Influence of Spatial Resolution of Rainfall Data on Flood Peak in Urban Areas

Due to climate change and urbanization, rainstorms are becoming more and more frequent and highly unevenly distributed in urban areas. However, it is difficult to have sufficient monitoring data for rainfall events because of insufficient installments of monitoring stations. Usually, the spatial resolution of meteorological forecast data is still coarse, which leads to the low accuracy of simulation and prediction results from urban flood models. Therefore, it is necessary to improve the spatial resolution of rainfall data to enhance the accuracy of urban flood simulation. In this paper, three catchment basins in Shenzhen River Basin are selected. Analytical data are extracted from 118 rainfall events lasting no less than 5 minutes from 31 meteorological stations within and around the study area during 2018-2020. The discrete observation data from meteorological stations are interpolated into continuous grids. According to the drainage zoning area, the grid size is initially set as 200 m×200 m. The gridded data is then aggregated into 400 m×400 m, 800 m×800 m, 1600 m×1600 m, 3200 m×3200 m, and 6400 m× 6400 m to achieve coarser spatial resolution rainfall datasets. The basin boundary is used during the aggregation to ensure that the average rainfall in the basin remains constant. The gridded rainfall data at different spatial resolutions are used as the input of an urban flood model to simulate the flooding processes of outlet sections in three watersheds. The model results of three watersheds are compared finally. By comparing the changes of flood peak at the outlets of each basin, the influence of rainfall spatial resolution on river flood peak is further analyzed by combing the spatial parameters such as rainfall center and basin center. The results of three watersheds show that for 118 rainfall processes, with the decrease of spatial resolution, the amplitude of flood at the watershed outlets gradually increases. When rainfall is evenly distributed in the whole basin, the variation ranges of flood peaks in the three basins are 5.04%, 8.73%, and 7.11%, respectively, compared to the rainfall data with 200 m× 200 m grid size. The influence of spatial resolution is strongly correlated with the relationship between rainfall center and watershed geometric center, which strongly relates to the direction from watershed geometric center to watershed outlet. The research provides a reference for improving the accuracy of urban flood models by involving the variable of rainfall, which enable researchers to better assess the risk of river floods. © 2022, Science Press. All right reserved.