Precise confidence limits of expected monthly rainfall must be regarded as fundamental to the specification of the climatic regions of an agricultural region. The science of weather forecasting is necessarily based on extensive statistics of past experience. However, unless account is taken of such factors as skewness, the means and standard deviations calculated from such statistics will lack the necessary precision and may often be misleading.This paper shows that frequency distributions of monthly rainfall, which were demonstrably skew, may be suitably transformed so that on the new scale they approximate closely to theoretical normal distributions. As a result precise fiducial or confidence limits—shown in fact to fit the data satisfactorily—may be estimated and afterwards reconverted to the original units.Practical applications of these confidence limits include the more accurate representation of rainfall patterns, valuable guides to the vital question of optimum time of planting, and finally, better estimates of the range of mean monthly rainfall where data are available only from a limited number of years.Application of these limits to some special cases is discussed in some detail.
Statistical Modeling of Monthly Rainfall in Selected Stations in Forest and Savannah Eco-climatic Regions of Nigeria
