Parameter Estimation Under Variable-Density Flow Conditions

Automated calibration and parameter estimation tools are being increasingly used in research and consulting environments in the groundwater modeling community.  We present an application of these tools to laboratory experiments that include the effects of variable-density flow and transport.  Since laboratory data are collected under controlled conditions we have good initial estimates for the numerical model.  Two experimental data sets were collected under similar conditions- one for the calibration phase and another for the prediction phase.  We show how the selection of the order of experimental data and various model input data and parameters affects the results obtained from the parameter estimation tools in terms of both the estimated parameter values and the uncertainty bounds, and the subsequent match between the prediction phase observations and model results.