Flood susceptibility mapping using multi-temporal SAR imagery and novel integration of nature-inspired algorithms into support vector regression

Flood has long been known as one of the most catastrophic natural hazards worldwide. Mapping flood-prone areas is an important part flood disaster management. In this study, a susceptibility mapping framework was developed based on novel integration nature-inspired algorithms into support vector regression (SVR). To end, various remote sensing (RS) and geographic information system (GIS) datasets were applied to hybridized SVR models map in Ahwaz township, Iran. The proposed two main steps: 1) updating inventory (historical flooded locations) using RS-based detection method within google earth engine (GEE) platform. mosaicked images multi-temporal Sentinel-1 synthetic aperture radar (SAR) data have used step; 2) producing standalone model SVR. methods derived from with meta-heuristic algorithms, hence forming SVR-bat algorithm (SVR-BA), SVR-invasive weed optimization (SVR-IWO), SVR-firefly (SVR-FA). A spatial database locations 11 conditioning factors (altitude, slope angle, aspect, topographic wetness index, stream power normalized difference vegetation index (NDVI), distance stream, curvature, rainfall, soil type, land use/cover) built for modelling. accuracy evaluated statistical sensitivity indices, such root mean square error (RMSE), receiver operating characteristic (ROC) area under ROC curve (AUROC) index. results indicated that all outperformed According AUROC values, predictive SVR-FA highest value 0.81, followed by SVR-IWO, SVR-BA, values 0.80, 0.79, 0.77, respectively.