Modelling rainfall in selected agricultural hubs in Nigeria: a comparative probability distributions study

Abstract

The most appropriate probability distribution for rainfall data varies from one region to the other. The need to research the best distribution for different locations is thus a continuous process. This research was designed to investigate rainfall variability and the best-t distribution model for monthly rainfall in Oke Ogun, the agricultural hub of Oyo state, Nigeria. Thirty-six years ' rainfall data (1982-2017) were statistically analysed and tted to ten different probability distributions for each calendar month. Parameters of each distribution were estimated from maximum likelihood estimates. Best t distribution models for each month were selected using Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC) and Corrected Akaike Information Criteria (CAIC) as model selection criteria. Gamma distribution had the best-t for six of the twelve-monthly rainfall data. Weibull distribution had best-t for three months, while Lognormal tted data for March and July best and Gumbel distribution had best-t for August. This study will enable adequate integrated water resources management as well as proper water resources evaluation and planning, for agricultural productivity in the Oke Ogun region of Oyo state. Nigeria.