Soil organic matter prediction method based on spectral shape characteristics

Abstract

Soil organic matter prediction method based on spectral shape characteristics. The present invention belongs to the field of soil organic matter spectral prediction, and specifically relates to a method for estimating soil organic matter. The method is as follows: 1. Soil sample collection and processing; 2. Soil sample spectral testing; 3. Soil spectral processing and spectral characteristic parameter extraction; 4. SOM prediction model construction to estimate SOM content. The model constructed by the present invention has high prediction accuracy, and the spectral characteristic parameters based on shape characteristics have great potential in SOM prediction.

Description

Technical Field

[0001] The invention belongs to the field of soil organic matter spectrum prediction, and in particular relates to a method for estimating soil organic matter. Background Art

[0002] Soil organic matter (SOM) is an important physical and chemical parameter. SOM is an important indicator for soil safety and fertility evaluation. Rapid and accurate SOM estimation is needed in precision agriculture and sustainable land use. People use spectral technology to understand the differences in the physical and chemical properties of different landforms. The specific spectral differences are mainly reflected in the reflection and absorption characteristics of different spectral wavelengths. Studies have shown that SOM has strong absorption characteristics in the visible near-infrared region, which can be used to quickly and accurately estimate the SOM content. As the SOM content increases, the spectral reflectance decreases. Therefore, SOM has a good negative correlation with the spectral reflectance. In order to improve the correlation between SOM and spectral reflectance, mathematical transformation methods such as differential, logarithmic, and inverse are applied to SOM prediction. Among them, differential transformation is the most commonly used spectral transformation method, which can not only improve the correlation with SOM, but also eliminate the influence of vegetation. In the current study, fractional differentials are applied to SOM prediction, which makes up for the situation that integer differentials cannot fully capture the spectral information in SOM prediction. However, for soil spectral SOM prediction, reflectances or mathematical transformation values ​​of many wavelengths are generally used as prediction variables, which increases the computational intensity.

[0003] Spectral characteristic parameters can characterize the shape characteristics of spectral curves. Generally, spectral characteristic parameters are extracted based on the spectral envelope curve, mainly including absorption area, absorption position, absorption depth and symmetry. Spectral characteristic parameters are mostly used in soil spectral classification research. The spectral characteristics of different soil types in visible light are quite different. Therefore, using spectral characteristic parameters for soil classification can obtain high classification accuracy. There are few studies on the use of spectral characteristic parameters for SOM prediction, but in visible light, as the SOM content increases and the reflectivity decreases, the shape of the spectral curve also changes. In order to study the potential of spectral characteristic parameters in SOM prediction, a SOM prediction method based on the shape characteristics of spectral curves is proposed. It is necessary to develop a fast, accurate and predictive SOM prediction method with few variables, which can greatly improve the prediction efficiency. While using spectral characteristic parameters for prediction, local modeling methods are also applied to improve the prediction ability of the model. Summary of the invention

[0004] The purpose of the present invention is to solve the technical problem that too many predictive variables are used in current SOM prediction, resulting in complex calculations, and to provide a soil organic matter prediction method based on spectral shape characteristics.

[0005] The soil organic matter prediction method based on spectral shape characteristics is as follows:

[0006] 1. Soil sample collection and processing:

[0007] Surface soil samples of cultivated land in Northeast China were collected in Northeast China. The soil samples were air-dried, ground and sieved, and the SOM content was obtained by the potassium dichromate heating method with a conversion factor of 1.724;

[0008] The operating steps of the potassium dichromate heating method are as follows:

[0009] 1.1. Weigh 0.1000g-0.5000g of air-dried soil sample by reduction method, accurate to 0.0001g, place the air-dried soil sample in a hard test tube through a 0.149mm sieve, add 0.1g of silver sulfate, add 5.00mL of 0.8000mol / L potassium dichromate standard solution, and then inject 5mL of sulfuric acid with a syringe, and rotate to mix well;

[0010] 1.2. First, heat the oil bath to 185℃-190℃, insert the hard test tube containing the soil sample into the wire cage in the oil bath to heat, control the temperature in the oil bath to 170℃-180℃, and keep the solution boiling for 5 minutes, then take out the wire cage, wait for the hard test tube to cool down slightly, and wipe the oil on the outside of the test tube with clean paper;

