In this paper, the spatial dynamic transfer process of soil organic carbon under hydraulic erosion was accurately described using geostatistics Kriging interpolation method to explore the relationship between erosion and carbon balance, and to provide scientific reference for effective control of soil erosion on the slope surface of Pisha sandstone area. The results indicated that: (1) Before and after the rainfall, the organic carbon of each soil layer decreased with the increase of soil layers, showing medium variation and positive spatial autocorrelation. The maximum autocorrelation occurred in the 0—10 cm soil layer. After rainfall, the organic carbon content of 0—20 cm soil layer decreased while that of 20—40 cm soil layer increased, and the variation degree of each soil layer was weaker than that before rainfall. (2) Before and after rainfall, soil organic carbon was fitted with Gaussian model and spherical model respectively, and the range increased with the increase of soil layers. Before and after rainfall, 0—10 cm soil layer showed strong spatial correlation and the nuggrt/sill increases after rain, while 10—40 cm soil layer showed moderate correlation and the nugget/sill decreased after rain. (3) After rainfall, the spatial heterogeneity weakened and the patch area became larger, which was generally manifested by the erosion in the same soil layer. Erosion occurred in the 0—20 cm soil layer, while sedimentation in the 20—40 cm soil layer. (4) Before and after the rainfall, soil organic carbon and clay contents showed a significant positive correlation. A very significant positive correlation was found in 0—10 cm layer before the rainfall, and the correlations increased after the rainfall, showing a very significant positive correlation in all soil layers. However, the correlation between soil organic carbon and silt contents before and after rainfall was weak and not significant, while the correlation between soil organic carbon and sand contents was basically negative.