PNNL

Abstract

Direct measurement of soil hydraulic properties often require restrictive initial and boundary conditions and can be time consuming, of limited range and often very expensive. An indirect method of increasing popularity is inverse modeling, which uses nonlinear regression methods to estimate hydraulic parameters. A major advantage of this approach is the inherent statistical analysis, which can be used diagnostically to measure the amount of information provided by the data and to infer the uncertainty with which values are calculated. In this research, we used the inverse modeling program, UCODE, with the flow simulator, STOMP, to estimate soil hydraulic properties from unsaturated flow experiments. The optimized parameter values and their corresponding 95% linear confidence intervals were estimated, their uniqueness and sensitivity analyzed, and the overall model fit evaluated. Results show that the STOMP/UCODE combination provides a convenient way to obtain the desired information from field-scale flow and transport experiments at the Hanford Site, WA.