A deep soil nitrogen content monitoring method, a server and a storage medium

Abstract

The application discloses a deep soil nitrogen content monitoring method, a server and a storage medium, and belongs to the field of soil component monitoring, and comprises the following steps: acquiring remote sensing image data of surface soil of a soil sampling point and collecting soil samples of the corresponding soil sampling point; performing pretreatment, spectral transformation and spectral feature extraction on the remote sensing image data, and constructing a multi-dimensional feature space; analyzing the corresponding soil samples to obtain variation rules of surface nitrogen and deep nitrogen, and obtaining a surface-deep nitrogen relationship model; filtering a feature band combination related to surface nitrogen content exceeding a threshold from the multi-dimensional feature space, and establishing a surface nitrogen content inversion model; in actual use, after collecting remote sensing image data, the surface nitrogen content inversion model and the surface-deep nitrogen relationship model are used for monitoring deep soil nitrogen content. The method avoids the problems of high cost and low timeliness of traditional sampling analysis.

Description

Technical Field

[0001] This application belongs to the field of soil composition monitoring, and specifically relates to a method, server and storage medium for monitoring nitrogen content in deep soil. Background Technology

[0002] Soil nitrogen content is closely related to the protein content of crop products and is an important component of soil nutrients. It is a crucial indicator of soil quality, and accurately determining soil nitrogen content is essential for crop growth, especially given the increasing importance of nitrogen in deep soils for both crop growth and environmental protection. Currently, soil nitrogen detection mainly relies on the following methods: Laboratory chemical analysis methods: Nitrogen content is determined by collecting soil samples and conducting chemical analysis (such as the Kjeldahl method, ultraviolet-visible spectroscopy, etc.). This method has high accuracy because it requires field collection of soil samples. Nitrogen that crops can directly absorb mainly exists in the forms of nitrate nitrogen and ammonium nitrogen, but these two forms of nitrogen are prone to change, thus requiring prompt laboratory testing. However, due to factors such as the distance between sampling points and laboratories, timely processing is often not possible, requiring low-temperature preservation. In cases with a large monitoring area and numerous soil samples, low-temperature preservation in the field is inconvenient, easily leading to inaccurate test results. This method results in a large workload for sample collection, high time costs, and limited spatial coverage, making it difficult to achieve large-area, continuous nitrogen monitoring.

[0003] Near-ground remote sensing analysis methods: These methods use hyperspectral sensors to acquire in-situ surface soil spectra and then employ empirical models or machine learning methods to invert and determine soil nitrogen content. However, this method is limited to surface soil, and there is currently a lack of effective methods to apply it to deep soil nitrogen monitoring.

[0004] Geophysical analysis methods: These methods use neutron detection or soil sensors to directly detect deep soil information. However, these methods involve high equipment costs and complex data acquisition processes, making them difficult to promote and use on a large scale.

[0005] Therefore, there is an urgent need for a technical solution for monitoring nitrogen content in deep soil that can solve the above problems. Summary of the Invention

[0006] To address the shortcomings of the existing technology, this application provides a method, server, and storage medium for monitoring nitrogen content in deep soil. It utilizes hyperspectral remote sensing to obtain the spectral reflectance of surface soil, eliminates bands with high noise levels, analyzes the correlation between the transformation forms of various spectral curves and surface soil nitrogen information, inverts the nitrogen content of the soil surface, and then indirectly inverts the nitrogen content of deep soil by establishing the relationship between nitrogen content in the soil surface and deep soil.

[0007] The technical effect to be achieved in this application is accomplished through the following solution: According to a first aspect of this application, a method for monitoring nitrogen content in deep soil is provided, comprising the following steps: Step 1: Obtain remote sensing image data of the surface soil at the soil sampling points, and collect soil samples from the corresponding soil sampling points; Step 2: Preprocess, transform, and extract spectral features from the remote sensing image data to construct a multidimensional feature space; Step 3: Analyze the corresponding soil samples to obtain the variation patterns of surface nitrogen and deep nitrogen, and obtain the surface-deep nitrogen relationship model; Step 4: Select feature band combinations from the multidimensional feature space that are correlated with surface nitrogen content by more than a threshold, and establish a surface nitrogen content inversion model; Step 5: In actual use, after collecting remote sensing image data, the nitrogen content in deep soil is monitored using the surface nitrogen content inversion model and the surface-deep nitrogen relationship model.

