Best Fit Probability Distribution /Parameter Estimation Procedure for Precipitation Frequency Analysis at Some Stations in Southern Nigeria

Abstract

This study investigates the best-fit probability distribution model and parameter estimation procedure namely EV1-MOM, EV1-MOM, GEV-LM, GPA-LM and GLO-LM suitable for at-site precipitation frequency analysis at rainfall gauging stations in Ikeja, Owerri and Port Harcourt in Nigeria. The best fit probability distribution - parameter estimation procedure at each station was selected based on a scoring and ranking scheme using the relative magnitudes of five goodness of fit measures namely: root mean square error (RMSE), relative root mean square error (RRMSE), mean absolute deviation index (MADI), maximum absolute error (MAE) and probability plot correlation coefficient (PPCC) obtained for the procedures. The results of the analyses indicate that the Generalized Extreme Value - L-Moments model (GEV-LM) procedure which is highest ranking is the best fit distribution model / estimation procedure for Ikeja and Port Harcourt stations. While Extreme Value type 1 – Method of Moments (EV1 MOM) is the best fit distribution model / estimation procedure for Owerri station. They were subsequently used to predict rainfall return values for return periods of engineering importance (from T = 2 years to 200 years) at the study stations. The obtained values were used to develop rainfall frequency curves for the stations. The rainfall frequency curves developed for the stations are useful for design of engineering infrastructure for flood mitigation at the studied stations and in addition, the results provide valuable insights for hydrological analysis and policymaking, particularly in addressing the escalating risk of extreme flood events in the southern Nigerian cities due to changing precipitation patterns and climate impacts