[Objective] To analyze the depth and distribution of karst soil in order to provide theoretical basis for using ground penetrating radar(GPR) in the research on soil distribution in karst peak-cluster depression area. [Methods] Through laboratory simulation experiment, this study established the relationship between GPR velocity of electromagnetic wave and soil water content in three types of soil texture(sandy clay loam, clay loam, silty clay) in karst region. Through the field measurement of soil texture and water content, the GPR image was calibrated and interpreted based on the right formula. Then the soil depth was estimated and verified by excavation. [Results] We found the theoretical models of cubic polynomial relationship between electromagnetic wave velocity(ν) and soil water content(θ) for the three soil textures. It showed that soil depth in the field was consistent with the results estimated with theoretical model, and the error range was 0—10 cm. The 2-dimension images reflecting soil depths in the different location under the survey line and the 3-dimension images reflecting soil depths distribution in the quadrant were generated using GPR software. Soil mainly distributed within the depth of 50 cm. [Conclusion] GPR technology was feasible to detect the depths and distribution of discontinuous soil in karst region.