[0011] 1.3. If the solution after boiling is green, it means that the amount of potassium dichromate standard solution is insufficient, and the soil sample should be weighed again; if the solution after boiling is orange-yellow or yellow-green, then after cooling, aspirate the mixture in the test tube into a 250mL conical flask, control the volume in the flask to 60mL-80mL, add 3-4 drops of o-phenanthroline indicator, and titrate with 0.2mol / L ammonium ferrous sulfate standard solution until the solution changes from orange-yellow to blue-green to brown-red; if N-phenylanthranilic acid indicator is used, the color change process changes from brown-red to purple to blue-green as the end point;

[0012] 1.4. When analyzing each batch of soil samples, perform 2-3 blank tests. In the blank tests, no soil samples are added, but 0.1g-0.5g of quartz sand is added. The other operating steps are exactly the same as those for soil sample analysis.

[0013] 2. Spectral test of soil samples:

[0014] ASD spectrometer was used to perform spectral testing in a dark room with a spectral range of 350-2500nm. The soil sample was placed in a container and the surface was scraped flat. A 50W halogen lamp was used as a light source to illuminate the surface of the soil sample. The probe of the ASD spectrometer was used to collect the soil spectrum. Ten spectral curves were collected each time, and the average was calculated as the spectral reflectance curve of a soil sample point.

[0015] 3. Soil spectrum processing:

[0016] The spectral reflectance curve is processed by nine-point smoothing, 10nm resampling and envelope removal. The absorption positions of the two absorption valleys of visible light are determined by removing the envelope curve. Based on the two absorption positions, the spectral characteristic parameters length L and area A are extracted from the spectral reflectance curve.

[0017] The formula for the nine-point smoothing is as follows:

[0018] R′ i =0.04R i-4 +0.08R i-3 +0.12R i-2 +0.16R i-1 +0.2R i +0.16R i+1 +0.12R i+2 +0.08R i+3 +0.04R i+4

[0019] R′ i It represents the reflectivity of a certain wavelength after smoothing, R represents the measured reflectivity, and i represents the wavelength;

[0020] The characteristic parameter length L formula is:

[0021]

[0022]

[0023] n=180°-arctan(α)

[0024] In the length L formula, λ represents the wavelength, a Indicates that the wavelength is a, R represents the reflectivity, β represents the radius, and π = 3.14;

[0025] The characteristic parameter area A formula is:

[0026] A c =A abc -A ▲abc

[0027]

[0028] A ▲abc =[(P2-P1)×(R c -R a )]×0.5

[0029] Where λ represents the wavelength, a Indicates wavelength is a, d is resampling interval, R is reflectivity, P is absorption position, n and α are angles (see the manual). Figure 1), β represents the radius, π = 3.14;

[0030] 4. Soil sample clustering: Use k-mean clustering to cluster the surface soil samples of cultivated land in Northeast China and determine the optimal number of clusters;

[0031] 5. Construction of SOM prediction model:

[0032] The prediction model uses the random forest model, takes the spectral characteristic parameters length L and area A as prediction variables, builds the random forest model in the software R, estimates the SOM content, and obtains the predicted soil organic matter.

[0033] In step 1, soil samples from the surface layer of 0.5-20 cm of cultivated land are collected.

[0034] In step 2, an ASD ground object spectrometer is used for spectral testing with a spectral range of 350-2500nm.

[0035] In step 5, the random forest model is constructed using the RandomForest package. When the number of trees is set to 500 and the optimal split node is set to one-third of the predictor variables, the out-of-bag error of the model is stable.

[0036] Step 5 After the SOM prediction model is constructed, the SOM prediction accuracy is evaluated using the coefficient of determination and root mean square error. The RMSE calculation formula is as follows:

[0037] (yp in this formula i and yo i Yes pi and oi )

[0038] where y p is the predicted SOC value, y o is the measured value of SOC, and m is the number of soil samples.

[0039] The model constructed by the present invention has high prediction accuracy, and the spectral characteristic parameters based on shape characteristics have great potential in SOM prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Schematic diagram of spectral characteristic parameter extraction in Experiment 1 of the present invention, wherein (a) is a length schematic diagram, and (b) is an area schematic diagram;

[0041] Figure 2 This is the training result of Nong'an County in Experiment 1 of the present invention;

[0042] Figure 3 This is the verification result of Nong'an County in Experiment 1 of the present invention;

[0043] Figure 4 It is the training result of Sanjiang Plain in Experiment 1 of the present invention;

[0044] Figure 5 This is the verification result of Sanjiang Plain in Experiment 1 of the present invention. DETAILED DESCRIPTION

[0045] The technical solution of the present invention is not limited to the specific implementation modes listed below, but also includes any combination of the specific implementation modes.

