Affiliations: [a] Department of Statistics, Federal University of Technology, Akure, Nigeria | [b] Agricultural and Environmental Engineering Department, University of Ibadan, Ibadan, Nigeria
Abstract: Reliable estimates of extreme flood flows are required for general flood risk management and planning. To assess the risk and nature of flood discharge, modified mixed-gamma distribution was applied to different plotting positions in taking care of shortfall in gamma distribution and how best it will estimate flood frequency at Moniya Gauging Station along Ona River. Goodness of fit, minimum absolute difference (MAD) and root mean square error (RMSE) between the observed and predicted flood flows were measured. Gamma distribution matched with Weibull plotting position which gave the highest coefficient of determination (R2) of 0.9966; mixed gamma with California gave 0.9947 while modified mixed gamma with Weibull gave 0.9964 respectively. The MAD of 49.653, 3.123 and 20.922 at return periods of 50, 100, and 200-year obtained under the gamma when matched with Weibull, California and Weibull plotting positions respectively. California gave minimal error with the RMSE of 92.85 for gamma with some information lost, correcting for error incurred, Weibull gave minimal errors of 159.20 and 93.91 for mixed gamma and modified mixed gamma distributions, respectively. Hence, we conclude that modified mixed gamma distribution with Weibull plotting position is the most suitable for prediction at the gauging station.
