Coal quality study using multivariate geostatistics

Abstract

The use of several variables or parameters is to express coal quality. The multivariate cases can classify coal quality variables. One can produce accurate estimates by using two variables with strong correlations. Estimates using two coal quality variables can solve the problem of OK (ordinary kriging) accuracy with the weak spatial correlation between data on one variable. The purpose of this study is to estimate the distribution of TS (total sulfur) content in coal seams using two coal quality parameters. Research methods include basic statistical analysis, variography, and estimation. Based on the basic statistical analysis, CV (coefficient of variation) of TS is 0.5. However, TS variogram is poor and indicates a low spatial correlation. Variographic analysis of other coal quality parameters (thickness, CV, TM, Ash, VM) can produce useful information. Cross variogram simulation shows that TS and thickness produce an informative cross variogram parameter. This research compares IDW (inverse distance weighting), OK, and cokriging estimation methods. Estimated accuracy uses linear regression analysis, probability curve, and visualization between data and estimates. The estimation using the cokriging method is more accurate than the IDW and OK methods.