Fusion of the Multisource Datasets for Flood Extent Mapping Based on Ensemble Convolutional Neural Network (CNN) Model

Floods, as one of the natural hazards, can affect environment, damage infrastructures, and threaten human lives. Due to climate change anthropogenic activities, floods occur in high frequency all over world. Therefore, mapping flood areas is prime importance disaster management. This research presents a novel framework for area based on heterogeneous remote sensing (RS) datasets. The proposed fuses synthetic aperture radar (SAR), optical, altimetry datasets areas, it applied three main steps: (1) preprocessing, (2) deep feature extraction multiscale residual kernel convolution neural network’s (CNN) parameter optimization by fusing datasets, (3) detection trained model. exploits two large-scale flooded Golestan Khuzestan provinces, Iran. results show that methodology has performance detection. visual numerical analyses verify effectiveness ability method detect with an overall accuracy (OA) higher than 98% both study areas. Finally, efficiency designed architecture was verified hybrid-CNN 3D-CNN methods.