Calculation of probability maps directly from ordinary kriging weights

Abstract

Probability maps are useful to analyze ores or contaminants in soils and they are helpful to make a decision during exploration work. These probability maps are usually derived from the indicator kriging approach. Ordinary kriging weights can be used to derive probability maps as well. For testing these two approaches a sample data base was randomly drawn from an exhaustive data set. From the exhaustive data set actual cumulative distribution functions were determined. Thus, estimated and actual conditional cumulative distribution functions were compared. The vast majority of correlation coefficients between estimated and actual probability maps is greater than 0.75. Not only does the ordinary kriging approach work, but it also gives slightly better results than median indicator kriging. Moreover, probability maps from ordinary kriging weights are much easier than the traditional approach based on either indicator kriging or median indicator kriging.