Runoff simulation is greatly impacted by the quantity and arrangement of rain gauges, especially in difficult terrain like the southwest mountainous area where there are frequent flash floods and unevenly distributed rain gauges. The Dumu River Basin, a common watershed in the upper levels of the Wujiang River that is vulnerable to flash floods, was chosen as the subject region for this investigation. The semi-distributed hydrological model TOPMODEL, which is easy to integrate and carry out a large number of stochastic simulations, was selected as the carrier for the rainfall and runoff observation data in the basin. This study quantitatively demonstrates the impact of the number and distribution characteristics of rain gauges on the simulation accuracy of typical karst basins using a variety of error analysis indicators and rain gage network spatial distribution description indicators. According to the findings, the range shows a negative exponential square decay, whereas the mean, standard deviation, coefficient of variation, minimum value, and maximum value all change exponentially. The maximum values of the NSE and correlation coefficient stay comparatively constant, suggesting that by optimizing the network distribution of rain gauges, great simulation accuracy can be attained even with a small number of rain gauges. All hydrological model indicators show a quadratic function relationship with the nearest neighbor index (NNI). Information entropy shows the same trend as correlation coefficients and NSE with regard to the number of rain gauges, and it has a greater correlation with error indicators than NNI. The maximum model accuracy is achieved when NNI falls between 1 and 2. A certain amount of uncertainty is present in both NNI and information entropy: the same value may accompany multiple simulation accuracies. This suggests that while establishing a rain gage network with a single indicator can effectively analyze areal rainfall, it may not always accurately simulate runoff. The results of the study can serve as technical references for water resource assessment, flood analysis and forecasting, and network optimization for regional rain gauges.