[0046] Specific implementation method 1: The soil organic matter prediction method based on spectral shape characteristics of this implementation method is as follows:

[0047] 1. Soil sample collection and processing:

[0048] Surface soil samples of cultivated land in Northeast China were collected in Northeast China. The soil samples were air-dried, ground and sieved, and the SOM content was obtained by the potassium dichromate heating method with a conversion factor of 1.724;

[0049] The operating steps of the potassium dichromate heating method are as follows:

[0050] 1.1. Weigh 0.1000g-0.5000g of air-dried soil sample by reduction method, accurate to 0.0001g, place the air-dried soil sample in a hard test tube through a 0.149mm sieve, add 0.1g of silver sulfate, add 5.00mL of 0.8000mol / L potassium dichromate standard solution, and then inject 5mL of sulfuric acid with a syringe, and rotate to mix well;

[0051] 1.2. First, heat the oil bath to 185℃-190℃, insert the hard test tube containing the soil sample into the wire cage in the oil bath to heat, control the temperature in the oil bath to 170℃-180℃, and keep the solution boiling for 5 minutes, then take out the wire cage, wait for the hard test tube to cool down slightly, and wipe the oil on the outside of the test tube with clean paper;

[0052] 1.3. If the solution after boiling is green, it means that the amount of potassium dichromate standard solution is insufficient, and the soil sample should be weighed again; if the solution after boiling is orange-yellow or yellow-green, then after cooling, aspirate the mixture in the test tube into a 250mL conical flask, control the volume in the flask to 60mL-80mL, add 3-4 drops of o-phenanthroline indicator, and titrate with 0.2mol / L ammonium ferrous sulfate standard solution until the solution changes from orange-yellow to blue-green to brown-red; if N-phenylanthranilic acid indicator is used, the color change process changes from brown-red to purple to blue-green as the end point;

[0053] 1.4. When analyzing each batch of soil samples, perform 2-3 blank tests. In the blank tests, no soil samples are added, but 0.1g-0.5g of quartz sand is added. The other operating steps are exactly the same as those for soil sample analysis.

[0054] 2. Spectral test of soil samples:

[0055] ASD spectrometer was used to perform spectral testing in a dark room with a spectral range of 350-2500nm. The soil sample was placed in a container and the surface was scraped flat. A 50W halogen lamp was used as a light source to illuminate the surface of the soil sample. The probe of the ASD spectrometer was used to collect the soil spectrum. Ten spectral curves were collected each time, and the average was calculated as the spectral reflectance curve of a soil sample point.

[0056] 3. Soil spectrum processing:

[0057] The spectral reflectance curve is processed by nine-point smoothing, 10nm resampling and envelope removal. The absorption positions of the two absorption valleys of visible light are determined by removing the envelope curve. Based on the two absorption positions, the spectral characteristic parameters length L and area A are extracted from the spectral reflectance curve.

[0058] The formula for the nine-point smoothing is as follows:

[0059] R′ i =0.04R i-4 +0.08R i-3 +0.12R i-2 +0.16R i-1 +0.2R i +0.16R i+1 +0.12R i+2 +0.08R i+3 +0.04R i+4

[0060] R′ i It represents the reflectivity of a certain wavelength after smoothing, R represents the measured reflectivity, and i represents the wavelength;

[0061] The characteristic parameter length L formula is:

[0062]

[0063] n=180°-arctan(θ)

[0064] In the length L formula, λ represents the wavelength, a Indicates that the wavelength is a, R represents the reflectivity, β represents the radius, and π = 3.14;

[0065] The characteristic parameter area A formula is:

[0066] A c =A abc -A ▲abc

[0067]

[0068] A ▲abc =[(P2-P1)×(R c -R a )]×0.5

[0069] Where λ represents the wavelength, a Indicates wavelength is a, d is resampling interval, R is reflectivity, P is absorption position, n and α are angles (see the manual). Figure 1 ), β represents the radius, π = 3.14;

[0070] 4. Soil sample clustering: Use k-mean clustering to cluster the surface soil samples of cultivated land in Northeast China and determine the optimal number of clusters;

[0071] 5. Construction of SOM prediction model:

[0072] The prediction model uses the random forest model, takes the spectral characteristic parameters length L and area A as prediction variables, builds the random forest model in the software R, estimates the SOM content, and obtains the predicted soil organic matter.

