Effect of Spectral Configuration on Soil Organic Matter and Electrical Conductivity Predicted by Optimal Band Combination Algorithm

Yi Zhang; Jianli Ding; Zipeng Zhang; Xiangyu Ge; Jinjie Wang

doi:10.3788/LOP202158.2128001

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 21 >Page 2128001 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 21, 2128001 (2021)

Effect of Spectral Configuration on Soil Organic Matter and Electrical Conductivity Predicted by Optimal Band Combination Algorithm

Yi Zhang, Jianli Ding^*, Zipeng Zhang, Xiangyu Ge, and Jinjie Wang

Author Affiliations

College of Resource and Environment Sciences, Xinjiang University, Urumqi , Xinjiang 830046, China

show less

DOI: 10.3788/LOP202158.2128001 Cite this Article Set citation alerts

Yi Zhang, Jianli Ding, Zipeng Zhang, Xiangyu Ge, Jinjie Wang. Effect of Spectral Configuration on Soil Organic Matter and Electrical Conductivity Predicted by Optimal Band Combination Algorithm[J]. Laser & Optoelectronics Progress, 2021, 58(21): 2128001 Copy Citation Text

show less

Fig. 1. Gaussian response function when N=3, R_SR=25 nm, I_SSI=25 nm

Download full size

Principal component projections of 255 samples for different spectral configurations. (a) ASD_1/1; (b) Con_3/10; (c) Con_10/10; (d) Con_20/20; (e) Con_40/40; (f) Con_60/60; (g) Con_100/100

Fig. 2. Principal component projections of 255 samples for different spectral configurations. (a) ASD_1/1; (b) Con_3/10; (c) Con_10/10; (d) Con_20/20; (e) Con_40/40; (f) Con_60/60; (g) Con_100/100

Download full size

Fig. 3. Soil characteristic box diagram of each spectral configuration after excluding outliers and the corresponding descriptive statistics. (a)‒(c) SOM Min, Mean, Max, Std, CV, Ske of entire, calibration, and validation datasets; (d)‒(f) EC Min, Mean, Max, Std, CV, Ske of entire, calibration, and validation datasets

Download full size

Fig. 4. Soil reflectance spectra. (a) SOM with different content; (b) EC with different content

Download full size

Fig. 5. Comparison of operation accuracy and operation time between SOM and EC under different band combinations.(a) SOM; (b) EC

Download full size

Fig. 6. Scatter plots of soil characteristics and measured values simulated by different PLSR models. (a)(b) Optimal spectra; (c)(d) full-band spectra

Download full size

Fig. 7. VIP values obtained from different PLSR models, best spectral parameters used as predictors to estimate soil properties. (a) SOM; (b) EC

Download full size

Configuration	N	400‒1000 nm		1000‒2400 nm
Configuration	N	SSI	SR	SSI	SR
ASD_1/1	2001	1	3	1	10
Con_3/10	339	3	3	10	10
Con_10/10	199	10	10	10	10
Con_20/20	99	20	20	20	20
Con_40/40	49	40	40	40	40
Con_60/60	32	60	60	60	60
Con_100/100	19	100	100	100	100

