A method and device for predicting soil organic carbon in three-dimensional space based on parameter scaling correction

Abstract

The application discloses a kind of based on parameter scaling correction soil organic carbon three-dimensional space prediction method and device, comprising: obtaining the target depth of soil organic carbon to be predicted;The target depth is substituted into the vertical distribution function of scaling coefficient constructed in advance, and the scaling coefficient under the target depth is obtained;The scaling coefficient under the target depth is used to correct the parameter prediction value of soil depth function of target depth, each parameter refers to the non-independent variable parameter in soil depth function;The target depth, the scaling coefficient under the target depth and the parameter prediction value of soil depth function of target depth are substituted into the soil organic carbon density function of pre-constructed soil depth function parameter scaling correction of adaptive depth, and the soil organic carbon density under the target depth is obtained.The application effectively reduces the error generated at different depths due to least square method and function overfitting when using soil depth function to predict soil organic carbon three-dimensional space.And this application can be applied to any depth interval, to obtain higher accuracy soil organic carbon three-dimensional space distribution in any depth interval.

Description

Technical Field

[0001] This invention relates to a method and apparatus for three-dimensional spatial prediction of soil organic carbon based on parameter scaling correction, belonging to the field of soil organic carbon density prediction technology. Background Technology

[0002] Spatial distribution data of soil organic carbon density are primarily obtained from soil datasets. Commonly used soil datasets globally include the Harmonized World Soil Database (HWSD) and SoilGrids250m; within China, commonly used datasets include the China High-Resolution National Soil Information Grid Basic Attributes Dataset and the China Soil Characteristics Dataset. However, these datasets provide soil data within fixed depth intervals. For example, HWSD's soil data is divided into two intervals: 0-30cm and 30-100cm, while NSIGC's soil data is divided into six intervals: 0-5cm, 5-15cm, 15-30cm, 30-60cm, 60-100cm, and 100-200cm. Fixed-depth soil data is not easily utilized for different research needs. Furthermore, because the data for each interval are predicted independently, a continuous distribution of soil organic carbon density with depth across a complete profile cannot be obtained. In addition, most of these datasets rely on machine learning methods for prediction, lacking interpretability.

[0003] Soil depth functions are commonly used to describe the vertical continuous distribution characteristics of soil organic carbon with depth. They clearly capture the complex variations of soil organic carbon at different depths, allowing for a comprehensive analysis of its continuous distribution within the soil profile. In soil mapping, soil depth functions have been widely applied to obtain the three-dimensional continuous distribution of soil organic carbon. However, because function fitting follows a global optimum principle, the prediction results of depth functions exhibit different errors at different depths, particularly in the surface and deep soil layers. Furthermore, soil samples with fewer layers in the training set are prone to overfitting during function fitting, which also affects the final parameter prediction results. Therefore, it is necessary to correct the predicted parameters of the soil depth function. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a method and device for three-dimensional spatial prediction of soil organic carbon based on parameter scaling correction.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] On one hand, this invention discloses a three-dimensional spatial prediction method for soil organic carbon based on parameter scaling correction, comprising:

[0007] Obtain the target depth of the soil organic carbon to be predicted;

[0008] Substituting the target depth into the pre-constructed vertical distribution function of the scaling factor, the scaling factor at that target depth is obtained;

[0009] The predicted values ​​of each parameter of the soil depth function at the target depth are corrected using the scaling factor at the target depth. Each parameter refers to the non-independent variable parameter in the soil depth function.

[0010] The target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth are substituted into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, to obtain the soil organic carbon density at the target depth.

[0011] Furthermore, the construction of the vertical distribution function of the scaling factor includes:

[0012] Collect measured profile data of soil properties to be predicted and corresponding environmental variable data for the target area;

[0013] The measured soil property profile data and the corresponding environmental variable data were divided into a training set and a validation set.

[0014] By fitting the vertical distribution of cumulative soil organic carbon density with depth for each profile in the training set using the soil depth function, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained.

[0015] Based on the corresponding environmental variable data, linear or nonlinear prediction models are trained using the parameter values ​​of the corresponding soil depth function of each profile in the training set, and the trained linear or nonlinear prediction models are obtained.

[0016] Based on environmental variables covering the entire domain, the values ​​of each parameter in the entire domain are inverted using a trained linear or nonlinear prediction model.

[0017] A correction method was established using the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function. The scaling correction coefficient at a fixed depth and the vertical distribution function of the scaling coefficient with depth were obtained.

