Multi-hazard susceptibility mapping in the Salt Lake watershed

Abstract

Multi-hazard risks are closely related to the sustainable management of society. An effective multi-hazard risk reduction requires analysis of the individual hazards and their interplay. The current study aimed to suggest a multi-hazard probability assessment in Salt Lake watershed, Iran. At first, we construct maps depicting the most effective factors on floods (11 factors), dust source (9 factors), and landslides (12 factors), and used the prioritize the impact of each respective factor on the occurrence of each hazard. Subsequently, flood, landslides, and dust source susceptibility maps prepared using a Generalized Linear Model (GLM) and Random Forest (RF) models in the R statistical software. Accuracy of the applied model was examined using receiver operating characteristic (ROC) and relative density (R-index). The result indicates that identified hazard areas cover about 73 % of the area of the Salt Lake watershed. This result indicates that the highest portion of the research area is affected by the three hazards of the dust source area, flood, and landslide, making it a very hazardous region. Landslide hazard share the largest percentage (32.1 %) of the entire research area. After landslides, floods and dust source areas (24.8 and 8.3 %, respectively) are the next most important hazard. The results showed that the RF model with AUC values of 93.5 %, 91.4 %, and 95.6 % has higher accuracy than GLM for the three dust, flood, and landslide hazards, This multi-hazard map serves as a valuable tool for land use planning and sustainable infrastructure development for the salt lake watershed.

Keywords

• Dust source areas
• Floods
• Landslides
• Generalized linear model (GLM)
• Random Forest (RF)