Soil organic carbon (OC) partitioning between mineral-associated organic carbon (MAOC) and particulate organic carbon (POC) is recognized as a major determinant of soil OC storage and its turnover rate. Infrared (IR) spectroscopy has long been used as a means of modeling and predicting soil OC contents due to distinct fingerprints of soil components in the mid-IR absorbance range. For the goal of modeling the OC content in soil and its fractions associated with MAOC and POC, this study applied a recently proposed novel approach [1] utilizing a “mathematical chromatography” of mid-IR spectra.
Collected mid-IR absorption spectra of forest and fallow soils in Israel and their size-separated fractions, <50µm and >50μm, associated with MAOC and POC, respectively, were analyzed using the nonnegative multivariate curve resolution (MCR). The nonnegative MCR technique following the Bouguer–Lambert–Beer law, decomposes spectra into additive contributions of chemically interpretable components, e.g., associated with clay-organic complexes, carbonates, and organic matter, which were then used to represent OC contents within a physicochemical model, as distinct from commonly employed empirical statistical models. Meaningful linear relationships were obtained between modelled and measured OC contents, with the R² values of 0.56 for MAOC, 0.72 for POC, and 0.69 for the whole soils. The major potential of the proposed approach is seen in the possibility to model the OC contents of soil and its fractions in terms of OC contents of the chemically meaningful independent components.
