Characterization of Biochar Properties Affected by Different Pyrolysis Temperatures Using Visible-Near-Infrared Spectroscopy

Table 4

Calibration and prediction results of partial least squares (PLS) regression models developed for ash, volatile matter (VM), fixed carbon (Fc) contents and calorific value (CV) with BOC-transformed spectra of biochar samples.

Biochar property	NFs^a	Calibration				Prediction
		𝑅^2	RMSEC^b	𝑅^2	RMSECV^c	𝑅^2	RMSEP^d	RPD^e
Ash (wt.%)	8	0.99	0.3	0.94	0.8	0.91	1.0	3.32
VM (wt.%)	3	0.89	2.2	0.87	2.4	0.86	2.4	2.77
Fc (wt.%)	3	0.90	1.8	0.88	2.0	0.86	2.1	2.70
CV (MJ/kg)	3	0.83	0.56	0.80	0.63	0.81	0.60	2.33

^aNumber of factors used in PLS model.
^bRoot mean squared error of calibration.
^cRoot mean squared error of cross-validation.
^dRoot mean squared error of prediction.
^eResidual predictive deviation, which is the ratio of standard deviation of reference values of the prediction set to the RMSEP.