Flood Inundation Mapping with Supervised Classifiers: 2021 Gediz Plain Flood

Generation of flood inundation maps is beneficial in risk assessment and evaluation. Flood mapping can be achieved by many remote sensing techniques like change detection (CD) with thresholding machine learning-based (ML) methods. Optical synthetic aperture radar (SAR) imagery are widely used, provided different satellite systems. This study used Sentinel-1 SAR Sentinel-2 MSI data Google Earth Engine (GEE) supervised ML algorithms. Gediz Plain, Turkey was selected as the area, which an agricultural area covered mostly croplands. A event that occurred on February 2, 2021, examined map for composed. Support Vector Machines (SVM), Random Forest (RF) K-Nearest Neighbor (KNN) algorithms were models trained manually created labelled GEE. Also, CD applied after before images a traditional approach. RF classifier performs best 94% overall classification accuracy where KNN gives 93.3% value dataset, indicating robustness all-weather conditions.