The state-of-the-art way to present geological information is no longer the paper map, but is the three-dimensional model (more about geological modelling systems). Such models are used by planners, engineers and regulators, but how reliable are they?
In statistical modelling measures of uncertainty are produced with all predictions, but three-dimensional geological models are produced by expert geological interpretation, not by statistics.
We have very little robust quantitative information about the uncertainty of these models. In order to fill this gap we undertook a designed experiment in which a group of modellers each modelled a common study area independently. The models were tested at validation boreholes, and the results analysed to quantify model precision and to identify the relative importance of different sources of uncertainty.
This is being developed in further work to benchmark model quality in different geological settings, and to develop routine methods for model assessment.
Contact Dr Murray Lark for more information