Pattern Analysis of Growing Season Precipitation in Southern Canada

This paper presents the results of climatic pattern analyses of three- and seven-day summer (May-August) precipitation totals for southern Canada. A range of eigenvectorial methods is applied to 1949-80 data for a fairly uniform network of 170 stations that extends to the east of the Rockies, to the north of the Great Lakes, and through the St Lawrence Valley. The major objectives are to identify the characteristic spatial patterns of the precipitation (including any dependence on the eigenvectorial model) and to suggest some potential applications of the results. The entire domain variance fraction cumulatively explained by the first eight unrotated Principal Components (PCs), and also the eight orthogonally-rotated (VARIMAX criterion) PCs derived from them, is approximately 55% for square-root transformations of both the three-and seven-day totals. The unrotated PC spatial loading patterns possess moderately strong large-scale (e. g. one-third to one-half domain) anomaly features through root 3, but thereafter decay progressively on both counts. The higher roots exhibit little structure and seem especially difficult to interpret physically. The VARIMAX PC loading patterns, in contrast, each emphasize a relatively strong anomaly in a different and somewhat restricted part of the domain, and are therefore considered more meaningful physically. They collectively yield a regionalization of the domain into eight subareas within which the three- and seven-day summer precipitation tends to be spatially coherent. This regionalization is strongly supported by (i) obliquely rotated (Harris-Kaiser Case II B'B criterion) PC and Common Factor analyses, (ii) further VARIMAX PC analyses on subsets of the data, (Hi) quantitative comparison with the interstation correlation matrix from which it is derived, and (iv) its close dovetailing with its central United States counterpart in Richman and Lamb (1985). The regionalization is suggested to be of considerable potential utility for crop yield modelling, short-range weather prediction, and research into climatic variation and change.