Rapid determination methods for field soil Droperties usine visible-near infrared spectroscopy
author: Gavin
2022-01-07
With the development of economy and technology, traditional extensive agriculture can no longer meet the comprehensive needs of modern people for the environment, resources, health, and economic benefits. Under such circumstances, the precision agriculture model has become an important direction in the development of modern agriculture. In precision agriculture, it is necessary to adjust the input of various agricultural materials to obtain the maximum economic and ecological benefits according to the different soil and crop conditions of each operation unit in the field. As the basis of agricultural production, soil provides the necessary nutrients and environmental conditions for crop growth, and rapid quantitative detection of its fertility is a prerequisite for the realization of precision agriculture. Traditional chemical method testing requires soil samples to be taken back to the laboratory for analysis. Although the testing accuracy is high, it requires a lot of labor cost and testing time, and the timeliness is low, which cannot meet the needs of quickly obtaining test results. In recent years, the rapid development of soil visible-near infrared (Vis-NIR) spectroscopy technology has overcome the disadvantages of high cost and long cycle of traditional methods, and has been widely used in agricultural production.
Principle
Based on visible-near-infrared full spectrum technology, combined with chemical test data, perform spectral modeling research on soil organic matter, total nitrogen and pH data of agricultural land, explore the applicability of different pretreatment methods and models in the prediction of soil properties in this area, and try Remove the influence of factors such as moisture in the field in-situ spectrum through spectral conversion.Use Optosky ATP9110-25 spectrometer to collect indoor spectral data for preprocessing method and model research. Three preprocessing methods and two models are selected for prediction accuracy comparison: the three preprocessing methods are SG smoothing, continuum removal, and SNV transformation. The latter two methods are all based on SG smoothing; two models The linear regression algorithm PLSR and the machine learning algorithm Cubisto were chosen to explore whether the DS spectral conversion method can play a role in removing the influence of moisture in the field spectrum and improve the prediction accuracy.
Solution
After the soil samples are transported to the laboratory, they need to be air-dried, ground and sieved (2mm aperture). After that, each sample is divided into two by the quarter method, and one is used for chemical analysis to determine the soil physical and chemical value. One copy is used for indoor spectrum test
Principle
Based on visible-near-infrared full spectrum technology, combined with chemical test data, perform spectral modeling research on soil organic matter, total nitrogen and pH data of agricultural land, explore the applicability of different pretreatment methods and models in the prediction of soil properties in this area, and try Remove the influence of factors such as moisture in the field in-situ spectrum through spectral conversion.Use Optosky ATP9110-25 spectrometer to collect indoor spectral data for preprocessing method and model research. Three preprocessing methods and two models are selected for prediction accuracy comparison: the three preprocessing methods are SG smoothing, continuum removal, and SNV transformation. The latter two methods are all based on SG smoothing; two models The linear regression algorithm PLSR and the machine learning algorithm Cubisto were chosen to explore whether the DS spectral conversion method can play a role in removing the influence of moisture in the field spectrum and improve the prediction accuracy.
Solution
After the soil samples are transported to the laboratory, they need to be air-dried, ground and sieved (2mm aperture). After that, each sample is divided into two by the quarter method, and one is used for chemical analysis to determine the soil physical and chemical value. One copy is used for indoor spectrum test
The test items of soil physical and chemical value are SOM, TN and pH value, and the test methods adopt conventional chemical analysis methods: the determination of soil organic matter content adopts the external heating-potassium bicarbonate volumetric method; the determination of soil total nitrogen content adopts the Kjeldahl method; The soil pH value is determined by the 5.0:1.0 water-soil ratio electrode measurement method, and the statistical data of the measurement results are shown in Table 2-1.
ATP9110 series broadband field spectroradiometer is a new portable hyperspectral fieldSpec from Optosky. Wavelength range of 300-11000 nm, suitable for geological research, mineral exploration, remote sensing, crop monitoring, forest research, oceanography and other fields of application.
