Stochastic conditional inverse modeling of subsurface mass transport: A brief review and the self-calibrating method

Abstract

Conditioning transmissivity realizations to state variable data is complex due to the non-linear dependence of transmissivity (or any univariate transform of it) and piezometric heads, concentrations or velocities. A review of the literature shows these complexities. The self-calibrating algorithm combines standard geostatistics and non-linear optimization in a way that allows the generation of multiple realizations of logtransmissivity, which are conditioned not only to logtransmissivity measurements but also to piezometric head and concentration data. The self-calibrating method is demonstrated in a two-dimensional synthetic exercise in which the trade-offs between transmissivity, piezometric head and concentration data are analyzed.