A typical coal resource characterization study calls for estimating thickness, density, and quality parameters over a block or node simultaneously. Traditionally, estimation has been done for each variable independent of the other variables. The methods range from the well known polygonal and distance weighting methods to the geostatistical method of kriging. None of these methods takes the correlations between the variables into account explicitly. A comprehensive study has been undertaken to determine whether the joint estimation technique of cokriging may be used to utilize intervariable correlations in increasing the accuracy of estimation.

Seam thickness, density, ash, calorific value, and sulphur have been studied to determine whether they are cross-correlated. Significant cross-correlations have been found to exist between ash content, density, and calorific value, where the rank of the coal is stable. A survey and a case study indicated that seam thickness may also be cross-correlated with ash content and density. Subsequently, separate kriging and cokriging results have been subjected to comparison via a cross-validation procedure. After normalization, cokriging has provided substantial improvements over kriging in estimating thickness, density, and ash content. Moreover, cokriging performed well in replicating the correlation schemes where kriging occasionally failed.

In this study, geostatistical methods have been found to produce results in compliance with their probabilistic premises.

A general purpose geostatistics software package has been written to carry out modeling and part of the research on a personal computer. This package has been designed to provide many advantages over the existing costly and black-box type software.