Summary

When using traditional discrete multiple indicator conditional simulation, semivariogram models are based on the spatial variance of data above and below selected thresholds (cut-offs). There are two problems though; 1) the spatial distribution of a threshold can be difficult to conceptualize, and 2) ordering of the indicators may influence the results, unfortunately to change the arbitrary order, to test sensitivity, involves substantial effort. If the conditional simulations instead are based on the indicators themselves (classes), rather than the thresholds separating the indicators, then the spatial statistics are more intuitive, and reordering the indicators becomes a trivial endeavor. When class indicators are used, the indicator order can be switched at any time without recalculating the semivariograms. If thresholds are used, and the ordering is changed, all the semivariograms must be recalculated. A final advantage of using the class approach is that semivariograms calculated from transition probabilities go directly into the simulation. Despite significant differences in the methods, the simulation results are nearly identical, for cases where ordering does not cause differences when using the threshold approach. Given the consistency resulting from the class approach and its ease of use, this approach is preferred.

AAPG Search and Discovery Article #90937©1998 AAPG Annual Convention and Exhibition, Salt Lake City, Utah