Application of principal component analysis in delineation of Tanzania into homogeneous rainfall regions.

This study attempts to delineate Tanzania into homogeneous rainfall regions for agricultural, hydrological purposes, etc.. In order to achieve this objective, error free monthly rainfall records from 150 rainfall stations distributed all over the country were subjected to Principal Component Analysis(PCA). The data used in the study covers 30 years period, 1961-1990. The preliminary results obtained by PCA varimax rotated solutions, Kaiser's criterion, screen test and the test of significance for common factors established 13 dominant PCA modes. Application of the more rigorous condition imposed on the results of the Burt-Bank formula at 1% level of significance reduced the dominant PCA modes to 9. However, all the 13 significant PCA modes depending on their level of significance contributed to the required factor loading patterns. These results, finally, established 17 unique common factors accounting for approximately 78% of the monthly rainfall data variance on the average. The unique common factors together with relief features delineated Tanzania into 17 homogeneous rainfall regions. The delimited regions have been described in terms of geographical and/or location with respect to major relief features, altitude, typical mean annual rainfall and seasonal rainfall distribution. The best representative station (station of highest communality) from each region provided the typical regional mean annual rainfall and seasonal rainfall distribution patterns. The importance of the delimited rainfall zones have also been highlighted.