ESTIMATION OF SEMIVARIOGRAM PARAMETERS AND EVALUATION OF THE EFFECTS OF DATA SPARSITY

Semivariogram parameters are estimated by a weighted least-squares met hod and a jackknife kriging method. The weighted least-squares method is investigated by differing the lag increment and maximum lag used in the fit. The jackknife kriging method minimizes the variance of the j ackknifing error as a function of semivariogram parameters. The effect s of data sparsity and the presence of a trend are investigated by usi ng 400-, 200-, and 100-point synthetic data sets. When the two methods yield significantly different results, more data may be needed to det ermine reliably the semivariogram parameters, or a trend may be presen t in the data.