[0008] Preferably, in step 1, a suitable remote sensing platform is selected according to the monitoring range to acquire remote sensing image data. The remote sensing platform includes satellite remote sensing, a hyperspectral sensor carried by an unmanned aerial vehicle, and a ground hyperspectral instrument.

[0009] Preferably, several soil samples at different depths are collected at soil sampling points, and each soil sample is divided into three parts. The first part is stored in an environment of -18°C. The second part is used to determine the nitrate nitrogen and ammonia nitrogen content in a timely manner using fresh soil samples. The third part is air-dried under natural conditions and the total nitrogen content is determined. The soil sampling point data with coordinate records obtained by GPS is converted into points with spatial coordinates.

[0010] Preferably, in step 2, the remote sensing image data is preprocessed and spectral features are extracted. Specifically, wavelet transform combined with sliding window filtering is used to eliminate low signal-to-noise ratio bands; the remaining bands are subjected to continuum removal, first-order differentiation, and logarithmic transformation to construct a multidimensional feature space.

[0011] Preferably, in step 3, the specific method for obtaining the deep nitrogen prediction model is as follows: construct a multiple regression equation, use surface nitrogen and soil physicochemical properties to fit and predict the nitrogen content at different depths, and obtain the deep nitrogen prediction model.

[0012] Preferably, the surface-deep nitrogen relationship model is constructed through regression analysis, and its general formula is defined as:

[0013] Where: E=[T,C,M] is the environmental covariate matrix; N deepNitrogen content in deep soil (target variable); N surface : Surface nitrogen content retrieved from remote sensing (core input); T: Soil type code (e.g., sandy soil = 0, clay = 1) or texture parameter (clay content %); C: Climate factor (normalized annual mean temperature / precipitation); M: Quantitative indicator of management measures (fertilizer application rate kg / ha).

[0014] Preferably, in step 4, the correlation between different transformed spectra and surface nitrogen content is evaluated using the random forest algorithm; XGBoost is used for feature importance weighting, and a surface nitrogen content inversion model is constructed based on the weighted multiple spectral features using the SVR kernel function; and cross-validation is used to evaluate and optimize the model performance.

[0015] Preferably, it further includes a reliability analysis step, specifically: Independent soil samples were collected from different regions to determine the deep nitrogen content, which was then compared with the predicted results. The error was calculated and analyzed using the coefficient of determination R0. 2 The mean square error (RMSE) is used to assess the reliability of the model.

[0016] in, It is the measured value of the i-th sample. It is the predicted value of the i-th sample. It is the average value of the actual measurements.

[0017] According to a second aspect of this application, a server is provided, comprising: a memory and at least one processor; The memory stores a computer program, and the at least one processor executes the computer program stored in the memory to implement the above-described method for monitoring nitrogen content in deep soil.

[0018] According to a third aspect of this application, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed, the computer program implements the above-described method for monitoring nitrogen content in deep soil.

[0019] According to one embodiment of this application, the beneficial effects of using this deep soil nitrogen content monitoring method are as follows: by indirectly inverting deep nitrogen through remote sensing technology, the high cost and low timeliness of traditional sampling and analysis are avoided, and non-destructive monitoring of deep nitrogen is achieved without damaging the soil, providing support for large-scale farmland nitrogen management. To improve monitoring efficiency and applicability, hyperspectral remote sensing is used to rapidly monitor soil within a region, greatly improving spatial and temporal resolution. It is applicable to different types of soil and agricultural environments, thus enhancing the versatility of the method. Optimizing data processing and modeling methods improves monitoring accuracy. By selecting appropriate spectral features and optimizing data processing methods, the accuracy of surface nitrogen inversion is improved. By combining multiple regression methods, a relationship model between surface nitrogen and deep nitrogen is constructed, which improves the reliability of deep nitrogen estimation. Promoting sustainable agricultural development and environmental protection, precise monitoring of soil nitrogen distribution can provide data support for optimizing farmland fertilization and precision agricultural management, reduce excessive application of chemical fertilizers, and improve nitrogen fertilizer utilization efficiency. Monitoring deep nitrogen can be used to assess the risk of nitrogen leaching, reduce groundwater pollution, and improve agricultural sustainability. Attached Figure Description

