Quantitative Prediction of Biochar Soil Amendments by Near-Infrared Reflectance Spectroscopy

Any sale of C credits in a cap-and-trade system based on soil biochar applications will require an effective and inexpensive audit system. We investigated the ability of near-infrared reflectance spectroscopy (NIRS) to predict biochar amendment levels, total C, and C/N ratios within and between two independent soil sample sets collected from an agricultural field in Boone County, Iowa. One set had high intrinsic diversity of soil properties, while the other set had low diversity. The calibration cross-validation procedure showed the ability of NIRS to quantitatively model total C and biochar C along with their normalized values with R2- values >0.82 and ratios of prediction deviation (RPDs) >2.2 for the high-diversity set. Modeling total C of the no-biochar controls (R-2 = 0.69, RPD = 1.56) and normalized total C (R-2 = 0.51, RPD = 1.31) from the low-diversity sample set was less accurate. Validation using the independent partial least squares regression models of the high-diversity sample set to predict constituent concentrations for the low-diversity sample set showed predictions of total C, biochar C, and normalized biochar C with bias-corrected RPDs >2.1 and R-2 values >0.87. Model validation R-2 (0.92) between the measured biochar C and model-predicted normalized biochar C was significantly greater (alpha < 0.05) than the autocorrelation between the measured total C and biochar C (R-2 = 0.89), thus demonstrating that NIRS responds to biochar C apart from total C. The results indicate that inexpensive NIRS analyses can be used to audit soil biochar applications.