A study published in Agrosystems Geosciences and Environment Journal, America by Federal University of Mato Grosso, Brazil and Research paper by Gizachew et al., has reported that the land use significantly affects soil properties and soil organic carbon (SOC) in the Brazilian Amazon. Forests have the highest SOC stocks, while pasture and cultivated lands show much lower values. SOC is strongly related to clay, silt, nitrogen, cation exchange capacity, and bulk density, and negatively related to pH. Machine learning models, especially random forest, predict SOC with very high accuracy using these key variables. These results support the use of ML-based SOC mapping for better land management and conservation planning.
Source: https://acsess.onlinelibrary.wiley.com/doi/10.1002/agg2.70356