In order to explore the main influence factors of the node water depth of rainwater pipe network located in a campus in Lanzhou city and to study the interpolation method suitable for the node water depth data, we analyze the data of the water depth of 246 nodes of the rainwater pipe network in the study area under the rainfall return periods of 7 years and 50 years (i.e., P=7 a and P=50 a, respectively). Using ArcGIS software, we make spatial difference analysis of node water depth data of rainfalls in the two return periods, and conduct a comparative study on the four Kriging interpolation models, i.e. stable model, spherical model, Gaussian model and index model, by means of cross-validation method. The results show that the Pearson correlations between the maximum depth of nodes and the depth of water accumulation for P=7 a and P=50 a are strong, and the correlation coefficients are 0.605 and 0.766, respectively. Through comparing the predicted values with the measured values of four Kriging semi-variation function models, it is found that the Kriged reduced mean error by Gaussian model is minimum (KRME=-0.87×10<sup>-4</sup> at P=7 a and KRME=0.87×10<sup>-3</sup> at P=50 a), and the Kriged reduced mean square error is the best (KRMSE=0.939 at P=7 a and KRMSE=0.947 at P=50 a). The Gaussian model is determined as the most suitable model for Kriging interpolation method of node water depth in the study area, which provides a more effective method to identify water logging in the study area by using limited waterlogging data, and provides a theoretical basis for the control and reduction measures of waterlogging.