CLIMATE ZONES OF THE CONTERMINOUS UNITED-STATES DEFINED USING CLUSTER-ANALYSIS

A regionalization of the conterminous United States is accomplished us ing hierarchical cluster analysis on temperature and precipitation dat a. The ''best'' combination of clustering method and data preprocessin g strategy yields a set of candidate clustering levels, from which the 14-, 25-, and 8-cluster solutions are chosen. Collectively, these are termed the ''reference clusterings.'' At the 14-cluster level, the bu lk of the nation is partitioned into four principal climate zones: the Southeast, East Central, Northeastern Tier, and Interior West cluster s. Many small clusters are concentrated in the Pacific Northwest. The 25-cluster solution can be used to identify the subzones within the 14 clusters. At that more detailed level, many of the areally more exten sive clusters are partitioned into smaller, more internally cohesive s ubgroups. The ''best'' clustering approach is the one that minimizes t he influences of three forms of bias-methodological, latent, and infor mation-for the dataset at hand. Sources of, and remedies for, these bi ases are discussed. Sensitivity tests indicate that some of the cluste rs in the reference clusterings lack robustness, especially those in t he Northeast quadrant of the United States. Some of the tests involve small and large alterations to the data preprocessing strategy. The ma jor shortcomings of the analysis procedure are that the clusters are u nnaturally constrained to be nonoverlapping and also that potentially important data from points outside of the political boundaries of the conterminous United States and over water are not included. Also, othe r variables that could be important or useful in characterizing climat e type could be added to, or used in place of, the temperature and pre cipitation variables used hemin. Further work on data preprocessing te chniques is also required. Remedies for these and other shortcomings a re proposed.