【Objective】The thickness of the loess layer overlying the Pisha sandstone area is crucial for regional ecological restoration; however, efficient detection methods are still lacking. Although ground-penetrating radar (GPR), owing to its rapid and non-destructive advantages, has been successfully applied in areas underlain by residual parent material, its suitability for detecting the soil–rock interface in regions covered by transported parent material (loess) has not yet been systematically validated. This study therefore aimed to evaluate the applicability of GPR for loess thickness determination in the Pisha sandstone area and to assess the performance of the Topp model in estimating the dielectric permittivity of loess in this region. 【Method】GPR was employed to measure loess thickness on three typical vegetation slopes (shrubland, arbor land, and grassland) in the Pisha sandstone area. The dielectric permittivity of loess was calibrated in situ through field excavation and pre?buried iron pipes. The measured permittivity values were compared with predictions from the Topp model to assess the model’s performance in this region, and the GPR-derived loess thicknesses were compared with actual excavation depths to evaluate the measurement accuracy of GPR. 【Result】The results showed that the measured relative dielectric permittivities of loess were 13.33, 19.57, and 13.92 for shrubland, arbor land, and grassland, respectively. The Topp model predicted loess dielectric permittivity with an accuracy ranging from 92.51% to 97.23%, showing no significant difference from the measured values. Also, the accuracy of GPR in determining loess thickness ranged from 87.16% to 97.78%, and the measured depths did not differ significantly from the actual burial depths. On all three slopes, loess thickness exhibited a clear increasing trend from the upper to the lower slope positions. The grassland had the largest average thickness (36.16 cm), while the arbor land had the smallest (19.06 cm). 【Conclusion】This study provides an efficient and accurate method for soil thickness investigation in the Pisha sandstone area and offers fundamental data support for soil and water conservation and hydrological process research in this region.