Flood Frequency Analysis in West Africa

Abstract

Devastating flood events are recurrently impacting West Africa. To mitigate flood impacts and reduce the vulnerability of populations, a better knowledge on the frequency of these events is crucial. The lack of reliable hydrometric datasets has hitherto been a major limitation in flood frequency analysis at the scale of West Africa. Utilising a recently developed African database, we perform a flood frequency analysis on the annual maximum flow (AMF) time series, covering 246 river basins in West Africa, between 1975 and 2018. Generalized extreme value (GEV) and Gumbel probability distributions were compared to fit AMF time series with the L-moments, Maximum Likelihood (MLE) and Generalized Maximum Likelihood (GMLE) methods. Results indicated that the GEV distribution with the GMLE method provided the best results. Regional envelope curves covering the entire West African region with unprecedented data coverage have been generated for the first-time providing insights for the estimation in flood quantiles for ungauged basins. The correlation between flood quantiles and watershed properties shows significant correlations with catchment area, groundwater storage, altitude and topographic wetness index. The findings from this study are useful for a better flood risk assessment and the design of hydraulic infrastructures in this region, and are a first step prior to the development of regional approaches to transfer the information from gauged sites to ungauged catchments.