Application of the L-moments approach to the analysis of regional flood frequency in Northern Algeria

Abstract

Algeria is periodically affected by severe and repeated flooding. Flood hazard studies require the estimation of extreme flood flows. To predict and estimate the characteristics of maximum flows, knowledge of their distribution law is essential. The L-moments method was used to determine homogeneous regions and the appropriate distribution for these regions. Three homogeneous regions were delineated. This technique produced a Pearson type III distribution for the coastal region and the Tell Atlas and a Pareto distribution for the steppe regions and high plateaus of Eastern and Western Algeria. To estimate maximum flow at ungauged sites for different return periods, the 'index-flood' method was used. An empirical formula was developed that uses the average annual maximum flow as a function of catchment area and the average slope of the main channel to estimate maximum flow at any point on the Oued. The coefficient of determination and the Nash criterion, for the three regions, are superiors to 88% and 83% respectively. The root mean squared error, for the three regions, is less than 12%. These values show the robustness of the developed models.