Application of spatial statistical techniques for predicting sulfur in the Pittsburgh No. 8 coal seam - SME Transactions 2014

Understanding the variability of sulfur and calorific value in coal seams is critical to planning mining operations and marketing the mined coal. In this study, statistical and geostatistical techniques were used to explore the underlying spatial trends present in a large coal deposit. Eight prediction models for the sulfur content of the coal seam were developed. Six of the models were created using geostatistical techniques, while the remaining two models were created using other techniques. The performance of each model was examined by conducting cross validation exercises. The results of the study showed that very pronounced trends and anisotropies were present in the sample data, which had to be accounted for in the modeling. The study found that a cokriging model using sulfur and ash is the best prediction model for predicting sulfur content in this coal reserve.