A COMPARISON OF THE SEQUENTIAL GAUSSIAN AND MARKOV-BAYES SIMULATION METHODS FOR SMALL SAMPLES

We compared the performance of sequential Gaussian simulation (sGs) an d Markov-Bayes simulation (MBs) using relatively small samples taken f rom synthetic datasets. Al moderate correlation (approximately r = 0.7 0) existed between a continuous primary variable and a continuous seco ndary variable. Given the small sample sizes, our objective was to det ermine whether MBs, with its ability to incorporate the secondary info rmation, would prove superior to SgS. A split-split-plot computer expe riment was conducted to compare the two simulation methods over a vari ety of primary and secondary sample sizes as well as spatial correlati ons. Using average mean square prediction error as a measure of local performance, sGs and MBs were roughly equivalent for random fields wit h short ranges (2 m). As range increased (15 m) the average mean; squa re prediction error for scs was less than or equal to that for MBs, ev en when number of noncollocated secondary observations was twice the n umber of collocated observations. Median variance within nonoverlappin g subregions was used as a measure of the local heterogeneity or surfa ce texture of the image. In most situations sGs images more faithfully reflected the true local heterogeneity, while MBs was more erratic, s ometimes oversmoothing and sometimes undersmoothing.