SEMIVARIOGRAM MODELING BY WEIGHTED LEAST-SQUARES

Permissible semivariogram models are fundamental for geostatistical es timation and simulation of attributes having a continuous spatiotempor al variation. The usual practice is to fit those models manually to ex perimental semivariograms. Fitting by weighted least squares produces comparable results to fitting manually in less time, systematically, a nd provides an Akaike information criterion for the proper comparison of alternative models. We illustrate the application of a computer pro gram with examples showing the fitting of simple and nested models. Co pyright (C) 1996 Elsevier Science Ltd