Application of selected FFA methods in extreme flood prediction of River Osse

Floods are far-flung natural disasters with significant socio-economic impacts and this demand appropriate management strategies and predictions. Flood prediction could be an enduring challenge due to inaccurate rainfall measurements. In this regard, this study applied three flood frequency probabilistic methods for Osse River discharge dataset on annual maximum flow data obtained from Iguoriakhi gauge station from 1989 to 2008. These distribution methods were utilized in predicting and comparing flood discharge estimated at return periods of 2, 5, 10, 20, 25, 50, 100, and 200 years, respectively. The results reveal that Gumbel distribution, Log Pearson Type III and Log normal estimated flood discharge for 2 and 200 years return period has graphical equations of y = 214.35ln(x) + 2029.7, y = 207.02ln(x) + 1999.1, and y = 157.97ln(x) + 2126.5 with coefficients of determination (ɤ²) of 0.9988, 0.969, and 0.983 and r values of 0.9994, 0.9844, and 0.9917 respectively. At 95% confidence limit, the statistical analysis using Extreme Value (EV11) revealed a co-linearity between the data and return period. Also, in the determination of the flood parameters using chi square, mean standard deviation index (MSDI) and coefficient of determination, Gumbel distribution emerged as the most suitable probability method for Osse River flood data analysis. However, for return periods ≤ 5, Log normal probability method can be used since it predicted higher discharge for these return periods while Gumbel or Log Pearson Type III methods can be used for return periods ≥ 10 as both methods give higher flood estimates which will be helpful in safe hydrological designs.