Abstract: 【Objective】 The purpose of this research was to study the application of digital soil mapping technology to making high definition soil map in land survey.【Method】 In this study, the soil type of the drainage basin in the Peilou Valley of the Qilian Mountains was taken as the research object, and the spectral index and topographic factor of remote sensing were extracted from Gaofen-2 remote sensing data and DEM data as environmental cooperative variables. The soil type mapping model was established based on C5. 0 decision tree algorithm, CART decision tree algorithm and support vector machine method respectively, and the soil type distribution map of the study area was made. The cartographic results of the three methods were compared and evaluated.【Result】 The results showed that the overall mapping accuracy of C5. 0 decision tree algorithm was 89%,which was significantly higher than that of the other two methods.【Conclusion】 Our results suggested that the proposed method was useful in the study area, and it could improve the accuracy of watershed digital soil mapping and provide scientific reference for the digital soil mapping technology in the area.