Method and system for estimating soil organic matter using modified three-dimensional spectral index

Abstract

The present invention proposes a method and system for estimating soil organic matter using a modified three-dimensional spectral index. First, the measured soil spectral data is obtained and the spectral data is preprocessed; then, the two-dimensional index and the modified three-dimensional index are respectively constructed using the Spearman correlation coefficient and the band combination optimization method, and the soil organic matter estimation model is constructed using the support vector machine modeling method; finally, the optimal spectral preprocessing method and optimal parameters are determined based on the correlation between the modified three-dimensional index constructed by each preprocessed spectrum under different correction factor values ​​and the soil organic matter and the change in model accuracy. The three indicators of absolute coefficient, root mean square error and residual prediction deviation are used to evaluate the model estimation performance, and the optimal soil organic matter estimation model (R 2 The soil organic matter content is estimated when the value is the largest, the RMSE value is the smallest, and the RPD value is greater than 2 and is the largest. The present invention can effectively weaken soil noise information and improve the estimation accuracy of soil organic matter.

Description

Technical Field

[0001] The present invention relates to the field of soil organic matter content estimation, and in particular to a method and system for estimating soil organic matter content by using a hyperspectral modified three-dimensional index (MNDI). Background Art

[0002] Soil organic matter content is a key indicator of soil fertility, and its rapid and accurate monitoring is of great significance to the rational use of land resources. Hyperspectral remote sensing technology is widely used in soil organic matter content estimation and environmental monitoring due to its advantages such as non-destructiveness, rapidity, and cost-effectiveness. However, the original soil reflectance spectrum usually contains information about other factors (such as water, salt, and vegetation, etc.) and is affected by the internal or environmental noise of the sensor, thereby masking part of the spectral information of soil organic matter, making it difficult to quantitatively estimate it. Therefore, it is necessary to develop a method and system for efficiently estimating soil organic matter content using hyperspectral data.

[0003] At present, many methods have emerged to estimate soil organic matter content using hyperspectral data. For example, in "Hong YS, Liu YL, Chen YY, et al. Application of fractional-order derivative in the quantitative estimation of soil organic matter content through visible and near-infrared spectroscopy [J]. Geoderma, 2018, 337: Pages 758-769.", different decomposition order differential algorithms were used to establish a model to estimate soil organic matter content; in "Tian YC, Zhang J, Yao X, et al. Laboratory assessment of three quantitative methods for estimating the organic matter content of soils in China based on visible / near-infrared reflectance spectra [J]. Geoderma, 2013, 202 (Complete): 161-170.", different pre-processed spectra were used to construct a two-dimensional index and model the estimation of soil organic matter content. Compared with one-dimensional spectral data, the two-dimensional index can enhance the relationship between soil organic matter and spectral characteristics to a certain extent. However, the optimization of this index is limited to a two-dimensional form, and few researchers have considered extending the bands to more than two. Summary of the invention

[0004] The purpose of the present invention is to weaken soil noise information, improve the estimation accuracy of soil organic matter, and provide a method and system for estimating soil organic matter content using a hyperspectral modified three-dimensional index (MNDI).

[0005] To achieve the above object, the present invention adopts the following technical solution:

[0006] The method for estimating soil organic matter content using hyperspectral modified three-dimensional index (MNDI) includes the following steps:

[0007] Step A: obtaining measured soil spectral data and preprocessing the spectral data;

[0008] Step B: The two-dimensional index (NDI) and the modified three-dimensional index (MNDI) were constructed using the Spearman correlation coefficient (r) and the band combination optimization method, and the soil organic matter estimation model was constructed using the support vector machine (SVM) modeling method;

[0009] Step C: According to the correlation between the modified three-dimensional index (MNDI) and soil organic matter formed by each preprocessed spectrum under different correction factors (k) and the change in model accuracy, the optimal spectrum preprocessing method and optimal parameters are determined. 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model was selected (the specific conditions were: R 2 The soil organic matter content was estimated by using the largest value, the smallest RMSE value, and the largest RPD value (greater than 2).

