Sen2GF3Floods: A Novel Multi-Source Remote Sensing Dataset for Flood Mapping

Sen2GF3Floods is a newly developed multi-source remote sensing dataset designed to strike a balance among data volume, annotation accuracy, and labeling efficiency. It is the first dataset to combine pre-disaster Sentinel-2 optical imagery with post-disaster Gaofen-3 (GF-3) synthetic aperture radar (SAR) data. The final dataset comprises 21,483 standardized samples, capturing nine significant flood events with high annotation quality.The dataset contains six spectral bands: four from Sentinel-2—red (Band 4), green (Band 3), blue (Band 2), and near-infrared (Band 8)—and two SAR bands from GF-3, specifically HH and HV polarizations. All bands are resampled to a uniform spatial resolution of 10 meters to ensure consistency across data sources.The dataset directory is organized into three main folders: sentinel2, gaofen3, and label. The Sentinel2 folder contains pre-disaster optical imagery acquired from Sentinel-2, while the gaofen3 folder includes post-disaster SAR imagery collected by Gaofen-3. The label folder provides the corresponding pixel-level annotation maps indicating flooded and non-flooded regions, ensuring clear spatial alignment with the input image pairs.

Sen2GF3Floods是一款新近研发的多源遥感数据集，旨在兼顾数据体量、标注精度与标注效率。该数据集是首个将灾前哨兵二号（Sentinel-2）光学影像与灾后高分三号（Gaofen-3, GF-3）合成孔径雷达（Synthetic Aperture Radar, SAR）数据进行融合的遥感数据集。最终数据集包含21483份标准化样本，覆盖9起重大洪涝事件，标注质量优异。数据集共包含6个光谱波段：其中4个源自哨兵二号，分别为红波段（Band 4）、绿波段（Band 3）、蓝波段（Band 2）与近红外波段（Band 8）；另外2个波段来自高分三号的SAR载荷，具体为HH与HV极化方式。所有波段均被重采样至统一的10米空间分辨率，以保障不同数据源间的数据一致性。数据集目录分为三个核心文件夹：sentinel2、gaofen3与label。其中sentinel2文件夹存储灾前哨兵二号光学影像，gaofen3文件夹收录灾后由高分三号获取的SAR影像，label文件夹则提供对应的像素级标注图，用于区分洪涝区域与非洪涝区域，且与输入影像对实现精准空间对齐。