Training method, prediction method and device of soil nitrogen mineralization rate prediction model

Abstract

The invention provides a training method, a prediction method and a device of a soil nitrogen mineralization rate prediction model, and relates to the technical field of remote sensing. Comprises: acquiring a remote sensing image sample and a soil sample of a monitored area sample; determining a soil nitrogen mineralization rate label corresponding to each observation point in the monitored area sample based on the soil sample; inputting the remote sensing image sample into the initial soil nitrogen mineralization rate prediction model, extracting a plurality of vegetation index characteristics corresponding to each observation point from the remote sensing image sample, and selecting a target vegetation index characteristic based on the plurality of vegetation index characteristics and the soil nitrogen mineralization rate label; and updating model parameters of the initial soil nitrogen mineralization rate prediction model based on the target vegetation index characteristics and the soil nitrogen mineralization rate labels corresponding to the observation points, so as to train and obtain the soil nitrogen mineralization rate prediction model, thereby achieving the purpose of predicting the soil nitrogen mineralization rate under the condition of not being limited by the area of the monitored region. And the calculation efficiency and the calculation precision of the soil nitrogen mineralization rate are effectively improved.

Description

Technical Field

[0001] The present application relates to the field of remote sensing technology, and in particular to a training method, a prediction method and a device for a soil nitrogen mineralization rate prediction model. Background Art

[0002] Soil nitrogen mineralization is an important nitrogen transformation process in the ecosystem, which plays a key role in plant growth and ecosystem function. Soil nitrogen mineralization rate (NMR) can be used to characterize important nitrogen transformation processes in the ecosystem.

[0003] At present, when determining the soil nitrogen mineralization rate, it is usually necessary to conduct field sampling and laboratory analysis of the soil in the monitoring area to determine the soil nitrogen mineralization rate. However, this method is usually only applicable to a small monitoring area and is difficult to apply to a large monitoring area. In addition, this method is time-consuming and labor-intensive, resulting in low calculation efficiency of the soil nitrogen mineralization rate.

[0004] Therefore, how to effectively improve the calculation efficiency of soil nitrogen mineralization rate without being restricted by the area of ​​the monitoring region is a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0005] The present application provides a training method, a prediction method and a device for a soil nitrogen mineralization rate prediction model, which can effectively improve the calculation efficiency of the soil nitrogen mineralization rate without being restricted by the area of ​​the monitoring area.

[0006] The present application provides a training method for a soil nitrogen mineralization rate prediction model, comprising: Acquire remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include multiple observation points; Based on the soil samples of the monitoring area samples, determining the soil nitrogen mineralization rate label corresponding to each of the observation points; Inputting the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extracting a plurality of vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and selecting a target vegetation index feature from the plurality of vegetation index features based on the plurality of vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; Based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each of the observation points, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train a soil nitrogen mineralization rate prediction model, which is used to predict the soil nitrogen mineralization rate.

[0007] According to a training method for a soil nitrogen mineralization rate prediction model provided by the present application, the target vegetation index feature is selected from the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, including: For each of the vegetation index characteristics, based on the vegetation index characteristics corresponding to each of the observation points, the average value of the vegetation index characteristics corresponding to all the observation points, the soil nitrogen mineralization rate label corresponding to each of the observation points, and the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points, determine the correlation coefficient corresponding to the vegetation index characteristics; Based on the correlation coefficients corresponding to the vegetation index features, the vegetation index feature corresponding to the maximum correlation coefficient is determined as the target vegetation index feature.

[0008] According to a training method for a soil nitrogen mineralization rate prediction model provided by the present application, the correlation coefficient corresponding to the vegetation index feature is determined based on the vegetation index feature corresponding to each observation point, the average value of the vegetation index feature corresponding to all observation points, the soil nitrogen mineralization rate label corresponding to each observation point, and the average value of the soil nitrogen mineralization rate labels corresponding to all observation points, including: according to Determining a correlation coefficient corresponding to the vegetation index feature; in, represents the correlation coefficient corresponding to the vegetation index feature, i represents the i-th observation point, N represents the total number of observation points, represents the vegetation index feature corresponding to the i-th observation point, represents the average value of the vegetation index characteristics corresponding to all observation points, represents the soil nitrogen mineralization rate label corresponding to the i-th observation point, It represents the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points.