[0073] Specific implementation method 2: This implementation method is different from the specific implementation method 1 in that a soil sample of 0.5-20 cm of the surface layer of cultivated land is collected in step 1. The rest is the same as the specific implementation method 1.

[0074] Specific implementation method three: This implementation method is different from specific implementation method one or two in that in step two, an ASD ground object spectrometer is used to perform spectrum testing, and the spectrum range is 350-2500nm. The rest is the same as specific implementation method one or two.

[0075] Specific implementation method 4: This implementation method is different from specific implementation methods 1 to 3 in that the random forest model is constructed using the RandomForest package in step 5, and when the number of trees is set to 500 and the optimal split node is set to one third of the predicted variable, the out-of-bag error of the model is stable. The rest is the same as specific implementation methods 1 to 3.

[0076] Specific implementation method 5: The difference between this implementation method and specific implementation methods 1 to 4 is that after the SOM prediction model is constructed in step 5, the SOM prediction accuracy is evaluated using the determination coefficient and root mean square error. The RMSE calculation formula is as follows:

[0077]

[0078] where y p is the predicted SOC value, y ois the measured value of SOC, i represents the quantity, and m is the number of soil samples. The rest is the same as that of the first to fourth embodiments.

[0079] The following experiments were used to verify the effect of the present invention:

[0080] Experiment 1:

[0081] The soil organic matter prediction method based on spectral shape characteristics is characterized in that the soil organic matter prediction method based on spectral shape characteristics is as follows:

[0082] 1. Soil sample collection and processing:

[0083] Surface soil samples of cultivated land in Northeast China were collected in Northeast China (surface soil samples of cultivated land were collected in Sanjiang Plain of Heilongjiang Province and Nong'an County of Changchun City of Jilin Province), and the soil samples were air-dried, ground and sieved. The potassium dichromate heating method was used to obtain the SOM content through a conversion coefficient of 1.724;

[0084] 2. Spectral test of soil samples:

[0085] ASD spectrometer was used to perform spectral testing in a dark room with a spectral range of 350-2500nm. The soil sample was placed in a container and the surface was scraped flat. A 50W halogen lamp was used as a light source to illuminate the surface of the soil sample. The probe of the ASD spectrometer was used to collect the soil spectrum. Ten spectral curves were collected each time, and the average was calculated as the spectral reflectance curve of a soil sample point.

[0086] 3. Soil spectrum processing:

[0087] The spectral reflectance curve is processed by nine-point smoothing, 10nm resampling and envelope removal. The absorption positions of the two absorption valleys of visible light are determined by removing the envelope curve. Based on the two absorption positions, the spectral characteristic parameters length L and area A are extracted from the spectral reflectance curve.

[0088] The formula for the nine-point smoothing is as follows:

[0089] R′ i =0.04R i-4 +0.08R i-3 +0.12R i-2 +0.16R i-1 +0.2R i +0.16R i+1 +0.12R i+2 +0.08R i+3 +0.04R i+4

[0090] R′ iIt represents the reflectivity of a certain wavelength after smoothing, R represents the measured reflectivity, and i represents the wavelength;

[0091] The characteristic parameter length L formula is:

[0092]

[0093] n=180°-arctan(θ)

[0094] In the length L formula, λ represents the wavelength, a Indicates that the wavelength is a, R represents the reflectivity, β represents the radius, and π = 3.14;

[0095] The characteristic parameter area A formula is:

[0096] A c =A abc -A ▲abc

[0097]

[0098] A ▲abc =[(P2-P1)×(R c -R a )]×0.5

[0099] Where λ represents the wavelength, a Indicates wavelength is a, d is resampling interval, R is reflectivity, P is absorption position, n and α are angles (see the manual). Figure 1 ), β represents the radius, π = 3.14;

[0100] 4. Soil sample clustering: Use k-mean clustering to cluster the surface soil samples of cultivated land in Northeast China (soil samples from Sanjiang Plain and Nong'an County) and determine the optimal number of clusters;

[0101] 5. Construction of SOM prediction model:

[0102] The prediction model uses the random forest model, takes the spectral characteristic parameters length L and area A as prediction variables, builds the random forest model in the software R, estimates the SOM content, and obtains the predicted soil organic matter.

