Bitemporal and Dual-Polarization Synthetic Aperture Radar (SAR) for Dynamic Flood Mapping and Socioeconomic Analysis Datasets in Bangladesh from 2015 to 2024

Dataset Overview: This dataset is designed for dynamic flood mapping and socioeconomic vulnerability analysis in Bangladesh, using bitemporal and dual-polarization Synthetic Aperture Radar (SAR) imagery collected between 2015 and 2024. It is aimed at deep learning applications to assess flood impacts and societal vulnerability. Key Components: SAR Data for Water Increase Segmentation: The "dataset_for_DeepLearning" includes pre-post flooding event of dual-polarization Sentinel-1 SAR images (VV and VH bands) at a 30-meter resolution, each with dimensions of 256x256 pixels. The data focuses on flood-prone areas, i.e., regions that have flooded relative to pre-flood conditions. These images are paired with labels indicating areas of water increase, suitable for deep learning model training, validation, and testing. National-scale Flood Binary Maps: The "Bangladesh_Data" component contains national-scale flood binary maps, with monthly resolution for 2015-2016 and bi-weekly resolution for 2017-2024, showing areas of flood presence and absence. Socioeconomic Data: The dataset also includes land-use, nighttime lights, and population data, essential for analyzing the impact of flooding on socioeconomic vulnerability and assessing risk across different regions.