INVERSE PARAMETER ESTIMATION IN A LAYERED UNSATURATED FIELD SOIL

Abstract

Identification of soil hydraulic parameters is an important step for modeling soil water flow and transport behavior. Parameter fitting by inverse modeling is becoming increasingly attractive as a way to avoid experimental difficulties. In this study, we linked the inverse program SUFI (Sequential Uncertainty domain FItting) with the one-dimensional flow model HYDRUS5 to investigate the applicability of inverse modeling to the water regime of a layered field soil. Measured pressure head and water content data from an irrigation and drainage experiment were used. The soil profile at the experimental site consisted of at least five soil layers, including a compacted pan. We formulated a total of eight inverse scenarios in an effort to explain the observed pressure head and water content data. Important processes in the experiment included, among others, air entrapment and preferential flow. We obtained a relatively precise simulation of pressure head and water content only after accounting for a pan layer and treating the irrigation and the drainage phases separately. We found that fitting alone, even with 20 to 30 parameters, could not account satisfactorily for the soil hydraulic behavior. Rather, correct accounting of the soil hydraulic processes as well as the soil system proved to be essential.