全球复杂场景洪水多模态遥感数据集

全球复杂场景洪水多模态数据集包含6大洲158个城市或地区的多光谱影像、SAR影像、DEM及洪水标签等数据集。该数据集覆盖了多样的土地覆盖类型、季节变化及高程范围，旨在全面反映全球复杂场景下的洪水特征，其被处理为易于存储和训练深度学习模型的数据结构。训练区域、验证区域与测试区域相互独立，保证了模型性能验证的客观性。基于该数据集进一步的开发多模态数据互补特征融合算法，可实现多模态数据互补的全球复杂场景下洪水精准检测。该数据集助力应急管理部门在洪涝灾害发生时迅速响应，开展洪涝灾情应急监测工作，帮助政府有效部署救援任务，减少洪涝灾害造成的人口损失和经济损失。其为实现可持续发展目标提供数据基础和技术支持。

The Global Multi-Modal Flood Dataset for Complex Scenarios encompasses multispectral imagery, SAR imagery, DEM data, flood labels and other related datasets from 158 cities and regions across six continents. This dataset covers diverse land cover types, seasonal variations and elevation ranges, aiming to comprehensively reflect flood characteristics in global complex scenarios, and has been structured into a format that facilitates storage and deep learning model training. The training, validation and test regions are mutually independent, which ensures the objectivity of model performance evaluation. Further development of multi-modal complementary feature fusion algorithms based on this dataset can enable accurate flood detection in global complex scenarios by leveraging complementary information from multi-modal data. This dataset supports emergency management departments in conducting rapid response and emergency monitoring of flood disasters when they occur, helping governments effectively deploy rescue missions and mitigate population and economic losses caused by floods. Moreover, it provides foundational data and technical support for the achievement of the Sustainable Development Goals (SDGs).