摘要
Block-to-block and block-to-point kriging predictions based on block data are proposed. Blocks may be regular (mesh data) or of more general shapes. Under the assumptions of second-order stationarity and isotropicity, we show how to lessen the number of calculations of relevant block-to-block or block-to-point covariances. As illustrations, a mesh data of population and a simulated block data on convex polygons are analyzed.