A comparative study of different statistical methods for Flood susceptibility assessment: A case study of N'fis basin, Marrakesh High Atlas (Morocco)

Abstract

Floods are one of the natural disasters with farreaching socio-economic and environmental consequences. A large part of the N'fis catchment area is vulnerable to the risk of flooding which causes an immense loss of people and infrastructure. Therefore, an accurate assessment for the susceptibility of natural risks remains essential. The main objective of this study is to evaluate the performance of frequency ratio (FR), informative value (IV) and weight of evidence (WoE) models in flood sensitivity mapping in the N'fis watershed, high atlas of Marrakech in central Morocco. A total of 87 paleo-floods sites were inventoried based on data from Tensift hydraulic agency and extensive field surveys and 11 causative factors were considered in this study. The results show high susceptibility in the northern part of the N'fis basin and moderate to low susceptibility in its southern parts. The maps were validated according to the receiver operating characteristic (ROC) curve and the area under the curves (AUC) was calculated. Then, the accuracy rates are of the order of 85.65%, 87.75% and 88.40% for the FR, IV and WoE models respectively. Thus, the WoE model proved to be the most significative model for the analysis of flood sensitivity in this region. The results of this work can be an important support for decisionmakers for flood risk-appropriate planning.