[0020] To more clearly illustrate the embodiments of this application or the existing technical solutions, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a method for monitoring nitrogen content in deep soil according to an embodiment of this application; Figure 2 This is a structural block diagram of a server according to one embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] like Figure 1 As shown, a method for monitoring nitrogen content in deep soil according to an embodiment of this application includes the following steps: Step 1: Obtain remote sensing image data of the surface soil at the soil sampling points, and collect soil samples from the corresponding soil sampling points; In this step, several soil sampling points are set up within the study area, and a suitable remote sensing platform is selected based on the monitoring range, for example: Satellite remote sensing (such as Zhuhai-1, Hyperion, GF-5, etc.) is suitable for large-scale, periodic monitoring; The drone is equipped with a hyperspectral sensor, which can perform low-altitude patrols or spectral information, and is suitable for high-precision monitoring at small and medium scales. Ground-based hyperspectral imagers can be handheld or mounted on a platform for small-scale experimental verification, improving model reliability.

[0024] Soil samples were collected from the corresponding soil sampling points. The specific method was as follows: soil samples were collected at the soil sampling points, and the actual nitrogen content was determined in the laboratory. The actual nitrogen content was then used to train the model.

[0025] During sampling, soil samples were collected at different depths (e.g., 0-5 cm, 5-10 cm, 10-20 cm, 20-40 cm, 40-60 cm). Each soil sample was divided into three parts. The first part was preserved at -18°C. The second part was used to determine the nitrate nitrogen and ammonia nitrogen content of fresh soil. The third part was air-dried naturally and then the total nitrogen content was determined. The GPS-measured sampling point data with coordinate records was converted into points with spatial coordinates. The coordinates of each sampling point and the nitrogen content data of the sample at that point were imported into ArcGIS to generate a data distribution with coordinate information, thereby mapping the specific raster in the remote sensing data to the ground data.

[0026] Step 2: Preprocess the remote sensing image data and extract spectral features to construct a multidimensional feature space; In this step, the hyperspectral data in the extracted remote sensing image data is preprocessed and the feature bands are optimized. The main method used is wavelet transform combined with sliding window filtering to eliminate low signal-to-noise ratio bands. The filtered bands are processed in various ways, such as continuum removal, first-order differentiation, logarithmic transformation, etc., to enhance the band features and construct a multi-dimensional feature space, that is, multiple spectral features formed by multiple spectral transformations.

[0027] Step 3: Analyze the corresponding soil samples to obtain the variation patterns of surface nitrogen and deep nitrogen, and obtain a deep nitrogen prediction model; In this step, laboratory chemical analysis methods were used to analyze the nitrogen content of each soil sample at the sampling points, and the statistical relationship between surface nitrogen (0-5 cm) and deep nitrogen (5-10 cm, 10-20 cm, 20-40 cm, 40-60 cm) was analyzed. Specifically, the contents of nitrate nitrogen, ammonia nitrogen, and total nitrogen in different soil layers were obtained. Nitrate nitrogen is commonly determined by ultraviolet spectrophotometry. The principle is that nitrate nitrogen has a characteristic absorption peak at 220 nm, but dissolved organic matter in the soil extract also has absorption interference at this wavelength. Therefore, it is necessary to simultaneously measure the absorbance at a wavelength of 275 nm to correct for interference (usually using A220 - (2 (A275) Calculate the corrected absorbance. Ammonia nitrogen is commonly determined using the indophenol blue colorimetric method (salicylic acid-hypochlorite method), where ammonia reacts with salicylic acid and hypochlorite under alkaline conditions to form a blue indophenol complex, and the absorbance is measured at a wavelength of 660 nm. Total nitrogen is determined using the Kjeldahl method: nitrogen compounds in the soil are hydrolyzed and oxidized using concentrated sulfuric acid and a mixed catalyst under strong heat and high temperature, converting nitrogen into ammonium ions; sodium hydroxide is added to the digested solution for distillation; the distilled ammonia gas (NH3) is absorbed by boric acid solution; the absorbent is then titrated with an acid standard solution (such as hydrochloric acid), using a bromocresol green and methyl red ethanol mixture as an indicator; at the titration endpoint, the solution color changes from green to grayish-red (or wine-red); finally, the total nitrogen content is calculated from the amount of acid standard solution consumed.

