Can a ramped high-temperature carbon analyser with thermal oxidation be used to quantify soil organic carbon pools?

Abstract

Soil organic carbon (SOC) pools range from easily decomposable particulate, slowly decomposable humic, and resistant organic carbon. These carbon pools, sometimes referred to as POC, HOC and ROC, respectively, can be used as inputs into carbon accounting applications such as the Roth-C model. This study examines whether ramped thermal oxidation, combined with evolved gas (CO₂) analysis (RTO-EGA) of soil samples at temperature intervals between 110 °C and 1000 °C, can quantify the allocation of SOC to these pools. RTO-EGA ramping profiles of 176 samples were combined with respective soil carbon pool data obtained from physical fractionation and ¹³C NMR analysis, into multiple linear regression (MLR) and partial least squares regression (PLSR) models. The samples were split into calibration (110) and validation (51) sets for modelling, and 15 samples omitted due to excessively low decomposition temperatures and associated high model leverage. The samples were also modelled by PLSR using mid-infrared diffuse reflectance (DRIFT) for comparison with RTO-EGA. RTO-EGA and DRIFT calibration and validation accuracies were generally high for POC and HOC, with R² values of 0.80-0.95 and RMSE values of 0.10-0.20 %C. Estimates of the ROC pool were relatively less accurate for samples not adequately covered by the calibration set. Since high-temperature carbon analysers are routinely used in most soil testing laboratories, the RTO-EGA approach can be easily adopted for carbon pool analysis. This proof-of-concept study demonstrated that the RTO-EGA method, combined with chemometric modelling, as a rapid and cost-effective method to quantify carbon pools.