OmbriaNet—Supervised Flood Mapping via Convolutional Neural Networks Using Multitemporal Sentinel-1 and Sentinel-2 Data Fusion

Regions around the world experience adverse climate-change-induced conditions that pose severe risks to normal and sustainable operations of modern societies. Extreme weather events, such as floods, rising sea levels, storms, stand characteristic examples impair core services global ecosystem. Especially floods have a impact on human activities, hence, early accurate delineation disaster is top priority since it provides environmental, economic, societal benefits eases relief efforts. In this article, we introduce OmbriaNet, deep neural network architecture, based convolutional networks, detects changes between permanent flooded water areas by exploiting temporal differences among flood events extracted different sensors. To demonstrate potential proposed approach, generated OMBRIA, bitemporal multimodal satellite imagery dataset for image segmentation through supervised binary classification. It consists total number 3.376 images, synthetic aperture radar from Sentinel-1, multispectral Sentinel-2, accompanied with ground-truth images produced data derived experts provided Emergency Management Service European Space Agency Copernicus Program. The covers 23 globe, 2017 2021. We collected, co-registrated preprocessed in Google Earth Engine. validate performance our method, performed benchmarking experiments OMBRIA compared several competitive state-of-the-art techniques. experimental analysis demonstrated formulation able produce high-quality maps, achieving superior over state-of-the-art. provide dataset, well OmbriaNet code at: https://github.com/geodrak/OMBRIA .