Table 1. Description of 7 spectral configurations

Configuration	Spectral variable					Maximum determination coefficient（R²）calibration/validation
ASD_1/1	TBI₁（R₂₁₇₀，R₁₈₈₃，R₂₃₂₂）	TBI₂（R₄₀₄，R₆₁₂，R₅₅₂）	TBI₃（R₁₈₇₆，R₂₁₇₀，R₈₄₇）	TBI₄（R₁₄₁₉，R₂₃₇₉，R₉₁₆）	TBI₅（R₁₈₈₀，R₆₅₃，R₂₁₇₀）	0.62/0.58	0.59/0.34	0.68/0.65	0.68/0.69	0.65/0.64
Con_3/10	TBI₁（R₁₇₈₀，R₁₄₈₀，R₁₇₆₀）	TBI₂（R₄₀₄，R₆₄₁，R₅₈₇）	TBI₃（R₁₄₈₀，R₁₇₇₀，R₆₀₈）	TBI₄（R₁₀₁₀，R₁₈₆₀，R₉₅₉）	TBI₅（R₁₄₈₀，R₁₇₇₀，R₇₂₂）	0.63/0.58	0.61/0.36	0.68/0.67	0.66/0.64	0.65/0.65
Con_10/10	TBI₁（R₁₈₈₀，R₂₁₇₀，R₁₉₁₀）	TBI₂（R₄₁₀，R₆₂₀，R₅₇₀）	TBI₃（R₁₈₈₀，R₂₁₇₀，R₇₅₀）	TBI₄（R₁₄₂₀，R₂₃₈₀，R₉₃₀）	TBI₅（R₁₈₈₀，R₂₁₇₀，R₆₂₀）	0.62/0.58	0.59/0.34	0.67/0.68	0.67/0.56	0.64/0.65
Con_20/20	TBI₁（R₁₈₈₀，R₂₁₈₀，R₂₃₂₀）	TBI₂（R₄₂₀，R₆₀₀，R₅₈₀）	TBI₃（R₁₈₈₀，R₂₁₈₀，R₆₂₀）	TBI₄（R₁₄₂₀，R₂₃₈₀，R₉₄₀）	TBI₅（R₁₈₈₀，R₆₈₀，R₂₁₈₀）	0.62/0.63	0.59/0.42	0.66/0.69	0.66/0.64	0.64/0.59
Con_40/40	TBI₁（R₁₈₈₀，R₂₁₆₀，R₁₈₄₀）	TBI₂（R₄₄₀，R₆₀₀，R₆₆₀）	TBI₃（R₁₄₀₀，R₂₁₆₀，R₉₆₀）	TBI₄（R₁₄₀₀，R₂₂₄₀，R₁₀₈₀）	TBI₅（R₁₄₀₀，R₉₆₀，R₂₁₆₀）	0.51/0.56	0.47/0.34	0.56/0.54	0.58/0.54	0.53/0.54
Con_60/60	TBI₁（R₂₂₇₀，R₁₉₁₀，R₆₅₀）	TBI₂（R₄₇₀，R₆₅₀，R₅₉₀）	TBI₃（R₉₅₀，R₂₂₁₀，R₁₄₉₀）	TBI₄（R₁₄₉₀，R₂₃₃₀，R₁₁₃₀）	TBI₅（R₈₉₀，R₂₂₁₀，R₁₄₉₀）	0.34/0.3	0.41/0.15	0.49/0.42	0.48/0.43	0.47/0.41
Con_100/100	TBI₁（R₁₄₀₀，R₇₀₀，R₂₃₀₀）	TBI₂（R₁₄₀₀，R₇₀₀，R₆₀₀）	TBI₃（R₁₈₀₀，R₁₉₀₀，R₂₂₀₀）	TBI₄（R₁₄₀₀，R₂₃₀₀，R₈₀₀）	TBI₅（R₁₅₀₀，R₆₀₀，R₅₀₀）	0.32/0.1	0.35/0.13	0.41/0.2	0.53/0.27	0.38/0.06

Table 2. Best frequency-spectrum parameters of SOM in each spectral configuration derived from optimal band combination algorithm

Configuration	Spectral variable					Maximum determination coefficient（R²）calibration/validation
ASD_1/1	TBI₁（R₂₂₉₂，R₂₃₅₅，R₆₃₄）	TBI₂（R₈₂₆，R₂₀₆₉，R₆₉₈）	TBI₃（R₈₀₃，R₁₀₉₉，R₂₁₆₉）	TBI₄（R₂₃₆₄，R₂₃₁₆，R₂₂₄₁）	TBI₅（R₂₁₇₉，R₂₁₃₄，R₂₁₀₂）	0.33/0.27	0.58/0.5	0.42/0.54	0.37/0.26	0.37/0.34
Con_3/10	TBI₁（R₁₉₆₀，R₁₈₉₀，R₁₉₇₀）	TBI₂（R₇₅₀，R₁₄₈₀，R₁₃₈₀）	TBI₃（R₁₉₂₀，R₉₅₀，R₈₉₃）	TBI₄（R₁₉₆₀，R₁₉₁₀，R₁₈₅₀）	TBI₅（R₇₂₈，R₁₇₃₀，R₄₁₃）	0.29/0.37	0.6/0.61	0.47/0.51	0.3/0.39	0.32/0.16
Con_10/10	TBI₁（R₂₂₉₀，R₂₃₆₀，R₆₉₀）	TBI₂（R₁₁₁₀，R₂₃₁₀，R₆₂₀）	TBI₃（R₂₁₈₀，R₁₁₅₀，R₈₉₀）	TBI₄（R₂₃₆₀，R₂₃₂₀，R₂₂₂₀）	TBI₅（R₂₁₈₀，R₂₁₃₀，R₂₁₀₀）	0.33/0.26	0.58/0.5	0.44/0.55	0.36/0.25	0.36/0.37
Con_20/20	TBI₁（R₂₂₈₀，R₂₃₆₀，R₆₆₀）	TBI₂（R₉₄₀，R₂₁₂₀，R₅₈₀）	TBI₃（R₂₁₂₀，R₁₀₈₀，R₇₆₀）	TBI₄（R₂₃₆₀，R₂₃₂₀，R₂₂₂₀）	TBI₅（R₂₁₈₀，R₂₁₄₀，R₂₁₀₀）	0.31/0.26	0.57/0.5	0.42/0.51	0.34/0.26	0.35/0.3
Con_40/40	TBI₁（R₂₂₈₀，R₂₃₆₀，R₆₄₀）	TBI₂（R₇₆₀，R₂₂₈₀，R₆₄₀）	TBI₃（R₂₁₂₀，R₁₀₈₀，R₇₆₀）	TBI₄（R₂₃₆₀，R₂₃₂₀，R₂₂₈₀）	TBI₅（R₂₂₈₀，R₂₁₆₀，R₁₈₀₀）	0.3/0.31	0.57/0.51	0.4/0.45	0.29/0.28	0.26/0.22
Con_60/60	TBI₁（R₂₂₇₀，R₂₂₁₀，R₂₀₃₀）	TBI₂（R₁₄₃₀，R₂₂₁₀，R₂₂₁₀）	TBI₃（R₁₃₇₀，R₁₄₉₀，R₁₀₇₀）	TBI₄（R₂₂₇₀，R₂₁₅₀，R₁₉₁₀）	TBI₅（R₁₀₁₀，R₂₂₁₀，R₄₇₀）	0.11/0.15	0.56/0.62	0.38/0.41	0.11/0.28	0.22/0.13
Con_100/100	TBI₁（R₁₈₀₀，R₂₂₀₀，R₆₀₀）	TBI₂（R₁₂₀₀，R₂₃₀₀，R₁₉₀₀）	TBI₃（R₁₃₀₀，R₂₂₀₀，R₁₄₀₀）	TBI₄（R₁₃₀₀，R₂₃₀₀，R₉₀₀）	TBI₅（R₁₉₀₀，R₂₁₀₀，R₁₈₀₀）	0.21/0.12	0.54/0.56	0.38/0.21	0.2/0.04	0.25/0.11