[0018] Furthermore, the construction of the soil organic carbon density function with adapted soil depth function parameter scaling correction includes:

[0019] The soil depth function f(x) is used to obtain the soil organic carbon to be predicted;

[0020] By using scaling factors to limit the magnitude of each parameter in the soil depth function f(x) at different depths, the soil organic carbon density function with scaled correction for the soil depth function parameters at the desired depth is obtained.

[0021] Furthermore, the method for establishing a correction by utilizing the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function, yields the scaling correction coefficient at a fixed depth and the vertical distribution function of the scaling coefficient with depth, including:

[0022] Obtain the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set;

[0023] Obtain the predicted values ​​of each parameter of the soil depth function;

[0024] Based on the measured values ​​of organic carbon density and the predicted values ​​of various parameters, several soil depth values ​​are selected, and the least squares method is used to globally fit the entire validation set to obtain the optimal scaling coefficients of the soil depth function for different soil depth values. The number of optimal scaling coefficients at each soil depth value is consistent with the number of non-independent parameters in the soil depth function. For each optimal scaling coefficient fitted at each soil depth value, its value at each soil depth value is used as the dependent variable, and each soil depth value is used as the independent variable. The least squares method is then used to fit the vertical continuous distribution of each scaling coefficient with depth.

[0025] Furthermore, the plurality of soil depth values ​​are at least three soil depth values.

[0026] Secondly, the present invention also discloses a three-dimensional spatial prediction device for soil organic carbon based on parameter scaling correction, comprising:

[0027] The acquisition module is used to acquire the target depth of soil organic carbon to be predicted;

[0028] The first substitution module is used to substitute the target depth into the pre-constructed vertical distribution function of the scaling factor to obtain the scaling factor at the target depth.

[0029] The correction module is used to correct the predicted values ​​of each parameter of the soil depth function at the target depth using the scaling factor at the target depth. The parameters refer to the non-independent variable parameters in the soil depth function.

[0030] The second input module is used to substitute the target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, so as to obtain the soil organic carbon density at the target depth.

[0031] Thirdly, the present invention also discloses a computer-readable storage medium for storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the method.

[0032] Fourthly, the present invention also discloses a computer device, comprising,

[0033] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing the method.

[0034] The beneficial effects achieved by this invention are as follows:

[0035] This invention effectively reduces the errors caused by least squares method and function overfitting at different depths when using soil depth functions to predict the three-dimensional spatial distribution of soil organic carbon. Furthermore, this invention can be applied to any depth range to obtain a more accurate three-dimensional spatial distribution of soil organic carbon within any depth range. Attached Figure Description

[0036] Figure 1 This is a flowchart illustrating the implementation of the present invention;

[0037] Figure 2 This corresponds to the vertical distribution of each scaling factor obtained in the verification set in embodiments 2 and 3. Figure 2 In this context, (a) corresponds to the scaling factor a in Example 2. Figure 2 (c) in the figure corresponds to the scaling factor k in Example 2; Figure 2 (b) in the example corresponds to the scaling factor a in Example 3. Figure 2 (d) in the figure corresponds to the scaling factor k in Example 3.

[0038] Figure 3 The spatial distribution of soil organic carbon density predicted by the logarithmic function after scaling factor correction in Example 2 at depth intervals of (a) 0-5cm, (b) 5-15cm, (c) 15-30cm, (d) 30-60cm and (e) 60-100cm. Detailed Implementation

[0039] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0040] Example 1: This example introduces a three-dimensional spatial prediction method for soil organic carbon, such as... Figure 1 As shown, the feature is that it includes:

[0041] Obtain the target depth of the soil organic carbon to be predicted;

[0042] Substituting the target depth into the pre-constructed vertical distribution function of the scaling factor, the scaling factor at that target depth is obtained;

[0043] The predicted values ​​of each parameter of the soil depth function at the target depth are corrected using the scaling factor at the target depth. Each parameter refers to the non-independent parameter in the soil depth function.

[0044] The target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth are substituted into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, to obtain the soil organic carbon density at the target depth.

[0045] The construction of the vertical distribution function of the scaling factor includes:

[0046] Collect measured profile data of soil properties to be predicted and corresponding environmental variable data for the target area;

[0047] The measured soil property profile data and the corresponding environmental variable data were divided into a training set and a validation set.

[0048] By fitting the vertical distribution of cumulative soil organic carbon density with depth for each profile in the training set using the soil depth function, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained.

[0049] Based on the corresponding environmental variable data, linear or nonlinear prediction models are trained using the parameter values ​​of the corresponding soil depth function of each profile in the training set, and the trained linear or nonlinear prediction models are obtained.

[0050] Based on environmental variables covering the entire domain, the values ​​of each parameter in the entire domain are inverted using a trained linear or nonlinear prediction model.

