Table 4 Prediction of GWP of biomass of fast-growing tree and agriculture residue by PLSR.

Variable selection method	Spectral pretreatment	Number of wavenumber	Factor	Calibration set		Prediction set
Variable selection method	Spectral pretreatment	Number of wavenumber	Factor	R²_c	RMSEC	R²_p	RMSEP	Bias	RPD
Full wavelength	Raw	1150	10	0.80	0.00054	0.76	0.00061	-0.0004	2.7
	SNV	1150	8	0.90	0.00053	0.82	0.00065	-0.0007	2.5
	FstDev	1150	9	0.89	0.00055	0.82	0.00062	-0.0006	2.6
	SecDev	1150	9	0.87	0.00054	0.84	0.00062	-0.0004	2.6
CM	Raw	220	10	0.80	0.00054	0.76	0.00063	-0.0002	2.6
	SNV	220	8	0.92	0.00053	0.80	0.00062	-0.0004	2.6
	FstDev	325	9	0.91	0.00054	0.74	0.00068	-0.0003	2.4
	SecDev	555	9	0.89	0.00053	0.83	0.00062	-0.0005	2.6
VM	Raw	520	10	0.93	0.00065	0.80	0.00071	-0.00005	2.3
	SNV	795	8	0.79	0.00082	0.85	0.00061	-0.00004	2.7
	FstDev	590	9	0.91	0.0006	0.74	0.00083	-0.00006	1.9
	SecDev	885	9	0.89	0.0006	0.82	0.00061	-0.00003	2.6
COVM	Raw	510	10	0.89	0.00061	0.82	0.00063	-0.00071	2.6
	SNV	895	8	0.95	0.00039	0.78	0.00076	-0.00011	2.1
	FstDev	325	9	0.92	0.00053	0.85	0.00063	-0.00014	2.6
	SecDev	450	9	0.92	0.00053	0.81	0.00071	3.4553e-05	2.3
VIP	Raw	950	10	0.80	0.00084	0.76	0.00062	-0.00014	2,6
	SNV	850	8	0.78	0.00084	0.84	0.00063	-0.00011	2.6
	FstDev	750	9	0.91	0.00053	0.74	0.00083	-.0.00012	1.9
	SecDev	985	9	0.89	0.00061	0.82	0.00068	-5.1928e-05	2.4

N: Number of samples in calibration set, R²_c: coefficient of determination of calibration set, n: number of samples in prediction set, R²_p: coefficient of determination of prediction set, RPD: ratio of prediction to deviation, CM: correlation method, VM: variance method, COVM: co-variance method, VIP: variable.
Significant values are in [bold].

Quick links

Search