[0028] Regression analysis was used to determine the variation patterns of nitrogen in the surface and deep layers. Specifically, the relationship model between surface and deep nitrogen was constructed through regression analysis, and its general formula is defined as:

[0029] Where: E=[T,C,M] is the environmental covariate matrix; N deep Nitrogen content in deep soil (target variable); N surface : Surface nitrogen content retrieved from remote sensing (core input); T: Soil type code (e.g., sandy soil = 0, clay = 1) or texture parameter (clay content %); C: Climate factor (normalized annual mean temperature / precipitation); M: Quantitative indicator of management measures (fertilizer application rate kg / ha); In practice, multiple linear regression, piecewise nonlinear regression, or mixed-effects models can be used. Model validation must meet the following requirements: significance test (p-value < 0.05); variance inflation factor (VIF) < 5 (to avoid multicollinearity); and residuals conforming to a normal distribution (Shapiro-Wilk test).

[0030] Step 4: Select feature band combinations from the multidimensional feature space that are correlated with surface nitrogen content exceeding a threshold to establish a surface nitrogen content inversion model; In this step, within the aforementioned multidimensional feature space, due to the high dimensionality, nonlinearity, and noise interference of soil spectral data, random forest is used to assess feature correlation to avoid overfitting. The correlation between different transformed spectra and surface nitrogen content is evaluated using the random forest algorithm. A correlation threshold can be set, and bands above the threshold are defined as high-correlation bands, such as 550 nm, 700 nm, 1450 nm, and 1950 nm. The specific steps include: The preprocessed spectral data X=[X1,X2,...,Xp] (p is the number of bands) and the measured surface nitrogen content Y are input into the random forest model; The contribution of each band to nitrogen content prediction is calculated based on Gini Importance. This importance measures the degree to which a feature reduces data impurity when splitting decision tree nodes. It is obtained by calculating the weighted average of the impurity reduction resulting from the splitting of all relevant nodes across all trees (with weights equal to the proportion of node samples). Specifically, the calculation is performed using the following formula:

[0031] Where: T: Total number of split nodes in the tree; ΔGini(t): Reduction in Gini impurity before and after splitting node t; Nt: Number of samples at node t; N: Total number of samples; M: Total number of decision trees in the random forest.

[0032] Sort by importance value in descending order and select the k bands with the highest correlation to nitrogen content (e.g., importance value > 0.01). During validation, permutation importance analysis was used to verify the key impact of the selected bands on the model performance, ensuring its robustness.

[0033] Then, XGBoost is used for feature importance weighting. Based on the weighted multiple spectral features, a surface nitrogen content inversion model is constructed using the SVR kernel function. This step employs an ensemble learning of Support Vector Regression (SVR) and XGBoost, using the selected high-importance spectral features to train the neural network model, thereby establishing the surface nitrogen content inversion model. Specifically: XGBoost is used for feature importance weighting. The feature weighting formula is as follows:

[0034] These are spectral indices for two specific wavelengths; The scientific basis for setting the RBF parameter is: based on the soil spectral characteristics, the RBF kernel bandwidth parameter is set.

[0035] Where σ is taken as the half-width of the soil characteristic peak (e.g., the half-width of the peak near 1450nm is ≈40nm), γ is calculated to be 0.0003".