[0051] A correction method was established using the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function. The scaling correction coefficient at a fixed depth and the vertical distribution function of the scaling coefficient with depth were obtained.

[0052] The construction of the soil organic carbon density function with adaptive depth function parameter scaling correction includes:

[0053] The soil depth function f(x) is used to obtain the soil organic carbon to be predicted;

[0054] By using scaling factors to limit the magnitude of each parameter in the soil depth function f(x) at different depths, the soil organic carbon density function with scaled correction for the soil depth function parameters at the desired depth is obtained.

[0055] The method establishes a correction method using the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function, obtaining the scaling correction coefficient at a fixed depth and the vertical distribution function of the scaling coefficient with depth, including:

[0056] Obtain the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set;

[0057] Obtain the predicted values ​​of each parameter of the soil depth function;

[0058] Based on the measured values ​​of organic carbon density and the predicted values ​​of each parameter, at least three soil depth values ​​are selected, and the least squares method is used to globally fit the entire validation set to obtain the optimal scaling coefficients of the soil depth function for different soil depth values. The number of optimal scaling coefficients at each soil depth value is consistent with the number of non-independent parameters in the soil depth function. For each optimal scaling coefficient fitted at each soil depth value, its value at each soil depth value is used as the dependent variable, and each soil depth value is used as the independent variable. The least squares method is then used to fit the vertical continuous distribution of each scaling coefficient with depth.

[0059] Example 2, based on the same inventive concept as Example 1, introduces a method for correcting the parameters of a soil depth function to adapt to different depths. This correction method uses scaling factors to adjust the parameters of the soil depth function at different depths. Taking the use of a logarithmic function as the soil depth function to fit the vertical continuous distribution of cumulative soil organic carbon density with depth as an example (exponential decay functions, power functions, inverse functions, etc., can be used), it can be expressed by equation (1):

[0060] ASOCD x =p1ln(x)+p2 (1)

[0061] Among them, ASOCD x is the cumulative soil organic carbon density at depth x in the soil profile, where x is the absolute depth to the soil surface, and p1 and p2 are the parameters of the function.

[0062] Our correction method can then be expressed as equation (2):

[0063]

[0064] in It is the measured cumulative soil organic carbon value at depth x in the validation set soil profile. These are the predicted values ​​of the function parameters corresponding to the soil profile, while a x ,k x It corresponds to and The scaling factor a,k located at depth x.

[0065] Based on the specifications defined by GlobalSoilMap, we selected the least squares method to fit the scaling factors at depths of 5, 15, 30, 60, and 100 cm. Furthermore, to ensure that the scaling factors adapt to depth variations and can be used to correct parameter predictions within any depth range of 0-100 cm, we additionally constructed a vertically continuous distribution of the scaling factors for each parameter with depth, such as... Figure 2 As shown, the predicted soil organic carbon density at a specific depth is obtained by the difference in cumulative soil organic carbon density at different depths. Therefore, the predicted soil organic carbon density at any depth within the range of 0-100 cm can be calculated using formula (3) after correction by a scaling factor.

[0066]

[0067] in, This refers to the soil organic carbon density at depth intervals x1-x2 cm corresponding to the grid. These correspond to the logarithmic function a, and k corresponds to the scaling factor obtained by fitting at depth x. These are the parameters predicted by the logarithmic function at the corresponding grid location.

[0068] By constructing a vertically continuous distribution of scaling factors with depth and combining it with the predicted depth function parameter map, a more accurate spatial distribution map of soil organic carbon density in any depth range can be generated. For example... Figure 3 As shown, this method was used to obtain the spatial distribution of soil organic carbon density at depth intervals of (a) 0-5cm, (b) 5-15cm, (c) 15-30cm, (d) 30-60cm and (e) 60-100cm in the Yangtze River Delta region.

[0069] The process of obtaining the scaling correction factor and the vertical distribution of the scaling factor with depth includes:

[0070] Collect measured profile data of soil properties and environmental variable data;

[0071] The measured soil property dataset and the corresponding environmental variable data were evenly divided into training set, validation set, and test set.

[0072] By using the logarithmic function as the soil depth function to fit the vertical distribution of the cumulative soil organic carbon density with depth for each profile in the training set, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained.

[0073] Based on environmental variable data, a nonlinear random forest model is trained using the parameter values ​​of each profile in the training set.

[0074] Based on environmental variables covering the entire domain, the trained random forest model is used to retrieve the parameter values ​​for the entire domain.

[0075] A correction method was established using the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function. The scaling correction coefficient at a fixed depth and the vertical distribution of the scaling coefficient with depth were obtained.