[0010] Preferably, in step A, measured soil spectral data are obtained and the spectral data are preprocessed: the obtained measured soil spectral data are preprocessed by moving window smoothing (SMO), Savitzky-Golay first-order derivative (SGD1), Savitzky-Golay second-order derivative (SGD2), standard normal variable transformation (SNV), standard normal variable transformation-detrending (SNV-DT), multivariate scatter correction (MSC) and maximum normalization (NOR).

[0011] Preferably, in step B, the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) are respectively constructed using the Spearman correlation coefficient (r) and the band combination optimization method, and the support vector machine (SVM) modeling method is used to construct a soil organic matter estimation model: the Spearman correlation coefficient (r) between the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) composed of each band combination and the soil organic matter is calculated, and the support vector machine (SVM) modeling method is used to construct a soil organic matter estimation model.

[0012] Preferably, in step C, the optimal spectral preprocessing method and the optimal parameters are determined according to the correlation between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum at different correction factors (k) and soil organic matter and the change in model accuracy: the optimal spectral preprocessing method and the optimal parameter value of the input correction factor (k) are determined according to the change in the Spearman correlation coefficient (r) between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum at different correction factors (k) and soil organic matter and the estimation model accuracy.

[0013] Preferably, the absolute coefficient (R 2 The model estimation performance was evaluated by three indicators: absolute coefficient (R 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) are used to evaluate the model estimation performance. 2 The larger the value and the smaller the RMSE value, the higher the accuracy of the estimation model. The PRD value is interpreted using a three-level classification standard. If the RPD is greater than 2, it means that the estimation model has ideal accuracy; if the RPD value is between 1.8 and 2, it means that the estimation model has the ability to quantitatively evaluate soil organic matter; if the RPD value is less than 1.4, the estimation model is considered unusable. The best model (with the largest R2 value, the smallest RMSE value, and the RPD value greater than 2 and the largest) is selected for soil organic matter content estimation.

[0014] In a second aspect, the present invention also proposes a system for estimating soil organic matter content using a hyperspectral modified three-dimensional index (MNDI), which is characterized by comprising:

[0015] Spectral preprocessing module: configured to obtain measured soil spectral data and preprocess the spectral data;

[0016] Index construction and modeling module: It is configured to construct the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) respectively using the Spearman correlation coefficient (r) and the band combination optimization method, and to construct the soil organic matter estimation model using the support vector machine (SVM) modeling method;

[0017] Spectral preprocessing method and parameter optimization module: It is configured to determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index (MNDI) and soil organic matter formed by each preprocessed spectrum under different correction factor (k) values ​​and the change of model accuracy; the absolute coefficient (R 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model was selected (the specific conditions were: R2 The soil organic matter content was estimated by using the largest value, the smallest RMSE value, and the largest RPD value (greater than 2).

[0018] Preferably, in the spectral preprocessing module, the acquired measured soil spectral data are preprocessed by moving window smoothing (SMO), Savitzky-Golay first-order derivative (SGD1), Savitzky-Golay second-order derivative (SGD2), standard normal variable transformation (SNV), standard normal variable transformation-detrending (SNV-DT), multivariate scatter correction (MSC) and maximum normalization (NOR);

[0019] Preferably, in the index construction and modeling module, the Spearman correlation coefficient (r) between the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) composed of each band combination and the soil organic matter is calculated, and the support vector machine (SVM) modeling method is used to construct a soil organic matter estimation model;

[0020] Preferably, in the spectrum preprocessing method and parameter optimization module, the optimal spectrum preprocessing method and the optimal parameter value of the input correction factor (k) are determined according to the Spearman correlation coefficient (r) between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum at different correction factors (k) and the soil organic matter and the change of the estimation model accuracy. 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model (R 2 The soil organic matter content was estimated by using the largest value, the smallest RMSE value, and the largest RPD value (greater than 2).

