Practical Application of Geostatistics to Problem of Mapping Regional Trends in Coal Quality

M. E. Hohn, K. C. Ashton, G. H. McColloch, Jr., C. J. Smith

The West Virginia Geological and Economic Survey (WVGES) has created a data base of coal quality information used for mapping and economic assessments. Three problems arise while using the data base. (1) Although samples were taken from more than 3,000 locations, coverage is uneven geographically and stratigraphically because most samples came from active mine sites at surface exposures. (2) Most parameters measuring coal quality show high local variability, even within mines. (3) Maps and statistics must neither over-estimate the extent of favorable areas nor condemn other areas, yet some means must be found to express local averages and local variability simultaneously.

Mine operators use geostatistics to assess specific properties from cores, plan mining programs, and predict the variability in coal quality during mining. The same methods can be used with the type of regional sampling programs represented by the WVGES coal data base. Linear kriging can be used to map the average coal quality over a selected block size, yielding maps that are smoothed to a degree controlled by the geologist. Local variability is displayed through maps of estimation variance. Non-parametric kriging provides estimates of the probability of exceeding selected cutoffs.

In one study where coal quality cutoffs were specified, maps of ash, BTU, and sulfur were heavily smoothed. Although they do not represent the true variability of each variable, they are easier to interpret regionally than unsmoothed maps. Large areas of the study area showed available tonnage meeting specifications, but some coal not meeting specifications could be expected in all areas.

AAPG Search and Discovery Article #91022©1989 AAPG Annual Convention, April 23-26, 1989, San Antonio, Texas.