[0076] Example 3: This example introduces a three-dimensional spatial prediction method for soil organic carbon, characterized by comprising:

[0077] Taking the use of a power function as a function of soil depth to fit the vertical continuous distribution of cumulative soil organic carbon density with depth as an example, it can be expressed by equation (4):

[0078]

[0079] Among them, ASOCD x is the cumulative soil organic carbon density at depth x in the soil profile, where x is the absolute depth to the soil surface, and p1 and p2 are the parameters of the function.

[0080] Our correction method can then be expressed as equation (5):

[0081]

[0082] in, It is the measured cumulative soil organic carbon value at depth x in the validation set soil profile. These are the predicted values ​​of the function parameters corresponding to the soil profile, while a x ,k x It corresponds to and The scaling factor a,k located at depth x.

[0083] Based on the specifications defined by GlobalSoilMap, we selected the least squares method to fit the scaling factors at depths of 5, 15, 30, 60, and 100 cm. Furthermore, to ensure that the scaling factors adapt to depth variations and can be used to correct parameter predictions within any depth range of 0-100 cm, we additionally constructed a vertically continuous distribution of the scaling factors for each parameter with depth, such as... Figure 2As shown, the predicted soil organic carbon density at a specific depth is obtained by the difference in cumulative soil organic carbon density at different depths. Therefore, the predicted soil organic carbon density at any depth within the range of 0-100 cm can be calculated using formula (6) after correction by a scaling factor.

[0084]

[0085] in, This refers to the soil organic carbon density at depth intervals x1-x2 cm corresponding to the grid. These correspond to the logarithmic function a, and k corresponds to the scaling factor obtained by fitting at depth x. These are the parameters predicted by the logarithmic function at the corresponding grid location.

[0086] By constructing a vertically continuous distribution of scaling factors with depth and combining it with the predicted depth function parameter map, a more accurate spatial distribution map of soil organic carbon density in any depth range can be generated.

[0087] The process of obtaining the scaling correction factor and the vertical distribution of the scaling factor with depth includes:

[0088] Collect measured profile data of soil properties and environmental variable data;

[0089] The measured soil property dataset and the corresponding environmental variable data were evenly divided into training set, validation set, and test set.

[0090] By using a power function as the soil depth function to fit the vertical distribution of the cumulative soil organic carbon density with depth for each profile in the training set, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained.

[0091] Based on environmental variable data, a linear multiple linear regression model is trained using the parameter values ​​of each profile in the training set.

[0092] Based on environmental variables covering the entire domain, the trained multiple linear regression model is used to deduce the parameter values ​​of the entire domain.

[0093] A correction method was established using the measured values ​​of cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set and the predicted values ​​of each parameter of the soil depth function. The scaling correction coefficient at a fixed depth and the vertical distribution of the scaling coefficient with depth were obtained.

[0094] Example 4, based on the same inventive concept as other examples, introduces a three-dimensional spatial prediction device for soil organic carbon, comprising:

[0095] The acquisition module is used to acquire the target depth of soil organic carbon to be predicted;

[0096] The first substitution module is used to substitute the target depth into the pre-constructed vertical distribution function of the scaling factor to obtain the scaling factor at the target depth.

[0097] The correction module is used to correct the predicted values ​​of the parameters of the soil depth function at the target depth using the scaling factor at the target depth.

[0098] The second input module is used to substitute the target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, so as to obtain the soil organic carbon density at the target depth.

[0099] Example 5, based on the same inventive concept as other examples, describes a computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the method.

[0100] Example 6, based on the same inventive concept as other examples, this example provides a computer device, including:

[0101] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing the method.