[0009] According to a training method for a soil nitrogen mineralization rate prediction model provided by the present application, the target vegetation index feature is selected from the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, including: Based on the multiple vegetation index features corresponding to each of the observation points, determining a covariance matrix between the multiple vegetation index features, wherein the covariance matrix is ​​used to characterize the correlation between the multiple vegetation index features; Determining a plurality of principal component vegetation index features from the plurality of vegetation index features based on the eigenvalues ​​and corresponding eigenvectors of the covariance matrix; Based on the correlation between the multiple principal component vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate labels corresponding to each of the observation points, the target vegetation index feature is selected from the multiple principal component vegetation index features.

[0010] The present application also provides a method for predicting soil nitrogen mineralization rate, comprising: Acquire a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; The target remote sensing image is input into a soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate corresponding to each target observation point is predicted through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is a soil nitrogen mineralization rate prediction model trained by any of the above embodiments.

[0011] The present application also provides a training device for a soil nitrogen mineralization rate prediction model, comprising: An acquisition unit, used to acquire remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include a plurality of observation points; A first processing unit is used to determine the soil nitrogen mineralization rate label corresponding to each of the observation points based on the soil samples of the monitoring area samples; A second processing unit is used to input the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extract a plurality of vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and select a target vegetation index feature from the plurality of vegetation index features based on the plurality of vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; An updating unit is used to update the model parameters of the initial soil nitrogen mineralization rate prediction model based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each of the observation points, so as to train a soil nitrogen mineralization rate prediction model, wherein the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

[0012] The present application also provides a device for predicting soil nitrogen mineralization rate, comprising: An acquisition unit, used for acquiring a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; A prediction unit is used to input the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predict the soil nitrogen mineralization rate corresponding to each target observation point through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is the soil nitrogen mineralization rate prediction model obtained by the above training.

[0013] The present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the training method of the soil nitrogen mineralization rate prediction model or the soil nitrogen mineralization rate prediction method as described in any one of the above-mentioned methods is implemented.

[0014] The present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for training a soil nitrogen mineralization rate prediction model or a method for predicting a soil nitrogen mineralization rate as described in any one of the above.

[0015] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the training method of the soil nitrogen mineralization rate prediction model or the soil nitrogen mineralization rate prediction method as described in any one of the above.

[0016] The training method, prediction method and device of the soil nitrogen mineralization rate prediction model provided in the present application are obtained by obtaining remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include multiple observation points; and based on the soil samples of the monitoring area samples, the soil nitrogen mineralization rate label corresponding to each observation point is determined; the remote sensing image samples of the monitoring area samples are input into the initial soil nitrogen mineralization rate prediction model, and the initial soil nitrogen mineralization rate prediction model is used to extract multiple vegetation index features corresponding to each observation point from the remote sensing image samples, and based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, a target vegetation index feature is selected from the multiple vegetation index features; and then based on the target vegetation index feature corresponding to each observation point and the soil nitrogen mineralization rate label, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train and obtain the soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate. In this way, remote sensing technology is used to obtain remote sensing image samples of the monitored area, and the soil nitrogen mineralization rate prediction model is trained by combining the remote sensing image samples and the soil nitrogen mineralization rate labels corresponding to multiple observation points in the monitored area samples. This makes it possible to dynamically predict the soil nitrogen mineralization rate through the soil nitrogen mineralization rate prediction model, and achieves the goal of effectively improving not only the calculation efficiency of the soil nitrogen mineralization rate but also the calculation accuracy of the soil nitrogen mineralization rate without being restricted by the area of ​​the monitored area. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A schematic flow chart of a method for training a soil nitrogen mineralization rate prediction model provided in an embodiment of the present application.

[0019] Figure 2 A schematic diagram of a training framework for a soil nitrogen mineralization rate model provided in an embodiment of the present application.

[0020] Figure 3 A schematic flow chart of a method for predicting soil nitrogen mineralization rate provided in an embodiment of the present application.

[0021] Figure 4 A schematic diagram of a soil nitrogen mineralization rate monitoring system provided in an embodiment of the present application.

[0022] Figure 5 A schematic diagram of a process flow of a training device for a soil nitrogen mineralization rate prediction model provided in an embodiment of the present application.

[0023] Figure 6 A schematic diagram of the flow of a device for predicting soil nitrogen mineralization rate provided in an embodiment of the present application.

