Exploratory Data Analysis of 1951–82 Summer Rainfall around Nelspruit, Transvaal, and Possible Effects of 1972–81 Cloud Seeding

Abstract

Eastern Transvaal summer rainfall from 1951 to 1982 is examined for possible changes coincident with the operational hail suppression program, which ran from 1972 to 1981. Rough exploratory methods are used initially and lead to fitting various models by standard regression techniques. The importance of station location (geography) and year of observation (trends over time), as well as possible seeding effects, are assessed by successively fitting models incorporating these effects and comparing their fits. Finally, graphical displays are used for further exploration of residuals from those models. The results from all these analyses show unusually large amounts of rainfall on the target area in six out of the ten seeded summers. These augmented amounts apparently occurred in summers when natural rainfall was plentiful. Since the allocation of the seeding was not randomized, one should not apply statistical significance tests, and definitive conclusions about the effect of seeding cannot be drawn.