History-matching of a contaminant transport model is fraught with difficulties. An advection-dispersion model often runs slowly; sometimes it can experience numerical problems.

Another problem is that comparison of measured concentrations with model-calculated concentrations is not straightforward. Nevertheless, measured concentrations are rich in information on local hydraulic properties and, of course, on sources of contamination. A GMDSI worked example shows the innovative use of particles in rapid probabilistic assessment of contaminant source locations and simultaneous hydraulic property inference. See also this open access article.

Programs that are provided with the PEST groundwater utility suite allow inclusion of information-rich but difficult-to-quantify measurements such as "proximity to plume" and "particle consistency status" in a stochastic inversion process. These are listed in the following table. Other programs from this suite are useful for translating particle track and location information to a format where it is amenable to plotting and visualisation.

A particle is awarded a status of "non-detect", "detect" or "inconsistent". History-matching minimises particle inconsistency.

Programs in the following table can be used as model postprocessors. As such, they support inclusion of information that can be calculated from particle tracks in calibration, and calibration-constrained uncertainty analysis, of groundwater models.

Probability that a contaminant source is undetected by present monitoring network.