A regional study of frequency maximum daily flows of the upper Senegal river basin (Guinea, Mali, Senegal and Mauritania)

Abstract

Frequency analysis of extreme events is among the preferred tools for estimating flood flows and their return periods. In a frequency analysis, the observations must be independent and identically distributed. These assumptions are not often respected, and the parameters of the frequency distributions to be adjusted are time-dependent or co-variable. The objective of this article is to make a statistical analysis of the maximum daily flow of six hydrometric stations located in the upper Senegal River basin. Four frequency distributions were thus retained in this analysis and adjusted in conjunction with the maximum daily flow: the generalized extreme value frequency distribution (GEV), the Pearson type 3 frequency distribution, the Gumbel frequency distribution, and the Log-normal frequency distribution. The comparison between these distributions was ensured through the calculation of adjustment indices, the robustness of simulation method, the visual test, and particularly the chi-square test. The results from the four methods are not significantly different. However, after classifying the frequency distributions, the statistical quality of the adjustment was taken into account. Thus, the Pearson type 3 frequency distribution was chosen to estimate the maximum flow quantiles in the upper Senegal River basin for different return periods.

Keywords: annual maximal flow, statistical adjustment, frequency analysis, estimation, Senegal River basin