[0021] The beneficial effects of the present invention are:

[0022] First, the present invention obtains measured soil spectral data and preprocesses the spectral data; then, the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) are constructed respectively using the Spearman correlation coefficient (r) and the band combination optimization method, and the soil organic matter estimation model is constructed using the support vector machine (SVM) modeling method; finally, the optimal spectral preprocessing method and optimal parameters are determined based on the correlation between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum under different correction factors (k) and the soil organic matter and the change in model accuracy. The absolute coefficient (R 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model (R 2The soil organic matter content is estimated when the value is the largest, the RMSE value is the smallest, and the RPD value is greater than 2 and is the largest. The present invention can effectively weaken soil noise information and improve the estimation accuracy of soil organic matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the method and system for estimating soil organic matter content using hyperspectral modified three-dimensional index (MNDI);

[0024] Figure 2(a)-Figure 2(h) The average reflectance graph of soil organic matter spectra with different pretreatments;

[0025] Figure 3(a)-Figure 3(h) It is the correlation coefficient matrix diagram between the two-dimensional index (NDI) of different pre-processed spectra and the soil organic matter content;

[0026] Figure 4(a)-Figure 4(h) It is the slice diagram of the correlation coefficient between the three-dimensional index (MNDI) of different pre-processed spectra and the soil organic matter content;

[0027] Figure 5 This is the maximum absolute correlation coefficient (MAr) diagram between the modified three-dimensional index (MNDI) constructed for each preprocessed spectrum at different values ​​of the correction factor (k) and the soil organic matter content. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0029] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the soil organic matter spectrum data collected from the field measured in the coastal plain of Laizhou Bay, Bohai Sea, Shandong Province is taken as an example to illustrate the specific implementation mode of the present invention with reference to the accompanying drawings. According to research, adding a third band within a specific wavelength domain can further enhance the sensitivity of the spectral index to the soil organic matter content and improve the estimation accuracy. In addition, spectral preprocessing can also effectively reduce the impact of background noise. Estimating the soil organic matter content based on the modified three-dimensional index (MNDI) of the preprocessed spectrum is a feasible way. In view of this, this embodiment proposes a method and system for estimating the soil organic matter content using a hyperspectral modified three-dimensional index (MNDI).

[0030] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0031] Example 1

[0032] The method for estimating soil organic matter content using hyperspectral modified three-dimensional index (MNDI) proposed in this embodiment includes the following steps:

[0033] Step A: Obtain field measured soil spectral data and preprocess the spectral data;

[0034] Step B: The two-dimensional index (NDI) and the modified three-dimensional index (MNDI) were constructed using the Spearman correlation coefficient (r) and the band combination optimization method, and the soil organic matter estimation model was constructed using the support vector machine (SVM) modeling method;

[0035] Step C: According to the correlation between the modified three-dimensional index (MNDI) and soil organic matter formed by each preprocessed spectrum under different correction factors (k) and the change in model accuracy, the optimal spectrum preprocessing method and optimal parameters are determined. 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model (R 2 The soil organic matter content was estimated by using the largest value, the smallest RMSE value, and the largest RPD value (greater than 2).

[0036] The present invention proposes to pre-process the soil spectrum, such as Figure 2(a)-Figure 2(h) As shown in the figure, the method performs moving window smoothing (SMO), Savitzky-Golay first-order derivative (SGD1), Savitzky-Golay second-order derivative (SGD2), standard normal variable transformation (SNV), standard normal variable transformation-detrending (SNV-DT), multivariate scatter correction (MSC) and maximum normalization (NOR) preprocessing on the acquired field measured soil spectral data.

[0037] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the moving window smoothing (SMO), Savitzky-Golay first-order derivative (SGD1) and Savitzky-Golay second-order derivative (SGD2) spectral preprocessing in this patent all use an 11mm window.

[0038] Figure 2(a) is a reflectance graph of the original spectrum; Figure 2(b)-Figure 2(h) These are the spectral reflectance maps preprocessed by moving window smoothing (SMO), Savitzky-Golay first derivative (SGD1), Savitzky-Golay second derivative (SGD2), standard normal variate transformation (SNV), standard normal variate transformation-detrended (SNV-DT), multivariate scatter correction (MSC) and maximum normalization (NOR).

[0039] The present invention proposes a method of using the Spearman correlation coefficient (r) and band combination optimization to construct a two-dimensional index (NDI) and a modified three-dimensional index (MNDI), and uses a support vector machine (SVM) modeling method to construct a soil organic matter estimation model, such as Figure 3(a)-Figure 3(h) and Figure 4(a)-Figure 4(h) , calculating the Spearman correlation coefficient (r) between the two-dimensional index (NDI) and the modified three-dimensional index (MNDI) composed of each band combination and the soil organic matter, and using the support vector machine (SVM) modeling method to construct a soil organic matter estimation model; the method comprises the following steps:

[0040] B: The two-dimensional index (NDI) and modified three-dimensional index (MNDI) were constructed using the Spearman correlation coefficient (r) and band combination optimization method, and the soil organic matter estimation model was constructed using the support vector machine (SVM) modeling method.

