Addressing Uncertainty in Contaminant Transport in Groundwater Using the Ensemble Kalman Filter

Nitrate in groundwater shows significant uncertainty which arises from sparse data and interaction among multiple geophysical factors such as source availability (land use), thickness and composition of the vadose zone, types of aquifers (confined or unconfined), aquifer heterogeneity (geological and alluvial), precipitation characteristics, etc. This work presents the fusion of the ensemble Kalman filter (EnKF) with the numerical groundwater flow model MODFLOW and the solute transport model MT3DMS. The EnKF is a sequential data assimilation approach, which is applied to quantify and reduce the uncertainty of groundwater flow and solute transport models. We conducted numerical simulation experiments for the period January 1990 to December 2005 with MODFLOW and MT3DMS models for variably saturated groundwater flow in various aquifers across Texas. The EnKF was used to update the model parameters, hydraulic conductivity, hydraulic head and solute concentration. Results indicate that the EnKF method notably improves the estimation of the hydraulic conductivity distribution and solute transport prediction by assimilating piezometric head measurements with a known nitrate initial condition. A better estimation of hydraulic conductivity and assimilation of continuous measurements of solute concentrations resulted in reduced uncertainty in MODFLOW and MT3DMS models. It was found that the observation locations and locations in spatial proximity were appropriately corrected by the EnKF. The knowledge of nitrate plume evolution provided an insight into model structure, parameters, and sources of uncertainty.