针对降雨时空不确定性严重制约洪涝模型模拟精度的问题，在深圳市布吉河流域研究降雨中心、降雨移动方向等时空特征对洪水过程的影响。利用2018—2020年的降雨监测数据，划分并筛选出116个大雨以上的降雨场次，进而对每个场次通过空间平均化处理和时间同步处理，分别构建空间平均和雨峰同步降雨过程；利用城市洪涝模型模拟116个降雨场次在实际降雨、空间平均、雨峰同步3种时空分布下的洪水过程；分析降雨中心和降雨移动方向对洪峰流量、洪水过程的影响。结果显示：经过空间平均化处理后，产生的洪峰流量较实际降雨模拟得到的洪峰流量变化范围为-43.9%～34.7%，而且当实际降雨中心位于流域几何中心下游时，模拟得到的洪峰倾向于偏高；经过雨峰同步处理的降雨较实际降雨模拟得到洪峰流量变化范围为-33.7%～20.0%，洪峰变化范围与降雨中心移动方向有较强的相关性；洪水过程与洪峰流量变化相对应，表现为洪峰增加、洪水持续时间缩短,或洪峰降低、洪峰持续时长增加。研究结论可为洪涝模型输入的降雨过程构建提供依据，为河道洪水风险评估、洪涝过程实时模拟提供借鉴。

There is an increase of extreme rainfall events frequency in cities due to global climate change and rapid urbanization,which brings urban flood disasters.The temporal and spatial heterogeneity of extreme rainfall events in urban areas became more and more obvious in recent years,mainly manifested in the aggravation of spatial heterogeneity and more concentration temporally and spatially in the rainfall process.Owing to temporal and spatial heterogeneity,it is difficult to evaluate and predict the flooding process in urban areas,especially with sudden rainstorm events,which becomes an important issue for improving the accuracy of the urban flood model. Buji River basin is located in Shenzhen,China,which covers 40.59 square kilometers area,and the built-up area in the region exceeds 70%.To improve the urban flood model concerning temporal and spatial heterogeneity,116 heavy rainfalls events were extracted from the monitoring data of 5-minute intervals of meteorological stations from 2018 to 2020,and the discrete station monitoring data was interpolated into spatially continuously distributed grid data by a spatial interpolation algorithm.The spatial average rainfalls and rain peak synchronous rainfalls were constructed with spatial averaging and time synchronization method.The spatial averaging method was to average the rainfall data every 5-minute for the whole basin.The time synchronous method was realized by adjusting the phase of the rainfall process in each grid,to ensure that the rain peaks of each grid appear at the same time.A urban flood model was used to simulate the flooding process of the outlet section of the watershed caused by 116 monitoring rainfall and 232 rainfall processes after spatial averaging and time synchronization,and the flood peak change index of each rainfall was extracted.The rainfall center and rainfall movement direction of 116 rainfall events were counted,and their effects on peak discharge and flood process were analyzed. The results showed that after spatial averaging,the variation of flood peak was -43.9%~34.7%,with an average of -3.3% compared with the flood peak obtained from the original rainfall.The flood peak was varied due to the spatial relationship between the rainfall center and geometric center of the basin.The flood peak was mainly reduced with the rainfall center located upstream of the geometric center of the basin,after spatial averaging,and the flood peak was mainly reduced with the rainfall center located in the opposite position of the geometric center of the basin.For the 65 rainfalls in the northeast (upstream) of the geometric center,the average variation of flood peak was -15.1%,and for the 51 rainfalls in the southwest side of the geometric center,the average change of flood peak was 3.1%.Compared with the original rainfalls,the variation of flood peak obtained by the rainfall peak synchronous processing is -33.7%~20.0%,with an average of -3.0%.The variation of flood peak was related to the direction of rainfall movement,with the average variation of flood peak 14 rainfalls moving from northwest to the southeast was -7.3%.The average variation of 12 rainfalls moving from northeast to southwest was -2.2% and the average variation of 20 rainfalls moving from southwest to northeast was 2%.For the flooding process,when the flood peak increased,the flood peak duration was shortened.On the contrary,when the flood peak decreased,the flood peak duration increased. Under the same total rainfall,compared with the original rainfall,the variation of flood peak caused by spatial average was -43.9%~34.7%,and the variation of flood peak caused by synchronous rainfall was -33.7%~20.0%.The variation of flood peak with spatial averaging was mainly due to the spatial relationship between the center of original rainfall and the spatial center of the basin.The variation of flood peaks caused by the synchronization was related to the moving direction of rainfall.The flood peak was more likely to increase after synchronization with the moving direction of rainfall opposite to the confluence direction.The references can be provided for flood simulation calculation and flood risk assessment in an urban watershed.