[0041] a: Calculate the Spearman correlation coefficient (r) between the two-dimensional index (NDI) composed of the two-band combination and soil organic matter;

[0042] Figure 3(a) is a matrix diagram of the Spearman correlation coefficient (r) between the two-dimensional index (NDI) composed of the original spectra and soil organic matter; Figure 3(b)-Figure 3(h) The matrix diagram shows the Spearman correlation coefficient (r) between the two-dimensional index (NDI) and soil organic matter composed of spectra preprocessed by moving window smoothing (SMO), Savitzky-Golay first derivative (SGD1), Savitzky-Golay second derivative (SGD2), standard normal variate transformation (SNV), standard normal variate transformation-detrended (SNV-DT), multivariate scatter correction (MSC) and normalization to the maximum (NOR).

[0043] b: Calculate the Spearman correlation coefficient (r) between the modified three-dimensional index (MNDI) composed of the three-band combination and soil organic matter;

[0044] Figure 4(a) is a slice diagram of the Spearman correlation coefficient (r) between the three-dimensional index (MNDI) composed of the original spectrum and soil organic matter; Figure 4(b)-Figure 4(h) Slice plots of the Spearman correlation coefficient (r) between the three-dimensional index (MNDI) and soil organic matter composed of spectra preprocessed by moving window smoothing (SMO), Savitzky-Golay first derivative (SGD1), Savitzky-Golay second derivative (SGD2), standard normal variate transformation (SNV), standard normal variate transformation-detrended (SNV-DT), multivariate scatter correction (MSC) and normalization to maximum (NOR) respectively.

[0045] c: Using the support vector machine (SVM) modeling method, soil organic matter estimation models were constructed based on the two-dimensional index (NDI) and the modified three-dimensional index (MNDI).

[0046] The calculation formula of the two-dimensional index (NDI) is as follows:

[0047]

[0048] The calculation formula of the modified three-dimensional index (MNDI) is as follows:

[0049]

[0050] Among them, R λ1 , R λ2 and R λ3 are the reflectance values ​​of bands λ1, λ2 and λ3 in the range of 325-1075nm, R λ1 ≠R λ ≠R λ3 k is a correction factor, and when k=0, the modified three-dimensional index (MNDI) is the two-dimensional index (NDI).

[0051] Table 1 Comparison of model estimation results

[0052]

[0053] The present invention proposes to determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index (MNDI) and soil organic matter formed by each preprocessed spectrum under different correction factor (k) values ​​and the change in model accuracy, such as Figure 5 As shown, the method determines the optimal spectral preprocessing method and the optimal parameter value of the input correction factor (k) according to the changes in the Spearman correlation coefficient (r) between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum at different correction factors (k) and soil organic matter and the accuracy of the estimation model; the method comprises the following steps:

[0054] C: Determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index (MNDI) constructed by each preprocessed spectrum under different correction factor (k) values ​​and soil organic matter and the change in model accuracy.

[0055] a: Calculate the Spearman correlation coefficient (r) between the modified three-dimensional index (MNDI) constructed from each preprocessed spectrum at different correction factor (k) values ​​and soil organic matter;

[0056] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the correction factor (k) in this patent ranges from -5 to 5 (step size 0.5). When k=0, the modified three-dimensional index (MNDI) is the two-dimensional index (NDI).

[0057] Figure 5 This is the maximum absolute correlation coefficient (MAr) diagram between the modified three-dimensional index (MNDI) constructed for each preprocessed spectrum at different values ​​of the correction factor (k) and the soil organic matter content.

[0058] b: The soil organic matter estimation model was constructed using the modified three-dimensional index (MNDI) constructed by each preprocessed spectra under different correction factor (k) values ​​and the support vector machine (SVM) modeling method;

[0059] c: Determine the optimal spectral preprocessing method and the optimal parameter value of the input correction factor (k) based on the Spearman correlation coefficient (r) and the change in the accuracy of the estimation model.

