Multiscale sources of spatial variation in soil. III. Improved methods for fitting the nested model to one-dimensional semivariograms

The previous paper in this series presented a one-dimensional stochastic nested model to account for superimposed sources of soil variation at various scales. This paper shows how the nested model can be fitted to experimental data using weighted or generalized least-squares methods that account for correlations between consecutive terms that had previously been neglected. This paper also presents a method of estimating “effective degrees of freedom” for each sampling interval and thus for estimating 90% confidence limits for the semivariogram of the nested model.