Abstract
As part of a national study of water management during periods of drought, the U.S. Army Corps of Engineers is underwriting the preparation of a national drought atlas. One of the variables being analyzed for the atlas is precipitation. A statistical technique known as L-moments is the basis for the analysis. Central to the L-moment technique is the aggregation of site-specific precipitation data into homogeneous regions. This paper concerns a methodology for defining regions of similar precipitation climates that are homogeneous with respect to the statistical distribution of annual precipitation. Included are a discussion of the data, of the necessity for regionalization, and of the iterative use of clustering and an L-moment-based homogeneity test to determine the regions.
The methodology resulted in 104 precipitation regions within the continental United States. The number of stations in each region varied from 1 to 97. Problems were encountered mainly in mountainous and in and areas. They were, however, resolved in all but three regions by examining the orography and / or the data.
