为研究降雨分布不均对水库近坝区洪水预报精度的影响，以五强溪近坝区流域为研究区域，选取 2014—2020 年 20 场历史洪水资料，采用反距离权重法绘制流域暴雨中心图，并据此进行子流域划分，其中，13 场 洪水用于模型率定，7 场洪水用于模型验证。选取 2021 年 4 场洪水进行检验，对比分析仅依据自然子流域划分和 考虑暴雨中心的洪水预报结果。结果表明：两种子流域划分方法在其最优参数下的洪水模拟精度相似，考虑暴雨 中心划分的洪水模拟确定性系数均值为 0.83，略高于仅依据自然子流域划分洪水模拟确定性系数，4 场洪峰误差 均在 10％以内，洪量误差均在 20％以内，达到了甲级精度标准，说明考虑暴雨中心对子流域进行划分是合理的。

Floods in near-dam areas due to spatial and temporally uneven rainfall are often difficult to forecast effectively with conventional forecasting schemes, where the non-uniformity of rainfall at the basin level is an important factor affecting the accuracy of flood forecasting. There have been many studies on sub-basin division methods, but the problem of uneven rainfall distribution has rarely been studied. The spatial distribution of rainfall represented by the location of the storm center is used to refine the sub-basin division, and to improve the accuracy of flood forecasting in the near-dam area.Taking the watershed near-dam area of Wuqiangxi by extracting from the SRTMDEM 90M digital elevation model, and based on the rainfall data, the storm center map was plotted using the inverse distance weighting method. Storm center locations were considered and used to precisely divide the sub-basins. A parameter calibration method combining an automatic selection of the Monte Carlo random sampling model and manual debugging was used in the parameter calibration process. Flood simulations were carried out based on the Xin'anjiang model, while the results were used to compare with those based on the natural sub-basin division method. Twenty historical floods from 2014 to 2020 were selected, of which 13 floods were used for model calibration and 7 floods for model validation. Four floods from 2021 were selected for testing, and the comparative analysis was based on the results of the natural sub-basin flood forecasts and the results of the flood forecasts considering the location of the storm center.The results showed that sub-flood simulation based on natural sub-basins only failed for one flood in terms of flood volume error and two floods in terms of flood peak error, while the sub-flood simulation taking into account the location of the center of the storm for the sub-basins passed in terms of flood volume error and flood peak error for both periods (rate and test). Four flood simulations for 2021 that were used for testing showed the mean value of the coefficient of certainty for the sub-basins considering the location of the storm center was 0.82, which was higher than the mean value of the coefficient of certainty for the sub-basins based on natural sub-basins, which was 0.72. The results of the four flood simulations considering the location of the storm center to divide the sub-basins were within 10% of the peak error and within 20% of the flood volume error, which reached the accuracy standard of Class A. This showed that it was reasonable to divide the sub-basins considering the location of the storm center.A new method is provided for the division of sub-basins in the near-dam area of reservoirs. The method of dividing sub-basins by considering the location of the storm center can be applied to flood forecasting in the near-dam area, thus achieving the effect of improving forecast accuracy, providing a basis for flood control departments to make decisions on disaster prevention and mitigation, and reducing the losses caused by floods in the basin, with significant economic and social benefits.