Modeling of Soil Organic Carbon Fractions Using Visible-Near-Infrared Spectroscopy

There is a pressing need for rapid and cost-effective tools to estimate soil C across larger landscapes. Visible-near-infrared diffuse reflectance spectroscopy (VNIRS) offers comparable levels of accuracy to conventional laboratory methods for estimating various soil properties. We used VNIRS to estimate soil total organic C (TC) and four organic C fractions in 141 samples collected in the Santa Fe River watershed of Florida. The C fractions measured were (in order of decreasing potential residence time in soils): recalcitrant C (RC), hydrolyzable C (HQ, hot-water-soluble C (SC), and mineralizable C (MC). Soil samples were scanned in the visible-near-infrared spectral range. Six preprocessing transformations were applied to the soil reflectance, and five multivariate techniques were tested to model soil TC and the organic C fractions: stepwise multiple linear regression (SMLR), principal components regression, partial least squares regression (PLSR), regression tree, and committee trees. Total organic C was estimated with the highest accuracy, obtaining a coefficient of determination using a validation set (R-v(2)) of 0.86, followed by RC (R-v(2) = 0.82), both using PLSR. The SC fraction was modeled best by SMLR (R-v(2) = 0.70), while PLSR produced the best models of MC (R-v(2) = 0.65) and HC (R-v(2) = 0.40). The addition of TC as a predictor improved the VNIRS models of the soil organic C fractions. Our study indicates the suitability of VNIRS to quantify soil organic C pools with widely varying turnover times in soils, which are important in the context of C sequestration and climate change.