Weights of Evidence (WoE) modeling has been applied in a number of GIS studies
of mineral deposits and other natural systems. One drawback to the technique is
all of the data points are equal in their contribution to the statistical results. The goal
of this project was to determine if a modification of the WoE technique could yield a
probabilistic assessment of the mineral potential of a region, and by inference, the
dollar value of a specific property position. While this approach is widely used in the
petroleum industry, it is only rarely mentioned in mineral property evaluations.
Weights of Evidence models are based on a ratio of the test points per unit area of
a test polygon to the total points per unit area of the entire study area. For this study,
two WoE values were calculated. The first used the standard WoE technique based on
the number of points (mineral occurrences) falling in a test polygon compared to the
total number of points in the study area. The second calculation was designed to
determine which attributes were associated with gold endowment by comparing the
total gold endowment of test polygons to the entire study area.
The study area for this project was the northern part of the state of Nevada, USA.
Total gold endowment for the study deposits exceeds 7,300 tonnes of metal (235 million
troy oz) with at least 20 deposits that are considered world class. Evidentiary layers
used in the study included geology, geochemistry, mineral occurrences,
geophysical studies, and isotopic data. Over six hundred individual WoE determinations
were winnowed down to four evidentiary evidence layers that showed the best
WoE contrasts combined with low conditional independence.
Gold endowment for the study area is dominated by sediment hosted gold deposits
and the WoE results reflect this. A combination of four layers proved to be the best
predictors for gold deposits on a regional basis. These were coincident multi-element
stream sediment anomalies (Au, Ag, As, Sb, and Ba >= 80%); a 4km buffer around
prospects with detectable Au and Ag or vein barite; a 5 km buffer around the
208Pb/204Pb = 39.7 isopleth; and transitional facies sediments. These four predictive layers
were combined using Bayesian probability calculations to produce a probability
map of the entire study area. This map compared well with the known deposits and
identified other target areas, one of which is currently showing excellent drill results.
The major drawback of the technique is that it is retrospective, in that it will lead
to the types of deposits that have already been found. If this study were to be done
with the deposits and gold endowments known in 1970, the study would point primarily
toward vein type deposits in volcanic rocks. Successful exploration consists of
looking beyond the obvious and following up on unexpected anomalies.