[0103] In step 1, soil samples from the surface layer of 0.5-20 cm of cultivated land are collected.

[0104] In step 2, an ASD ground object spectrometer is used for spectral testing with a spectral range of 350-2500nm.

[0105] In step 5, the random forest model is constructed using the RandomForest package. When the number of trees is set to 500 and the optimal split node is set to one-third of the predictor variables, the out-of-bag error of the model is stable.

[0106] Step 5: After the SOM prediction model is constructed, the SOM prediction accuracy is calculated using the coefficient of determination (R 2 ) and root mean square error (RMSE) evaluation, the RMSE calculation formula is as follows:

[0107]

[0108] where y p is the predicted SOC value, y o is the measured value of SOC, i represents the quantity, and m is the number of soil samples.

[0109] pass Figure 1-Figure 5 It can be seen that for the soil data of the Sanjiang Plain in Heilongjiang, the verification accuracy of the model is R 2 =0.69; RMSE =3.76 g kg -1 ; For Nong'an County, Jilin Province, the validation accuracy of the model is R 2 =0.76; RMSE = 7.43 gkg -1 The model has high prediction accuracy, and the spectral feature parameters based on shape features have great potential in SOM prediction.

Claims

1. Soil organic matter prediction method based on spectral shape characteristics, characterized by The soil organic matter prediction method based on spectral shape characteristics is as follows:

1. Soil sample collection and processing: Surface soil samples from cultivated land in Northeast China were collected in Northeast China. The soil samples were air-dried, ground and sieved. The potassium dichromate heating method was used to obtain the soil organic matter (SOM) content through a conversion coefficient of 1.

724.

2. Spectral test of soil samples: ASD spectrometer was used to perform spectral testing in a dark room with a spectral range of 350-2500nm. The soil sample was placed in a container and the surface was scraped flat. A 50W halogen lamp was used as a light source to illuminate the surface of the soil sample. The probe of the ASD spectrometer was used to collect the soil spectrum. Ten spectral curves were collected each time, and the average was calculated as the spectral reflectance curve of a soil sample point.

3. Soil spectral processing: The spectral reflectance curve is processed by nine-point smoothing, 10nm resampling and envelope removal. The absorption positions of the two absorption valleys of visible light are determined by removing the envelope curve. Based on the two absorption positions, the spectral characteristic parameters length L and area A are extracted from the spectral reflectance curve. The formula for the nine-point smoothing is as follows: R′ i =0.04R i-4 +0.08R i-3 +0.12R i-2 +0.16R i-1 +0.2R i +0.16R i+1 +0.12R i+2 +0.08R i+3 +0.04R i+4 R′ i It represents the reflectivity of a certain wavelength after smoothing, R represents the measured reflectivity, and i represents the wavelength; The characteristic parameter length L formula is: n=180°-arctan(α) In the length L formula, λ represents the wavelength, a and λ c denote the wavelengths at points a and c, respectively, R denotes the reflectivity, β denotes the radius, and π = 3.14; The characteristic parameter area A formula is: A c =A abc -A ▲abc A ▲abc =[(P2-P1)×(R c -R a )]×0.5 Where λ represents the wavelength, a and λ c denote the wavelengths of points a and c, respectively, d denotes the resampling interval, R denotes the reflectivity, P denotes the absorption position, n and α denote the angles, β denotes the radius, and π = 3.14; 4. Soil sample clustering: Use k-mean clustering to cluster the surface soil samples of cultivated land in Northeast China and determine the optimal number of clusters; 5. Construction of SOM prediction model: The prediction model uses the random forest model, takes the spectral characteristic parameters length L and area A as prediction variables, builds the random forest model in the software R, estimates the SOM content, and obtains the predicted soil organic matter; In step 5, the random forest model was constructed using the RandomForest package. The number of trees was set to 500, the optimal split node was set to one-third of the predictor variables, and the out-of-bag error of the model was stable.

2. The soil organic matter prediction method based on spectral shape characteristics according to claim 1 is characterized in that In step 1, soil samples from the surface layer of 0.5-20 cm of cultivated land are collected.

3. The soil organic matter prediction method based on spectral shape characteristics according to claim 1 is characterized in that Step 5 After the SOM prediction model is constructed, the SOM prediction accuracy is evaluated using the coefficient of determination and root mean square error. The RMSE calculation formula is as follows: where y p is the predicted SOC value, y o is the measured value of SOC, i represents the quantity, and m is the number of soil samples.

Images