Analysis of alternative measurement strategies for the inverse optimization of the hydraulic properties of a volcanic soil

Data of two or more state variables are recommended for porous media flow parameter estimation by inverse modeling. In this context, it is desirable to minimize efforts and cost when measuring these variables. A method to analyze the suitability of different sampling alternatives to identify the flow parameters from transient flow experiments by inverse modeling is proposed and tested in a large undisturbed volcanic soil column. Alternative measurement strategies are defined by combining observed data from different hydraulic state variables with a variable depth resolution. The state variables considered are soil water content ðuÞ; matric pressure head ðhÞ; and the water flux at the bottom of the soil column ðqÞ: The performance of the inverse analysis is measured by means of a factorial evaluation index, encompassing a measure of the goodness-of-fit and parameter uncertainty. To test this approach, two experiments were conducted where u and h were measured at a maximum measurement depth frequency of seven observation depths per profile. A first irrigation experiment served to make an initial direct estimate of the flow parameters for the four soil horizons in the column. From this initial experiment, the hydraulic parameters selected for inverse estimation are reduced to the saturated water content and the van Genuchten’s suction curve shape parameters. Results from the second transient irrigation experiment show that although inverse modeling using data from all the state variables considered (u; h and q) give the best results, monitoring of u in combination with either h or q proofs to be sufficient, even when only four observation depths are considered. q 2004 Elsevier B.V. All rights reserved.