Satellite and multi algorithm capability for flood hazard assessment and mapping in northern Iran

Iran is one of the most flood prone areas in the world. The spring flood of 2019 was recorded one of the most devastating flood events in northern region of Iran. In this study, Sentinel-1, Sentinel-2, Sentinel-3 and Landsat-8 images were used to extract the flood map. Then, flood maps of these areas were prepared using Random Forest (RF) algorithm for Sentinel images and Support Vector Machine (SVM) algorithm for Landsat-8 images. In addition, flooding in these areas was assessed using the Fuzzy Best Worse Model - Weighted Multi-Criteria Analysis (FBWM-WMCA). The results of FBWM model showed that the criteria of precipitation, slope, height, land use, drainage density and distance from channel were the highest and the criteria of Curvature, Geology, Topographic Wetness Index (TWI), Stream Transport Index (STI), Stream Power Index (SPI) and The Topographic Ruggedness Index (TRI) played the lowest role in flooding in these areas. According to the FBWM-WMCA model, 38% of the Gorgan watershed in the northern, northwestern, western and southwestern parts and 45% of the Atrak watershed in the eastern, northeastern, northern and western parts are in high flood risk. The overall accuracy of the 2019 flood maps in Gorgan watershed for Sentinel-1, Sentinel-2, Sentinel-3 and Landsat-8 images is 89, 87, 80 and 85% and for Atrak is 91, 88, 82 and 86 percentages respectively. In general, based on the results of this study, FBWM and FBWM-WMCA models are effective and efficient for determining the weight of criteria and preparing flood risk maps, respectively.