1984 Volume 62 Issue 5 Pages 783-790
Meteorological network planning involves the study of multivariate phenomenon in both space and time domains. In order to intensify the existing meteorological network it is essential to consider all possible multivariate combinations of the existing stations. In a study area with a large number of stations, it is computationally infeasible to consider all possible multivariate combinations. To resolve such complexities, Shannon's information theory is used such that the information transmitted by a set of stations in one subregion about the other subregion is minimized. The total information at a point is decomposed into internal information common throughout the subregion and the information contributed by other stations at that point. To gain a better understanding of the significance of the partition results, and to improve the spatial resolution of the methodology, the partitioning of three-station subgroup is investigated.