[0060] d: Absolute coefficient (R 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) are used to evaluate the model estimation performance. 2 The larger the value and the smaller the RMSE value, the higher the accuracy of the estimation model. The PRD value is interpreted using a three-level classification standard. If the RPD is greater than 2, it means that the estimation model has ideal accuracy; if the RPD value is between 1.8 and 2, it means that the estimation model has the ability to quantitatively evaluate soil organic matter; if the RPD value is less than 1.4, the estimation model is considered unusable. The best model (with the largest R2 value, the smallest RMSE value, and the RPD value greater than 2 and the largest) is selected for soil organic matter content estimation.

[0061] Example 2

[0062] Based on the same inventive concept, this embodiment also provides a system for estimating soil organic matter content using a hyperspectral corrected three-dimensional index. Since the principle of solving the problem by this system is similar to the aforementioned method for estimating soil organic matter content using a hyperspectral corrected three-dimensional index, the implementation of this system can be achieved according to the specific steps of the aforementioned method, and the repeated parts will not be repeated.

[0063] The system proposed in this embodiment for estimating soil organic matter content using a hyperspectral corrected three-dimensional index includes:

[0064] Spectral preprocessing module: configured to obtain field measured soil spectral data and preprocess the spectral data;

[0065] Index construction and modeling module: It is configured to use the Spearman correlation coefficient and the band combination optimization method to construct a two-dimensional index and a modified three-dimensional index respectively, and use the support vector machine modeling method to build a soil organic matter estimation model;

[0066] Spectral preprocessing method and parameter optimization module: It is configured to determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index and soil organic matter formed by each preprocessed spectrum under different correction factor values ​​and the change in model accuracy. The absolute coefficient (R 2 ), root mean square error (RMSE) and residual prediction deviation (RPD) were used to evaluate the model estimation performance, and the best soil organic matter estimation model (R 2 The soil organic matter content was estimated by using the largest value, the smallest RMSE value, and the largest RPD value (greater than 2).

[0067] The specific data processing process of the above-mentioned spectrum preprocessing module, index construction and modeling module, spectrum preprocessing method and parameter optimization module is exactly the same as that of the previous embodiment 1, so it will not be repeated here.

[0068] A person skilled in the art can understand that all or part of the processes in the above embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0069] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0070] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for estimating soil organic matter using a modified three-dimensional spectral index, characterized in that: The steps include: Step A: obtaining measured soil spectral data and preprocessing the spectral data; Step B: Using the Spearman correlation coefficient and the band combination optimization method to construct a two-dimensional index and a modified three-dimensional index, respectively, and using the support vector machine modeling method to build a soil organic matter estimation model; The two-dimensional index NDI: ； The modified three-dimensional index MNDI: ； Among them, R λ1 , R λ2 and R λ3 are the reflectance values ​​corresponding to the wavelength bands λ1, λ2 and λ3 in the range of 325~1075 nm, k is the correction factor, the correction factor k ranges from -5 to 5, the step length is 0.5, k≠0; Step C: Determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index formed by each preprocessed spectrum under different correction factor values ​​and soil organic matter and the change in model accuracy; use the three indicators of absolute coefficient, root mean square error and residual prediction deviation to evaluate the model estimation performance, select the optimal model, and estimate the soil organic matter content; the conditions for the optimal model are: the absolute coefficient value is the largest, the root mean square error value is the smallest, and the residual prediction deviation value is greater than 2 and is the largest.

2. The method according to claim 1, characterized in that: In the step A, measured soil spectral data are obtained and the spectral data are preprocessed: the measured soil spectral data are respectively preprocessed by moving window smoothing, Savitzky-Golay first-order derivative, Savitzky-Golay second-order derivative, standard normal variable transformation, standard normal variable transformation-detrending, multivariate scattering correction and maximum normalization.

3. The method according to claim 1, characterized in that: In the step B, the two-dimensional index and the modified three-dimensional index are respectively constructed using the Spearman correlation coefficient and the band combination optimization method, and the soil organic matter estimation model is constructed using the support vector machine modeling method: the Spearman correlation coefficient between the two-dimensional index and the modified three-dimensional index composed of each band combination and the soil organic matter is calculated, and the soil organic matter estimation model is constructed using the support vector machine modeling method.

