AbstractAccurate prediction of subsoil compaction in cropland areas is important for food security and sustainable agricultural production development. This study evaluated the validity and advantages of the soil compaction stratification ratio (SR) as an indicator for assessing the risk of subsoil compaction and presented a theoretical framework for predicting and evaluating subsoil compaction risk. Geographic information system and remote sensing technologies were utilized in combination with a state‐of‐the‐art ensemble machine learning model to quantitatively predict the spatial distribution and uncertainty of subsoil compaction risk in croplands. The results showed that the subsoil bulk density (BD) was significantly greater than the topsoil BD in Shaanxi Province, China. Compared with that of the subsoil BD, the stratification ratio of the bulk density (SRBD) exhibited greater sensitivity to subsoil compaction, and it accurately reflected spatial variations in subsoil compactness. The SRBD exhibited obvious spatial heterogeneity, and the mean annual precipitation, elevation, and mean land surface temperature substantially influenced its spatial distribution. The critical thresholds of the SRBD for the low‐risk, medium‐risk, and high‐risk groups were 1.01, 1.21, and 1.32, respectively. Low subsoil compaction dominated in the mountainous and hilly areas, whereas the Guanzhong Plain and Hanzhong Basin exhibited medium levels of subsoil compaction. Some croplands in the central Guanzhong Plain experienced severe subsoil compaction issues. In conclusion, the soil compaction SR was found to be an effective indicator of subsoil compaction by effectively eliminating soil background differences. This approach has potential applications in cropland management, digital soil mapping, and early warning of soil compaction risk.
Read full abstract