iTOUGH2 (inverse TOUGH2) is a computer program that provides inverse modeling capabilities for the TOUGH codes. iTOUGH2 solves the inverse problem by automatically calibrating a TOUGH2 model (or any other model) against observed data. Essentially any TOUGH2 input parameter can be estimated based on any observation for which a corresponding TOUGH2 output can be calculated. An objective function measures the difference between the model calculation and the observed data, and a minimization algorithm proposes new parameter sets that iteratively improve the match. Once the best estimate parameter set is identified, iTOUGH2 performs an extensive analysis, which provides statistical information about residuals, estimation uncertainties, and the ability to discriminate among model alternatives. Furthermore, an uncertainty propagation analysis can be performed to quantify prediction errors.
Independent forward simulations conducted as part of an iTOUGH2 simulation-optimization run can be performed in parallel using PVM. Certain geostatistical interpolation and simulation routines of the GSLIB library are integrated into iTOUGH2 for the generation of spatially correlated random property fields in combination with the pilot point method.
While specifically developed for the calibration of TOUGH2 models, iTOUGH2 also supports the PEST protocol, i.e., iTOUGH2 capabilities can be used for the analysis of any user-supplied model or series of models.