INVESTIGATION OF TRANSPORT PROCESSES AT THE MADE SITE USING THE ENSEMBLE KALMAN FILTER

Over the past two decades, the Macro-Dispersion Experiment (MADE) site in Columbus, Mississippi has been studied extensively to understand solute transport processes in a heterogeneous fluvial aquifer. Despite the extremely large amount of data available at the site, controversies remain on the key transport processes that control the plume behaviors observed in previous natural-gradient tracer tests. Using a three-dimensional inverse model, Barlebo et al. (2004) represented the conductivity field as seven uniform zones, although the quality of data match and estimated conductivity values raised some further discussions (Molz et al., 2006; Hill et al., 2006). In this work, we apply the ensemble Kalman filter (EnKF) as a sequential data assimilation approach to investigate the MADE site conductivity distributions. The biggest advantage of EnKF, as compared to traditional inverse methods, is that neither the covariance nor the Jacobian of the state variables is explicitly needed, thereby improving significantly the computational efficiency in parameter estimation. Preliminary results indicate that the hydraulic head observations are essential in order to identify large-scale conductivity features at the site, while subsequent assimilation of concentration data with time can further refine the conductivity estimates to higher resolution and accuracy. Different modeling theories, including the advective-dispersive and dual-domain mass transfer, will be further examined under the EnKF framework in an attempt to provide some new insight into the transport processes at the MADE site.