Abstract
Among some sophisticated interpolation techniques, Kriging is an attractive method because it is constrained by intrinsic hypothesis. In order to apply Kriging to actual data, however, it is required to choose the variogram or covariance function, which are representationo f spatial correlation.I t is well known that spatial autocorrelation of error terms in regression model is often encountered, for which error component models (ECM) have been suggested and applied. In this paper, an interpolation technique coherent with Kriging system by expanding ECM is developed.
