Table 2.

Performance of PLSR models for predicting soil physicochemical properties under different spectral preprocessing methods.

Properties	Preprocessing	PCs	Rc	RMSEC	Rp	RMSEP	RPD
pH	Raw	12	0.78	0.46	0.71	0.47	1.219
	S-G	8	0.67	0.55	0.74	0.43	1.075
	SNV	11	0.73	0.48	0.71	0.50	1.079
	Medfilt	13	0.76	0.46	0.85	0.41	1.416
	Normalize	12	0.71	0.49	0.83	0.42	1.311
SOM	Raw	10	0.92	3.52	0.93	3.80	2.466
	S-G	10	0.93	3.43	0.93	3.75	2.535
	SNV	9	0.93	3.66	0.92	3.84	2.592
	Medfilt	12	0.95	3.07	0.95	3.04	3.118
	Normalize	9	0.92	3.71	0.94	3.50	2.815
AN	Raw	11	0.82	31.77	0.80	38.71	1.330
	S-G	9	0.80	32.61	0.79	40.07	1.282
	SNV	10	0.82	33.91	0.80	34.38	1.498
	Medfilt	12	0.85	29.17	0.85	36.13	1.429
	Normalize	10	0.82	33.66	0.84	33.01	1.536
AP	Raw	10	0.75	23.25	0.79	23.93	1.330
	S-G	10	0.76	22.77	0.79	23.94	1.360
	SNV	10	0.73	23.33	0.73	28.22	1.030
	Medfilt	13	0.79	22.70	0.79	22.09	1.397
	Normalize	10	0.73	23.80	0.79	24.46	1.221
AK	Raw	11	0.80	54.24	0.81	70.16	1.212
	S-G	10	0.80	54.13	0.80	71.26	1.214
	SNV	10	0.77	65.96	0.81	56.47	1.372
	Medfilt	14	0.82	53.82	0.84	67.31	1.481
	Normalize	10	0.78	61.58	0.83	59.87	1.386

Bold values indicate the optimal preprocessing method for each indicator.