Table 3. Best frequency-spectrum parameters of EC in each spectral configuration derived from optimal band combination algorithm

Configuration	LVs	$R_{c}^{2}$	RMSE_CV /%	RPIQ_c	$R_{v}^{2}$	RMSE_P /%	RPIQ_v	Time /s
ASD_1/1	5/12	0.76/0.66	0.349/0.416	2.78/2.33	0.76/0.69	0.350/0.382	2.78/2.55	>24 h/174.806
Con_3/10	5/12	0.77/0.66	0.339/0.415	2.87/2.35	0.76/0.69	0.332/0.381	2.94/2.56	553.184/138.728
Con_10/10	5/14	0.76/0.66	0.348/0.417	2.79/2.33	0.77/0.69	0.359/0.380	2.71/2.56	117.189/109.263
Con_20/20	5/14	0.74/0.67	0.357/0.410	2.72/2.37	0.77/0.68	0.325/0.392	3.00/2.48	15.941/55.348
Con_40/40	5/15	0.70/0.65	0.385/0.427	2.52/2.27	0.72/0.68	0.693/0.390	1.41/2.50	2.973/5.535
Con_60/60	5/12	0.63/0.64	0.431/0.427	2.26/2.28	0.6/0.64	0.464/0.417	2.11/2.34	1.353/2.222
Con_100/100	5/14	0.60/0.59	0.452/0.458	2.14/2.12	0.35/0.54	0.567/0.463	1.72/2.11	0.674/0.736

Table 4. Predicted results of SOM obtained by combining best spectral parameters/all-optical spectral parameters in different spectral configurations

Configuration	LVs	$R_{c}^{2}$	RMSE_CV / $(d S \cdot m^{- 1})$	RPIQ_c	$R_{v}^{2}$	RMSE_P / $(d S \cdot m^{- 1})$	RPIQ_v	Time /s
ASD_1/1	5/4	0.70/0.52	7.12/9.05	2.12/1.67	0.68/0.51	7.06/8.79	2.15/1.73	>24 h/172.844
Con_3/10	5/4	0.71/0.51	7.27/9.42	2.14/1.65	0.70/0.51	7.04/9.17	2.15/1.65	554.677/128.822
Con_10/10	5/4	0.70/0.52	7.17/9.08	2.11/1.67	0.69/0.49	6.88/8.96	2.21/1.70	120.039/105.332
Con_20/20	5/4	0.72/0.52	6.94/9.09	2.18/1.66	0.69/0.49	6.92/8.95	2.19/1.70	15.864/57.023
Con_40/40	5/4	0.67/0.52	7.46/9.05	2.03/1.67	0.65/0.48	7.33/9.10	2.07/1.67	2.431/5.236
Con_60/60	5/4	0.67/0.52	7.74/9.32	2.02/1.68	0.66/0.49	7.87/8.96	1.93/1.70	1.062/2.169
Con_100/100	5/4	0.66/0.50	7.61/9.27	1.99/1.63	0.66/0.49	7.21/8.98	2.11/1.69	0.566/0.739

Table 5. Prediction results of EC obtained by combining best spectral parameters/all-optical spectral parameters in different spectral configurations

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information