Abstract: Taking the rock slope of the subway deep foundation pit as the research object, elastic wave CT is used to preliminarily judge the type and spatial distribution of rock mass diseases in foundation pit, which is then verified by geological radar. Furtherly the a watershed algorithm is adopted to analyze the wave velocity distribution obtained by elastic wave CT in order to extract the detailed sound velocity change at the rock cave and delineate its range. On this basis, based on the coordinate information of elastic wave CT three-dimensional spatial wave velocity database, combined with the spatial coordinate information obtained by watershed algorithm, the modeling database is constructed, after imported into GOCAD software, the three-dimensional geological visualization model is established after processing and analysis. The rock mass disease model established in this paper is highly consistent with geological drilling, peephole imaging and field observation. The proposed rock disease identification method and modeling technology can not only provide an important reference for rock mass support but also provide a basis for water disaster prevention of urban underground engineering.