Flood susceptibility mapping in northern regions of Iran using advanced data mining algorithms (Case study: Haraz watershed)

Floods are one of the phenomena in nature that human beings have been witnessing for a long time. In Iran, due to the large area, diffident climates, temporal and spatial density of rainfall in most watersheds, we see huge floods every year. Flood susceptibility mapping is one of the basic strategies to reduce the loss of life and property due to floods. In this study, Bagging and Shannon Entropy methods have been used to prepare flood susceptibility maps. In the current study, 201 floodplain locations were prepared. Of the 201 positions, 70% were used for modeling and map preparation. The remaining 30%, which were randomly generated, were used to validate the maps produced. Furthermore, ten effective factors including slope, land curvature, distance to river, elevation, rainfall, stream power index (SPI), topographic wetness index (TWI), lithology, land use and normalized difference vegetation index (NDVI) were used. The mentioned models determined the effect weight of each factor affecting the occurrence of floods. The ROC curve was drawn and the area below the curve (AUC) was calculated to validate the flood susceptibility map. The results showed that Bagging model has a higher accuracy than Shannon Entropy model. Therefore, the high accuracy of this model indicates that it is reliable for preparing a flood susceptibility map in areas without statistics. The results showed that Bagging model has a higher accuracy than Shannon Entropy model. Therefore, the high accuracy of this model indicates that it is reliable for preparing a flood susceptibility map in areas without statistics.