As regional prospectivity models become more common across the sector we need to act with caution. Although they are often commissioned with the worthy aim of optimising the probability of exploration success, when wielded incorrectly they can have the opposite effect.
The typical output of a prospectivity model is a heat map ranging from 1 (most prospective) to 0 (least prospective); the important point is that this is a relative scale. As a prospective terrane approaches full maturity, the number of deposits left for discovery becomes small, and exploration becomes progressively less viable. A relative scale fails to capture this dynamic and could be used to justify exploration where none is warranted. The solution is to develop an absolute prospectivity model, in which the scale reflects the actual estimated probability that a cell hosts undiscovered mineralisation.
Furthermore, in exploration, size matters, and it is often the case that only the small probability of a very large discovery makes exploration worthwhile. Maturity tends to disproportionately diminish the probability of large discoveries as these are easier to find; consequently, the probability on any discovery diminishes with maturity and that any new discovery is small increases with maturity. This effect should also be captured, so that an optimal absolute prospectivity model encapsulates a probability distribution of undiscovered mineralisation of varying size within each cell.
A related problem is the misidentification of prospectivity as a sensitivity issue. There is often an expectation that prospectivity models can identify novel targets that traditional targeting cannot, by being more sensitive to subtle signals in the data. This may be true in some cases, but it is worth asking what kind of targets are revealed only through a prospectivity model; the answer is often second- or third-tier targets with weak expressions. To the extent that a prospectivity model prioritises novel, weak targets over stronger, recognised ones is likely the extent to which it diverts exploration capital toward poorer targets. Rather, the focus of a prospectivity model should be on noise and false-positive reduction.
To address these challenges, SRK have developed M-VAP, a Bayesian-based absolute prospectivity model. The model balances a predictive, mineral-systems-based prior model against an evidential model derived from known mineralisation and previous exploration activity. Because the model is also size-segmented, the output is a map in which each cell contains a probability distribution representing the estimated probability of undiscovered mineralisation. These probabilities can be summed across an area of interest to assess the overall likelihood that a project will yield different discovery outcomes. This approach enables better exploration decision-making and more effective capital allocation across a portfolio.