When collecting the field spectrum, it is necessary to collect the soil sample with a square sampler with a depth of 20cm and a sampling area of 10cm×10cm. After sampling, the side of the sampler is opened, and the spectrum is collected on the vertical section of the soil. A reflection probe is used to collect 10 spectra at each of 3 different depth positions at the upper, middle, and lower ends of the profile. After the collected 30 spectral data are arithmetic averaged, the field in-situ spectral data of the sampling point can be obtained. In the process of field spectrum detection, additional attention should be paid to ensure that the measured soil surface is flat but not too smooth, and the probe should try to avoid crushed rocks, roots and large pores in the soil that will affect the test results.
When collecting the field spectrum, it is necessary to collect the soil sample with a square sampler with a depth of 20cm and a sampling area of 10cm×10cm. After sampling, the side of the sampler is opened, and the spectrum is collected on the vertical section of the soil. A reflection probe is used to collect 10 spectra at each of 3 different depth positions at the upper, middle, and lower ends of the profile. After the collected 30 spectral data are arithmetic averaged, the field in-situ spectral data of the sampling point can be obtained. In the process of field spectrum detection, additional attention should be paid to ensure that the measured soil surface is flat but not too smooth, and the probe should try to avoid crushed rocks, roots and large pores in the soil that will affect the test results.
Test result
The content of various substances in the soil, such as moisture, organic matter, nitrogen, phosphorus, potassium, iron oxide, etc., will affect its spectral reflectance characteristics. Different types of soils have different reflectance characteristics. Through proper pretreatment, some of the noise effects generated during the spectral collection process can be removed and the spectral reflectance characteristics can be further highlighted. After collecting 240 soil samples in the study area, remove the first and last ends of the spectrum (high high-frequency noise), leaving the wavelength range of 400~2450nm to obtain the average spectral curve, and analyze the basic spectral reflectance characteristics of the fluvo-aquic soil in this area. The original average spectrum is shown in Figure 3-1 (a). Generally speaking, in the wavelength range of 400^1200nm, the reflectance increases rapidly with wavelength, and the growth tends to be gentle at 1200~2450nm, and at 1400nm. 1900nm. 2200nm three There are obvious absorption valleys near these wavelengths.The main reason for the absorption troughs at 1400nm and 1900nm is the absorption of O-H groups in water molecules. The absorption trough at 1900 nm is significantly higher than that at 1400 nm, and the 2200nm absorption trough is mainly caused by O-H groups in organic matter. SG smoothing, SNV transformation, and continuum removal are used to preprocess the original average spectrum. SG smoothing uses a second-order polynomial and the window size is set to 11. SNV transformation and continuum removal are the basis of SG smoothing. Finished above, the processing results are shown in order of 3-1 (b), 3-1 (c), 3-1 (d). From the figure, it can be seen that SG smoothing does not change the overall trend of the spectrum much, but Reduce the "burr" noise, make the spectral curve smoother. The SNV transformation makes the spectral absorption characteristics significantly magnified. After the continuum removal process, in addition to the three obvious absorption valleys being highlighted, another absorption valley is formed near 500 nm, which is difficult to observe with the naked eye in the original spectrum. Information related to the absorption of iron oxides in the soil
The content of various substances in the soil, such as moisture, organic matter, nitrogen, phosphorus, potassium, iron oxide, etc., will affect its spectral reflectance characteristics. Different types of soils have different reflectance characteristics. Through proper pretreatment, some of the noise effects generated during the spectral collection process can be removed and the spectral reflectance characteristics can be further highlighted. After collecting 240 soil samples in the study area, remove the first and last ends of the spectrum (high high-frequency noise), leaving the wavelength range of 400~2450nm to obtain the average spectral curve, and analyze the basic spectral reflectance characteristics of the fluvo-aquic soil in this area. The original average spectrum is shown in Figure 3-1 (a). Generally speaking, in the wavelength range of 400^1200nm, the reflectance increases rapidly with wavelength, and the growth tends to be gentle at 1200~2450nm, and at 1400nm. 1900nm. 2200nm three There are obvious absorption valleys near these wavelengths.The main reason for the absorption troughs at 1400nm and 1900nm is the absorption of O-H groups in water molecules. The absorption trough at 1900 nm is significantly higher than that at 1400 nm, and the 2200nm absorption trough is mainly caused by O-H groups in organic matter. SG smoothing, SNV transformation, and continuum removal are used to preprocess the original average spectrum. SG smoothing uses a second-order polynomial and the window size is set to 11. SNV transformation and continuum removal are the basis of SG smoothing. Finished above, the processing results are shown in order of 3-1 (b), 3-1 (c), 3-1 (d). From the figure, it can be seen that SG smoothing does not change the overall trend of the spectrum much, but Reduce the "burr" noise, make the spectral curve smoother. The SNV transformation makes the spectral absorption characteristics significantly magnified. After the continuum removal process, in addition to the three obvious absorption valleys being highlighted, another absorption valley is formed near 500 nm, which is difficult to observe with the naked eye in the original spectrum. Information related to the absorption of iron oxides in the soil
Conclusion
In this study, the influence of environmental factors such as moisture on the field in-situ spectra was removed. The DS spectrum conversion method was selected to establish a conversion model between the field spectra and the corresponding indoor spectra to complete the conversion of the field in-situ spectra to the indoor spectra, and finally effectively improve the field The prediction accuracy of in-situ spectroscopy for the three attributes. It is preliminarily proved that the DS spectrum conversion method is feasible in field in-situ spectroscopy to remove the influence of interference factors such as moisture.
