Application of DEMATEL-AHP and SVM in identifying flood-prone areas (Case study: Barzak-e-Kashan basin)

Articles in Press

Document Type : Research Paper

Authors

2- Associate Professor of Watershed Sciences and Engineering & Corresponding Author, Department of Natural Engineering, Faculty of Natural Resources and Earth Sciences, Kashan University of Kashan, Kashan, Iran.

Abstract

Flood is one of the most frequent and costly natural disasters, and the financial and human losses caused by it every year affect a wide range of countries, especially Iran. Therefore, one of the fields of research to control flood risks is to identify the flooding points of the region. For this reason, the aim of the current research is to identify flood-prone areas in Barzak-e-Kashan basin using DEMATEL-AHP and SVM models. For this purpose, 100 flooding points were identified and recorded during field visits. In the following, 12 factors affecting the occurrence of flood including precipitation, lithology, land use, distance from stream, slope, drainage density, topographic position index, topographic wetness index, topographic roughness index, stream power index, curve number and runoff coefficient were selected for preparing maps of flood-prone areas and their layers were prepared in ArcGIS 10.7.1 and SAGA GIS softwares’ environment. The results showed that the precipitation component with the highest weight equal to 0.211 is the most effective variable on flooding and the runoff coefficient is the most influential factor and has the most relationship with other factors. Also, according to AUC=0.859, the AHP efficiency was evaluated very well and the SVM accuracy was good (AUC= 0.751) in validation phase. The map of flood-prone area also showed that the north, northwest, and west lands of Barzak basin have the highest potential for flooding. The results of the present research can be used as a road map for managers and policy makers to manage flood.

Keywords

Main Subjects

Iranian Journal of Soil and Water Research

Articles in Press, Accepted Manuscript
Available Online from 11 September 2024

Files

Share

How to cite

Statistics