太湖流域是我国经济最发达地区之一，因其河网复杂、地势平坦，洪灾频发且损失严重，建立该流域洪水淹没要素的高精度快速预测方法是其防洪应急的重大需求。针对太湖流域洪灾前高分辨率数字表面模型DSM缺乏、降雨径流量预测精度低、高分辨率DSM地形条件下洪水淹没连通区域分析精度及效率有待提升的问题，本项目研究ICESat-2辅助国产新型高分辨率卫星影像的灾前高分辨率DSM构建方法，为太湖流域洪水淹没快速预测提供精细的地形参数；构建改进LSTM时序深度网络的降雨径流量高精度预测方法，为太湖流域洪水淹没快速预测提供高精度的降雨产水量参数；构建与高分辨率DSM相适应的洪水淹没连通区域分析和水面高程推算方法，实现洪水淹没要素的快速精准预测；以太湖流域杭嘉湖区2018年“云雀”台风暴雨洪水为例，验证所提方法的有效性。研究成果将为太湖流域防汛抗旱总指挥部提供高精度高时效性的洪水淹没预测信息，为科学防洪减灾提供支撑。

The Taihu Lake Basin is one of the most economically developed regions in China. Due to its complicated river network, flat terrain, it floods frequently and often causes severe losses. The establishment of an accurate and rapid prediction method for flood inundation elements in the basin is a major requirement for flood control and emergency response. Facing the problems that the Taihu Lake Basin is in the lack of high-resolution Digital Surface Model (DSM) before the flood, the accuracy of rainfall runoff prediction is low, and the accuracy and efficiency of the analysis of flood-connected areas under the terrain condition of high resolution DSM need to be improved, three main contents will be investigated in this project: (1) the method for constructing high-resolution and high-precision DSM in Taihu Lake Basin using Chinese new high-resolution stereo satellite images assisted by the Ice, Cloud and Land Elevation Satellite-2 (ICESat-2) laser points, providing high-resolution topographic parameter for the fast flood inundation prediction; (2) an improved Long Short-term Memory (LSTM) neural network model for high-precision prediction of rainfall runoff, providing accurate rainfall yield parameter for the fast flood inundation prediction; (3) a flood-connected area search and water surface elevation calculation method that adapts to the high-resolution DSM, realizing the fast and accurate flood inundation prediction based on the high-resolution DSM and flood volume. The effectiveness of the proposed method will be verified by the example of the flood caused by Typhoon Jongdari in the Hangjiahu area of Taihu Lake Basin in 2018. The research results will provide high-accuracy and timely flood prediction information to Taihu Lake Basin Flood Control and Drought Relief Headquarters, and provide support for scientific flood control and disaster reduction.