[0036] Then, the nonlinear relationship is fitted using the support vector regression (SVR) kernel function (radial basis function, RBF); In this method, due to the deep soil sampling depth, the cost is higher than that of traditional sampling, resulting in a smaller sample size to control costs. Therefore, XGBoost, which is robust to missing values, and SVR, which has excellent generalization performance in small sample scenarios, are selected. The integration of these two technologies can complement each other's advantages and improve the recognition effect of the surface nitrogen content inversion model.

[0037] Cross-validation is used to evaluate model performance, optimize hyperparameters, and improve prediction accuracy.

[0038] Step 5: In actual use, after collecting remote sensing image data, the nitrogen content in deep soil is monitored using the surface nitrogen content inversion model and the deep nitrogen prediction model.

[0039] This step also includes a reliability analysis step, specifically: Independent soil samples were collected from different regions to determine the deep nitrogen content, which was then compared with the predicted results. The error was calculated and analyzed using the coefficient of determination R0. 2 The reliability of the model is assessed by using the Coefficient of Determination and the Root Mean Squared Error (RMSE).

[0040]

[0041]

[0042] in, It is the measured value of the i-th sample. It is the predicted value of the i-th sample. It is the average value of the actual measurements.

[0043] Based on the actual measured data, the prediction accuracy was further improved by improving the data preprocessing method and optimizing the parameters of the surface nitrogen content inversion model and the deep nitrogen prediction model.

[0044] In other embodiments of this application, since the hyperspectral instrument has a high spectral resolution, it affects the improvement of spatial resolution and spatial scale. Therefore, multispectral data is used. In order to avoid the problem of less spectral information and limited sensitive bands in multispectral data, the parameters in the modeling process are adjusted, such as enhancing spectral feature extraction and data fusion, so as to achieve similar technical effects.

[0045] According to a second aspect of this application, a server is provided, comprising: a memory 201 and at least one processor 202; The memory 201 stores a computer program, and the at least one processor 202 executes the computer program stored in the memory 201 to implement the above-mentioned method for monitoring nitrogen content in deep soil.

[0046] According to a third aspect of this application, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed, the computer program implements the above-described method for monitoring nitrogen content in deep soil.

[0047] According to one embodiment of this application, the beneficial effects of using this deep soil nitrogen content monitoring method are as follows: by indirectly inverting deep nitrogen through remote sensing technology, the high cost and low timeliness of traditional sampling and analysis are avoided, and non-destructive monitoring of deep nitrogen is achieved without damaging the soil, providing support for large-scale farmland nitrogen management. To improve monitoring efficiency and applicability, hyperspectral remote sensing is used to rapidly monitor soil within a region, greatly improving spatial and temporal resolution. It is applicable to different types of soil and agricultural environments, thus enhancing the versatility of the method. Optimizing data processing and modeling methods improves monitoring accuracy. By selecting appropriate spectral features and optimizing data processing methods, the accuracy of surface nitrogen inversion is improved. By combining multiple regression methods, a relationship model between surface nitrogen and deep nitrogen is constructed, which improves the reliability of deep nitrogen estimation. Promoting sustainable agricultural development and environmental protection, precise monitoring of soil nitrogen distribution can provide data support for optimizing farmland fertilization and precision agricultural management, reduce excessive application of chemical fertilizers, and improve nitrogen fertilizer utilization efficiency. Monitoring deep nitrogen can be used to assess the risk of nitrogen leaching, reduce groundwater pollution, and improve agricultural sustainability.

[0048] It should be noted that the above detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0049] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0050] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that the embodiments of this application described herein can be implemented in sequences other than those illustrated or described herein.

[0051] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0052] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways, such as rotated 90 degrees or in other orientations, and the spatial relative descriptions used herein will be interpreted accordingly.

[0053] In the detailed description above, reference has been made to the accompanying drawings, which form part of this document. In the drawings, similar symbols typically identify similar parts unless the context otherwise indicates otherwise. The illustrated embodiments described in the detailed specification, drawings, and claims are not intended to be limiting. Other embodiments may be used and other changes may be made without departing from the spirit or scope of the subject matter presented herein.

