One of the difficulties and challenges in research on urban safety disaster mitigation and prevention is the risk assessment of urban storm waterlogging. The numerical simulation evaluation model of urban flooding by high-resolution remote sensing is useful for determining pre-disaster risk， simulating mid-disaster scenario， and determining post-disaster loss of flooding. It also has significant application value for assisting in urban planning and managing flooding emergencies. A numerical simulation model of urban flooding is constructed using the GF-7 satellite stereo photogrammetry， object-oriented classification， and hydrodynamic coupling technique， and the risk evaluation is constructed by combining water depth， flow velocity， impact area， storm frequency， and other characteristics. The model accuracy is better than 82% after testing on the middle and lower reaches of the Dasha River in Shenzhen， and when compared to historical measured data for verification， the flooding risk area is decreased by 3.36% in 2021 compared to 2017. According to the study， a high-precision DEM is essential for increasing the precision of an examination of urban storm waterlogging. The waterlogging depth accuracy of the 2.5 m resolution DEM is 5.31%，23.88%， and 58.09% higher than that of the 5 m，10 m， and 20 m， respectively， based on the results of the 1h simulation. The research results have been applied to the evaluation of sponge city in Shenzhen and achieved better effects， providing scientific means and basis for urban flood disaster risk management.