[0102] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0104] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0106] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A three-dimensional spatial prediction method for soil organic carbon based on parameter scaling correction, characterized in that, include: Obtain the target depth of the soil organic carbon to be predicted; Substituting the target depth into the pre-constructed vertical distribution function of the scaling factor, the scaling factor at that target depth is obtained; The predicted values ​​of each parameter of the soil depth function at the target depth are corrected using the scaling factor at the target depth. Each parameter refers to the non-independent variable parameter in the soil depth function. Substitute the target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, and obtain the soil organic carbon density at the target depth. The construction of the vertical distribution function of the scaling factor includes: Collect measured profile data of soil properties to be predicted and corresponding environmental variable data for the target area; The measured soil property profile data and the corresponding environmental variable data were divided into a training set and a validation set. By fitting the vertical distribution of cumulative soil organic carbon density with depth for each profile in the training set using the soil depth function, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained. Based on the corresponding environmental variable data, linear or nonlinear prediction models are trained using the parameter values ​​of the corresponding soil depth function of each profile in the training set, and the trained linear or nonlinear prediction models are obtained. Based on environmental variables covering the entire domain, the values ​​of each parameter in the entire domain are inverted using a trained linear or nonlinear prediction model. A correction method is established using the measured values ​​of cumulative soil organic carbon density at fixed depths in each soil profile of the validation set and the predicted values ​​of various parameters of the soil depth function. This yields the scaling correction coefficients at fixed depths and the vertical distribution function of the scaling coefficients with depth, specifically including: Obtain the measured cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set; Obtain the predicted values ​​of each parameter of the soil depth function; Based on the measured values ​​of organic carbon density and the predicted values ​​of various parameters, several soil depth values ​​are selected, and the least squares method is used to globally fit the entire validation set to obtain the optimal scaling coefficients of the soil depth function for different soil depth values. The number of optimal scaling coefficients at each soil depth value is consistent with the number of non-independent parameters in the soil depth function. For each optimal scaling coefficient fitted at each soil depth value, its value at each soil depth value is used as the dependent variable, and each soil depth value is used as the independent variable. The least squares method is then used to fit the vertical continuous distribution of each scaling coefficient with depth.

2. The method for three-dimensional spatial prediction of soil organic carbon based on parameter scaling correction according to claim 1, characterized in that, The construction of the soil organic carbon density function with adaptive depth function parameter scaling correction includes: Soil depth function used to obtain the soil organic carbon to be predicted f (x); Scaling factors are used to constrain the soil depth function at different depths. f The values ​​of each parameter in (x) are used to obtain the soil organic carbon density function with the soil depth function parameter scaling correction for the adaptive depth.

3. The method for three-dimensional spatial prediction of soil organic carbon based on parameter scaling correction according to claim 1, characterized in that, The soil depth values ​​are at least three soil depth values.

4. A three-dimensional spatial prediction device for soil organic carbon based on parameter scaling correction, characterized in that, include: The acquisition module is used to acquire the target depth of soil organic carbon to be predicted; The first substitution module is used to substitute the target depth into the pre-constructed vertical distribution function of the scaling factor to obtain the scaling factor at the target depth. The correction module is used to correct the predicted values ​​of each parameter of the soil depth function at the target depth using the scaling factor at the target depth. The parameters refer to the non-independent variable parameters in the soil depth function. The second input module is used to substitute the target depth, the scaling factor at the target depth, and the predicted values ​​of each parameter of the soil depth function at the target depth into the pre-constructed soil organic carbon density function with scaled correction for soil depth function parameters adapted to the depth, so as to obtain the soil organic carbon density at the target depth. The construction of the vertical distribution function of the scaling factor includes: Collect measured profile data of soil properties to be predicted and corresponding environmental variable data for the target area; The measured soil property profile data and the corresponding environmental variable data were divided into a training set and a validation set. By fitting the vertical distribution of cumulative soil organic carbon density with depth for each profile in the training set using the soil depth function, the parameter values ​​of the soil depth function corresponding to each profile in the training set are obtained. Based on the corresponding environmental variable data, linear or nonlinear prediction models are trained using the parameter values ​​of the corresponding soil depth function of each profile in the training set, and the trained linear or nonlinear prediction models are obtained. Based on environmental variables covering the entire domain, the values ​​of each parameter in the entire domain are inverted using a trained linear or nonlinear prediction model. A correction method is established using the measured values ​​of cumulative soil organic carbon density at fixed depths in each soil profile of the validation set and the predicted values ​​of various parameters of the soil depth function. This yields the scaling correction coefficients at fixed depths and the vertical distribution function of the scaling coefficients with depth, specifically including: Obtain the measured cumulative soil organic carbon density at a fixed depth in each soil profile of the validation set; Obtain the predicted values ​​of each parameter of the soil depth function; Based on the measured values ​​of organic carbon density and the predicted values ​​of various parameters, several soil depth values ​​are selected, and the least squares method is used to globally fit the entire validation set to obtain the optimal scaling coefficients of the soil depth function for different soil depth values. The number of optimal scaling coefficients at each soil depth value is consistent with the number of non-independent parameters in the soil depth function. For each optimal scaling coefficient fitted at each soil depth value, its value at each soil depth value is used as the dependent variable, and each soil depth value is used as the independent variable. The least squares method is then used to fit the vertical continuous distribution of each scaling coefficient with depth.

5. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods of claims 1 to 3.

6. A computer device, characterized in that, include, One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing the method of any of claims 1 to 3.

Images