Spatial prediction of flood-susceptible zones in the Ourika watershed of Morocco using machine learning algorithms

Purpose The purpose of the paper is to predict mapping areas vulnerable flooding in Ourika watershed High Atlas Morocco with aim providing a useful tool capable helping mitigation and management floods associated region, as well whole. Design/methodology/approach Four machine learning (ML) algorithms including k-nearest neighbors (KNN), artificial neural network, random forest (RF) x-gradient boost (XGB) are adopted for modeling. Additionally, 16 predictors divided into categorical numerical variables used inputs Findings results showed that RF XGB were best performing algorithms, AUC scores 99.1 99.2%, respectively. Conversely, KNN had lowest predictive power, scoring 94.4%. Overall, predicted over 60% was very low flood risk class, while high class accounted less than 15% area. Originality/value There limited, if not non-existent studies on modeling using AI tools ML region flooding, making this study intriguing.