Studying the tendency of soil temperature and moisture can provide scientific support for setting soil parameters for land surface hydrological model. Previous research has shown that there is long‐range correlation of soil temperature or moisture; however, driving factors of long‐range correlation variations and the cross correlation between temperature and moisture under multiple timescales have rarely been discussed. To explore the driving factors, we compared long‐range correlation between shady and sunny slopes. Adaptive fractal analysis was utilized to analyze long‐range cross correlation (LRCC) and long‐range correlation (LRC). Experimental data were collected between 9 August 2013 and 3 December 2014 at A'rou observation stations, Babao River Basin. Results show that two slopes have some common characteristics. (1) Soil temperature and moisture display the consistent change. (2) The persistence intensity of moisture, temperature, and their interaction intensity change at 9, 6, and 6‐day timescales, respectively. (3) LRC and LRCC indices demonstrate the stratification heterogeneity. Results also show that LRC and LRCC indices on two slopes have differences. The soil system was stratified into three layers to explain these differences over large timescales. (1) At 4‐ to 20‐cm depth, with significant influence of solar radiation and evapotranspiration on sunny slope, moisture shows weaker persistence. There is stronger antipersistence of temperature and stronger hydrothermal interaction on sunny slope. (2) At 20‐ to 80‐cm depth, there is weaker persistence of soil moisture on shady slope because of unstable groundwater and side runoff supply. (3) At 80‐ to 160‐cm depth, with seasonal melt‐water effect on sunny slope, persistence of moisture was considerably weaker.