SPECTRAL MODELING FOR THE QUANTIFICATION OF PARTICULATE ORGANIC CARBON AND MINERAL-ASSOCIATED ORGANIC MATTER IN SOIL

This study explored the use of visible near-infrared (vis-NIR-SWIR) and mid-infrared (Mid-IR) spectroscopy to quantify particulate organic carbon (POM) and mineral-associated organic carbon (MAOM), essential for understanding soil carbon dynamics and health. Conducted in São Paulo, Brazil, with 35 soil samples, the study applied vis-NIR-SWIR and Mid-IR spectroscopy to identify spectral bands associated with these carbon fractions. Data analysis with machine learning algorithms indicated greater accuracy in predicting MAOM, with a Spearman correlation index of 0.74, highlighting its stability. In contrast, POM, more susceptible to environmental variations, showed a lower correlation of 0.28. Spatial predictions revealed that Mid-IR spectroscopy was especially effective for mapping MAOM, with a validation R² of 0.65, compared to 0.53 for POM. These findings support spectroscopic methods as efficient tools for mapping stable carbon fractions, promoting sustainable soil management and agricultural practices that enhance carbon conservation.