and parameter estimation. By use of a parameter identification algorithm and synthesized data, it is possible to isolate the effects of data availability and data uncertainty. This approach was implemented using the U.S. Geological Survey's method of characteristics (USGS‐MOC) model for a hypothetical aquifer. A parameter identification scheme attached to the USGS‐MOC model was used to determine unknown transmissivities and dispersivities. The study results showed that the predictive ability of the USGS‐MOC model (and, by implication, similar models) is limited unless relatively extensive and good quality data are available. For the example tested, it was found that extending the length of the observation series was more effective in improving parameter estimates and resolution of the contaminant plume prediction than adding observation wells. Further, when the boundary conditions were known, the contaminant predictions were much more sensitive to accurate estimation of transmissivity than to the estimation of dispersivities. The numerical results also showed that after a relatively short simulation period (less than 4 years), predicted contaminant concentrations could be significantly in error. This suggests the importance of integrating uncertainty analysis into the prediction of long‐term contaminant transport.