The assignment of Rainfall stations into homogeneous groups: An application of principal component analysis

Abstract The calculation of eigenvectors has been used to describe annual precipitation totals over an area of large extent. Rotation of the eigenvectors brought about further clarification of the system of recording stations. By noting the eigenvector of most importance to each station, delimitation of an area was made possible. Reduction in records, a necessary prelude to our future work on forecasting models, was obtained by discarding those stations most closely related to eigenvectors whose eigenvalue was less than 0·6.