Estimation of Soil Organic Matter Content Based on Characteristic Variable Selection and Regression Methods

Guanwen Li; Xiaohong Gao; Nengwen Xiao; Yunfei Xiao

doi:10.3788/AOS201939.0930002

Acta Optica Sinica, Volume. 39, Issue 9, 0930002(2019)

Estimation of Soil Organic Matter Content Based on Characteristic Variable Selection and Regression Methods

Guanwen Li^1,2、**, Xiaohong Gao^、*, Nengwen Xiao², and Yunfei Xiao¹

¹ Qinghai Provincial Key Laboratory of Physical Geography and Environmental Process, School of Geography Sciences, Qinghai Normal University, Xining, Qinghai 810008, China

² Chinese Research Academy of Environmental Sciences, Beijing 100012, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(16)

Fig. 1. Location of the study area and distribution of soil sampling sites

Download full size

Fig. 2. Spectral reflectance curves of soil samples. (a) Raw spectra; (b) spectra after MSC-MF-1st Derivative pre-processing

Download full size

Fig. 3. Variable selection process by sCARS. (a) Changing trend of variables; (b) 10-fold RMSECV values; (c) regression coefficients of variables

Download full size

Fig. 4. Characteristic variable selection process by GA

Download full size

Fig. 5. Characteristic variable selection process by sCARS-SPA from the pre-processing spectrum. (a) Number of variables in the model; (b) variable index

Download full size

Fig. 6. Distribution of characteristic variables with different variable selection methods

Download full size

Fig. 7. Scatter plot for the measured and predicted value by sCARS-PLSR model

Download full size

Fig. 8. Scatter plot for the measured and predicted value by SPA-SVM model

Download full size

Fig. 9. Scatter plot for the measured and predicted value by IRIV-RF model

Download full size

Fig. 10. Results of PLSR, SVM and RF models with different variable selection methods

Download full size

Fig. 11. Scatter plots for the measured and predicted value by sCARS-RF model before and after artificially eliminating outliers. (a) Contain outliers; (b) eliminate outliers

Download full size

Table 1. Soil organic matter content statistics of calibration set and validation set
View table
Table 1. Soil organic matter content statistics of calibration set and validation set
Sample set Samplenumber Min /(g·kg^-1) Max /(g·kg^-1) Mean /(g·kg^-1) SD
Calibration set 268 4.86 148.74 32.47 23.52
Validation set 133 8.26 133.56 32.16 22.44

Table 2. Accuracies of PLSR model with different variable selection methods

View table

Table 2. Accuracies of PLSR model with different variable selection methods

Selection method	Variable number	PC	Calibration set		Validation set
Selection method	Variable number	PC	$R_{cal}^{2}$	R_MSECAL	${R^{2}}_{val}$	R_MSEVAL	R_PD
Full-spectrum	2000	5	0.842	9.326	0.835	9.069	2.5
sCARS	51	5	0.874	8.327	0.883	7.797	2.9
SPA	5	5	0.850	9.103	0.858	8.525	2.6
GA	186	4	0.842	9.342	0.861	8.415	2.7
IRIV	63	6	0.843	9.300	0.875	8.043	2.8
sCARS-SPA	17	4	0.765	11.391	0.848	8.791	2.6

Table 3. Accuracies of SVM model with different variable selection methods

View table

Table 3. Accuracies of SVM model with different variable selection methods

Selection method	Variable number	Optimal parameter		Calibration set		Validation set
Selection method	Variable number	g (nuclear function)	c (punishment coefficient)	$R_{cal}^{2}$	R_MSECAL	$R_{val}^{2}$	R_MSEVAL	R_PD
Full-spectrum	2000	0.036	3.031	0.91	7.221	0.74	11.546	1.9
sCARS	51	0.021	3.031	0.881	8.116	0.877	7.918	2.8
SPA	5	0.004	1.741	0.858	8.855	0.889	7.477	3.0
GA	186	0.012	1.741	0.867	8.577	0.871	8.093	2.8
IRIV	63	0.021	3.031	0.869	8.493	0.864	8.307	2.7
sCARS-SPA	17	0.011	2.858	0.877	8.246	0.873	8.052	2.8

Table 4. Accuracies of RF model with different variable selection methods

View table

Table 4. Accuracies of RF model with different variable selection methods

Selection method	Variable number	Calibration set		Validation set
Selection method	Variable number	$R_{cal}^{2}$	R_MSECAL	$R_{val}^{2}$	R_MSEVAL	R_PD
Full-spectrum	2000	0.942	5.817	0.957	4.840	4.6
sCARS	51	0.942	5.781	0.958	4.780	4.7
SPA	5	0.930	6.585	0.954	5.082	4.4
GA	186	0.939	5.894	0.959	4.699	4.8
IRIV	63	0.941	5.927	0.960	4.656	4.8
sCARS-SPA	17	0.940	5.910	0.955	4.971	4.5

Table 5. Model accuracy after manually eliminating outliers
View table
Table 5. Model accuracy after manually eliminating outliers
Model Calibration set Validation set
$R_{cal}^{2}$ R_MSECAL $R_{val}^{2}$ R_MSEVAL R_PD
sCARS-PLSR 0.943 5.538 0.926 5.987 3.7
sCARS-SVM 0.926 6.092 0.957 4.903 4.6
sCARS-RF 0.985 3.204 0.988 2.865 7.8

Tools

Get Citation

Copy Citation Text

Guanwen Li, Xiaohong Gao, Nengwen Xiao, Yunfei Xiao. Estimation of Soil Organic Matter Content Based on Characteristic Variable Selection and Regression Methods[J]. Acta Optica Sinica, 2019, 39(9): 0930002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites