GIS-Based Flood Susceptibility Mapping of Srinagar District, India Using Weights-of-Evidence (WofE), Frequency Ratio (FR) and Fuzzy Gamma Operator (FGO)

The Valley of Kashmir has been the site of more than fifty disastrous floods since 635 AD. Srinagar district has been inundated numerous times, since it emerged after the discharge of the large Karewa Lake from the Jhelum valley. In September 2014, cumulative effect of the south-west monsoon and the western disturbances, i.e., the storms that originate in the Caspian or Mediterranean Sea and deliver non-monsoon rainfall to north-west India, exasperated the discharge of Jhelum River, causing widespread inundation in the Kashmir valley. In this research, flood susceptibility zonation of Srinagar district using weights-of-evidence (WofE), frequency ratio (FR) and fuzzy gamma operator (FGO) has been carried out in a geographic information system. WofE and FR models are applied as autonomous data-driven techniques. FGO, on the other hand, is applied in the FR-based objective version and decision-makers knowledge-based subjective version. The novelty of the present study is that it augments FGO with subjective knowledge of a group of decision-makers using a fuzzy simple additive weighting system. Thirteen flood conditioning factors were considered for analysis including elevation, slope, slope aspect, profile curvature, plan curvature, Topographic Wetness Index, Stream Power Index, Sediment Transport Index, Topographic Roughness Index, geology, Modified Fournier Index, Land Use/Land Cover and distance from natural streams. In order to create and validate flood susceptibility index (FSI) maps, a flood inventory database obtained by digital change detection analysis of Landsat 8 OLI/TIRS images was segmented into training and test datasets and utilized for developing and validating the maps. All the resultant maps rationally illustrate the spatial susceptibility of the Srinagar district to flood. Area under the curve (AUC) was used to calculate prediction and success rates in order to determine the efficacy of the proposed methodologies. It shows that all FSI maps have ‘good’ success and prediction performance. However, the FR map has the highest success rate (AUC = 0.873), while the subjective FGO map achieves the highest prediction performance (AUC = 0.869). As such, the study reveals that the accuracy standards of FGO could be enhanced by means of incorporating of the subjective judgment. Additionally, the study indicates that the aforementioned flood susceptibility mapping techniques are significantly reliable, especially in the Himalayan regions.