4. The method according to claim 1, characterized in that: The Spearman correlation coefficient between the two-dimensional index composed of the two-band combination and the soil organic matter was calculated; the Spearman correlation coefficient between the modified three-dimensional index composed of the three-band combination and the soil organic matter was calculated; and the support vector machine modeling method was used to construct soil organic matter estimation models based on the two-dimensional index and the modified three-dimensional index respectively.

5. The method according to claim 1, characterized in that: In the step C, the optimal spectral preprocessing method and the optimal parameters are determined according to the correlation between the corrected three-dimensional index constructed by each preprocessed spectrum under different correction factor values ​​and the soil organic matter and the change in model accuracy: the optimal spectral preprocessing method and the optimal parameter value of the input correction factor are determined according to the change in the Spearman correlation coefficient between the corrected three-dimensional index constructed by each preprocessed spectrum under different correction factor values ​​and the soil organic matter and the estimation model accuracy.

6. The method according to claim 1, characterized in that: In step C, the larger the absolute coefficient value of the model and the smaller the root mean square error value, the higher the accuracy of the estimation model is considered to be; the residual prediction deviation value is interpreted using the three-level classification standard. If the residual prediction deviation value is greater than 2, it means that the estimation model has ideal accuracy; if the residual prediction deviation value is between 1.8-2, it means that the estimation model has the ability to quantitatively evaluate soil organic matter; if the residual prediction deviation value is less than 1.4, the estimation model is considered to be unusable.

7. A system for estimating soil organic matter using a modified three-dimensional spectral index, characterized in that: include: Spectral preprocessing module: configured to obtain measured soil spectral data and preprocess the spectral data; Index construction and modeling module: It is configured to use the Spearman correlation coefficient and the band combination optimization method to construct a two-dimensional index and a modified three-dimensional index respectively, and use the support vector machine modeling method to build a soil organic matter estimation model; The two-dimensional index NDI: ； The modified three-dimensional index MNDI: ； Among them, R λ1 , R λ2 and R λ3 are the reflectance values ​​corresponding to the wavelength bands λ1, λ2 and λ3 in the range of 325~1075 nm, k is the correction factor, the correction factor k ranges from -5 to 5, the step length is 0.5, k≠0; Spectral preprocessing method and parameter optimization module: It is configured to determine the optimal spectral preprocessing method and optimal parameters based on the correlation between the modified three-dimensional index constructed by each preprocessed spectrum under different correction factor values ​​and soil organic matter and the change in model accuracy; the three indicators of absolute coefficient, root mean square error and residual prediction deviation are used to evaluate the model estimation performance, and the optimal model is selected to estimate the soil organic matter content; the conditions for the optimal model are: the absolute coefficient value is the largest, the root mean square error value is the smallest, and the residual prediction deviation value is greater than 2 and is the largest.

8. The system according to claim 7, characterized in that: In the spectral preprocessing module, the acquired measured soil spectral data are preprocessed by moving window smoothing, Savitzky-Golay first-order derivative, Savitzky-Golay second-order derivative, standard normal variable transformation, standard normal variable transformation-detrending, multivariate scattering correction and maximum normalization.

9. The system according to claim 7, characterized in that: In the index construction and modeling module, the two-dimensional index and the modified three-dimensional index are constructed respectively using the Spearman correlation coefficient and the band combination optimization method, and the support vector machine modeling method is used to construct a soil organic matter estimation model: the Spearman correlation coefficient between the two-dimensional index and the modified three-dimensional index composed of each band combination and the soil organic matter is calculated, and the soil organic matter estimation model is constructed using the support vector machine modeling method.

10. The system according to claim 7, characterized in that: In the spectral preprocessing method and parameter optimization module, the optimal spectral preprocessing method and the optimal parameter value of the input correction factor are determined according to the changes in the Spearman correlation coefficient between the modified three-dimensional index constructed by the correction factor of each preprocessed spectrum with different values ​​and the soil organic matter and the accuracy of the estimation model.

Images