[0054] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring nitrogen content in deep soil, characterized in that, Includes the following steps: Step 1: Use satellite remote sensing to acquire remote sensing image data of the surface soil at the soil sampling points, and collect soil samples from the corresponding soil sampling points; Step 2: Preprocess, transform, and extract spectral features from the remote sensing image data to construct a multidimensional feature space; The specific method is as follows: wavelet transform combined with sliding window filtering is used to eliminate low signal-to-noise ratio bands; the remaining bands are subjected to continuum removal, first-order differentiation and logarithmic transformation to construct a multidimensional feature space. Step 3: Analyze the corresponding soil samples to obtain the variation patterns of surface nitrogen and deep nitrogen, and obtain the surface-deep nitrogen relationship model; The specific method for obtaining the surface-deep nitrogen relationship model is as follows: construct a multiple regression equation, use surface nitrogen and soil physicochemical properties to fit and predict nitrogen content at different depths, and obtain the surface-deep nitrogen relationship model. Step 4: Select feature band combinations from the multidimensional feature space that are correlated with surface nitrogen content exceeding a threshold, and establish a surface nitrogen content inversion model; specifically including the following steps: The correlation between different transformed spectra and surface nitrogen content was evaluated using the random forest algorithm, including the following steps: The processed spectral data X=[X1,X2,...,Xp] and the measured surface nitrogen content Y were input into the random forest model, where p is the number of bands; the contribution of each band to the nitrogen content prediction was calculated based on Gini importance. This importance measures the degree to which a feature reduces data impurity when splitting decision tree nodes. It is obtained by calculating the weighted average of the impurity reduction caused by the splitting of this feature across all relevant nodes in all trees, specifically using the following formula: ； Where: T: the total number of split nodes in the tree; ΔGini(t): the reduction in Gini impurity before and after splitting node t; Nt: the number of samples at node t; N: the total number of samples; M: the total number of decision trees in the random forest; sorted in descending order of importance value, the top k bands with the highest correlation to nitrogen content were selected. Then, XGBoost is used to weight the features based on their importance. The feature weighting formula is as follows: ,in, These are spectral indices for two specific wavelengths; A surface nitrogen content inversion model was constructed based on weighted spectral features using the SVR kernel function; cross-validation was employed to evaluate and optimize the model performance; the SVR kernel function used was the radial basis function. Step 5: In actual use, after collecting remote sensing image data, the nitrogen content in deep soil is monitored using the surface nitrogen content inversion model and the surface-deep nitrogen relationship model.

2. The method for monitoring nitrogen content in deep soil according to claim 1, characterized in that, Several soil samples were collected at different depths at soil sampling points, and each soil sample was divided into three parts. The first part was kept in an environment of -18°C for preservation. The second part was used to determine the nitrate nitrogen and ammonia nitrogen content in a timely manner using fresh soil samples. The third part was air-dried under natural conditions and the total nitrogen content was determined. The soil sampling point data with coordinate records obtained by GPS was converted into points with spatial coordinates.

3. The method for monitoring nitrogen content in deep soil according to claim 1, characterized in that, The surface-deep nitrogen relationship model is constructed through regression analysis, and its general formula is defined as: Where: E=[T,C,M] is the environmental covariate matrix; N deep Nitrogen content in deep soil; N surface : Surface nitrogen content obtained from remote sensing inversion; T: Soil type code or texture parameter; C: Climate factor; M: Quantitative indicator of management measures.

4. The method for monitoring nitrogen content in deep soil according to claim 1, characterized in that, It also includes a reliability analysis step, specifically: Independent soil samples were collected from different regions to determine the deep nitrogen content, which was then compared with the predicted results. The error was calculated and analyzed using the coefficient of determination R0. 2 The mean square error (RMSE) is used to assess the reliability of the model. in, It is the measured value of the i-th sample. It is the predicted value of the i-th sample. It is the average value of the actual measurements.

5. A server, characterized in that, include: Memory and at least one processor; The memory stores a computer program, and the at least one processor executes the computer program stored in the memory to implement the deep soil nitrogen content monitoring method according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the method for monitoring nitrogen content in deep soil as described in any one of claims 1 to 4.

Images