Classification of climate by pattern analysis with Australasian and southern African data as an example

Abstract Pattern analysis shows promise in the classification of climate. As a specific example 206 African and 94 Australasian climatic stations have been grouped by these techniques. The stations were treated as a single set and data were available or calculated for 16 climatic attributes on a monthly basis giving 192 attributes per station. Separate analyses based on the whole year, summer, and winter, were carried out and groupings at the 20-, 30- and 40-group level were made. The groupings formed by pattern analysis showed geographic contiguity enabling boundaries to be drawn subjectively. For the whole year analysis 10 out of 40 groups were common to both Africa and Australasia. Means and standard deviation of some group attributes are presented. The special advantages of the numerical method used in this paper include reproducibility, the possibility of using large numbers of attributes, and particular periods of the year, and delineation of explicit hierarchical relationships between groups.