ESTIMATING PARAMETERS OF VAN GENUCHTEN’S CHARACTERISTIC EQUATIONS FROM FIELD MEASUREMENTS AND NUMERICAL INVERSION OF A TIME-DEPENDENT UNSATURATED FLOW MODEL

An in situ method for estimating parameters necessary for characterizing unsaturated flow in porous media has been developed and applied to 33 individual data sets from the same study site (a sand dune on the edge of a wetland). The measured field data include beginning and ending soil moisture profiles, and continuous records of net surface flux (infiltration/evaporation) and pressure head at the base of the of the profile. For each study period, the initial profile was used along with the upper and lower boundary conditions to simulate flow and storage changes in the profile from a form of the Richard’s equation. Following each model run, simulated profiles for the last time step were compared to the measured profile and if the two differed by more than a specified tolerance, the parameters in van Genuchten’s characteristic equations were adjusted and the forward modeling process was repeated. Parameter adjustment was accomplished using the Levenberg-Marquardt procedure for solving non-linear least squares problems.

A statistical analysis of the resulting (optimized) parameter estimates indicates that repeated experiments using the same procedure of analysis produce very consistent optimum parameter values. Frequency histograms of the parameter values have strong modal peaks with values that are consistent with those considered to be valid for uniform sand. The optimization procedure also produced reasonable (and unique) values of parameter estimates for a buried clay-rich layer within the otherwise homogeneous profile of dune sand.