As a dynamic simulated urban stormwater accumulation model, SWMM is widely used in urban accumulation load estimation research and has achieved good results in many small-scale studies. As China"s urban development continues to accelerate, the spatial scale of urban accumulation research has also become larger. How to construct the model scientifically and reasonably, and to ensure the accuracy of the model has become an important issue in research. This paper mainly analyzes the impact of geographic national condition monitoring data on model accuracy when SWMM was applied to urban area storm accumulation simulation. On the basis of this, in the three aspects of watershed division, pipe network generalization and node selection, the parameters were optimized. Using the optimized model to simulate under three real rainfalls, the results showed that the overflow node and the overloaded pipe section were basically in line with the street flooding situation. Finally, this model was used to divide the risk area of Yan"an City, which further illustrates the practical value of the optimized model.