Improved remediation processes through cost-effective estimation of soil properties from surface measurements

A wide range of technologies is presently available for the remediation of contaminated soils. The optimal selection depends on a number of soil characteristics. However, if the depth of the contaminated layer is considerable, the direct measurement of these properties can be costly and sometimes outright infeasible. In this paper, a method originally developed for the early detection of leaks in landfill liners has been properly modified to accommodate the estimation of soil characteristics. In particular, while the soil properties were considered known parameters in the previous model, they are now present as non-linear parameters and their estimation constitutes the main goal of the article. The resulting algorithm consists in the optimization of a suitable objective function with respect to both linear and non-linear variables and makes it possible to estimate soil characteristics from surface measurements. In particular, it is shown that partitioning linear and non-linear variables into two different sets and regularizing the inverse problems resulting from the discretization of the Richards' problem with unknown boundary conditions provides a robust numerical procedure. As an example, the method has been applied to the estimation of soil porosity, demonstrating its robustness and reliability potential. Including an inexpensive estimation procedure for proper soil parameters in the preliminary analysis can greatly improve the performances of remediation technologies whose convenience depends critically on the parameters selected case by case. (C) 2017 Elsevier Ltd. All rights reserved.