Related products
Based on visible-near-infrared full spectrum technology, combined with chemical test data, perform spectral modeling research on soil organic matter, total nitrogen and pH data of agricultural land.
SWIR Hypespectral Camera:ATH1010-25
Fieldspec Spectroradiometer:ATP9110-25
Related articles
1.Wang Y Veltkamp D J, Kowalski B R. Multivariate instrument standardization[J].Analytical Chemistry, 1991, 63(23):2750-2756.
2.Wang Y, Huang T, Liu J, et al. Soil pH value, organic matter and macronutrients contents prediction using optical diffuse reflectance spectroscopy [J]. Computers & Electronics in Agriculture, 2015, 111:69-77.
3.Vohland M, Besold J, Hill J, et al. Comparing different multivariate calibration methods fbr the determination of soil organic carbon pools with visible to near infrared spectroscopy [J]. Geoderma, 2011, 166(1):0-205.
4.Ackerson J P, Morgan C L S, Ge Y Penetrometer-mounted VisNIR spectroscopy: Application of EPO-PLS to in situ VisNIR spectra[J]. Geoderma,2017, 286:131- 138.
In this study, the influence of environmental factors such as moisture on the field in-situ spectra was removed. The DS spectrum conversion method was selected to establish a conversion model between the field spectra and the corresponding indoor spectra to complete the conversion of the field in-situ spectra to the indoor spectra, and finally effectively improve the field The prediction accuracy of in-situ spectroscopy for the three attributes. It is preliminarily proved that the DS spectrum conversion method is feasible in field in-situ spectroscopy to remove the influence of interference factors such as moisture.
Related products
Based on visible-near-infrared full spectrum technology, combined with chemical test data, perform spectral modeling research on soil organic matter, total nitrogen and pH data of agricultural land.
SWIR Hypespectral Camera:ATH1010-25
Fieldspec Spectroradiometer:ATP9110-25
Related articles
1.Wang Y Veltkamp D J, Kowalski B R. Multivariate instrument standardization[J].Analytical Chemistry, 1991, 63(23):2750-2756.
2.Wang Y, Huang T, Liu J, et al. Soil pH value, organic matter and macronutrients contents prediction using optical diffuse reflectance spectroscopy [J]. Computers & Electronics in Agriculture, 2015, 111:69-77.
3.Vohland M, Besold J, Hill J, et al. Comparing different multivariate calibration methods fbr the determination of soil organic carbon pools with visible to near infrared spectroscopy [J]. Geoderma, 2011, 166(1):0-205.
4.Ackerson J P, Morgan C L S, Ge Y Penetrometer-mounted VisNIR spectroscopy: Application of EPO-PLS to in situ VisNIR spectra[J]. Geoderma,2017, 286:131- 138.
Optosky Fieldspec NIR Analyzers 
Application of Field Spectrometer in Estimating Leaf Chlorophyll Content 
Comparison test of manual measure surface reflectance 
Military rescue technology of UAV equipped with FieldSpec 
Application of Fieldspec in Soil Salinization 
How to quickly and nondestructively identify jade by ATP9100? 
What is the advantage of design of Field Spectradiometer? 
How to use FieldSpec to detect field spectral?