[0024] Figure 7 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] In the embodiments of the present application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of the present application, the character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0027] The technical solution provided in the embodiments of the present application can be applied to the fields of agricultural production, ecological environment protection, etc., and provide a basis for scientific decision-making. Since the soil nitrogen mineralization rate can be used to characterize the important nitrogen transformation process in the ecosystem, by calculating the soil nitrogen mineralization rate, a scientific monitoring and evaluation tool can be provided for agricultural production fertilization management, thereby realizing the refined management of crop fertilization.

[0028] At present, it is usually necessary to conduct field sampling and laboratory analysis of the soil in the monitoring area to calculate the soil nitrogen mineralization rate. However, this method is usually only applicable to a small monitoring area and is difficult to apply to a large monitoring area. In addition, this method is time-consuming and labor-intensive, resulting in low calculation efficiency of the soil nitrogen mineralization rate.

[0029] In order to effectively improve the calculation efficiency of soil nitrogen mineralization rate without being restricted by the area of ​​the monitoring area, an embodiment of the present application provides a training method for a soil nitrogen mineralization rate prediction model, which can make full use of remote sensing technology to obtain remote sensing image samples of the monitoring area, and combine the remote sensing image samples and the soil nitrogen mineralization rate labels corresponding to multiple observation points in the monitoring area samples to train the soil nitrogen mineralization rate prediction model, so that the soil nitrogen mineralization rate can be dynamically predicted by the soil nitrogen mineralization rate prediction model, achieving the goal of not only effectively improving the calculation efficiency of the soil nitrogen mineralization rate but also improving the calculation accuracy of the soil nitrogen mineralization rate without being restricted by the area of ​​the monitoring area, thereby better realizing the refined management of crop fertilization.

[0030] It can be understood that the executor of the present method can be an electronic device such as a computer or a server, or it can be a training device for a soil nitrogen mineralization rate prediction model set in the electronic device. The training device for the soil nitrogen mineralization rate prediction model can be implemented by software, hardware, or a combination of both.

[0031] Below, the training method of the soil nitrogen mineralization rate prediction model provided by the present application will be described in detail through the following specific embodiments. It is understandable that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0032] Figure 1 A flow chart of a method for training a soil nitrogen mineralization rate prediction model provided in an embodiment of the present application is provided. For example, see Figure 1 As shown, the training method of the soil nitrogen mineralization rate prediction model may include: S101. Obtain remote sensing image samples of monitoring area samples and soil samples of monitoring area samples; wherein the monitoring area samples include multiple observation points.

[0033] For example, the remote sensing image samples may be multispectral remote sensing images of different resolutions, or may be hyperspectral remote sensing images, etc., and may be specifically set according to actual needs.

[0034] For example, when obtaining remote sensing image samples of monitored area samples, satellite images of monitored area samples can be collected by satellite and used as remote sensing image samples; drone images of monitored area samples can also be collected by drone and used as remote sensing image samples; of course, satellite images and drone images of monitored area samples can also be collected by satellite and drone respectively, and the collected satellite images and drone images can be fused to obtain remote sensing image samples of monitored area samples. The specific settings can be made according to actual needs.

[0035] The remote sensing image samples are used for subsequent training of soil nitrogen mineralization rate models. For example, see Figure 2 As shown, Figure 2 A schematic diagram of a training framework for a soil nitrogen mineralization rate model provided in an embodiment of the present application. For example, when obtaining remote sensing image samples of monitoring area samples for training the soil nitrogen mineralization rate model, remote sensing images of the monitoring area samples can be first collected, and the collected remote sensing images can be preprocessed, such as denoising, radiation correction, geometric correction, atmospheric correction, etc., and the preprocessed remote sensing images can be used as remote sensing image samples of the monitoring area samples.

[0036] It is understandable that when training the soil nitrogen mineralization rate model, in addition to obtaining remote sensing image samples of the monitoring area samples, it is also necessary to determine the soil nitrogen mineralization rate label corresponding to each observation point, that is, to execute the following S102.

[0037] S102. Based on the soil samples in the monitoring area, determine the soil nitrogen mineralization rate label corresponding to each observation point.

[0038] For example, the soil samples of the monitoring area samples may include soil samples of respective multiple observation points, such as soil samples of 0-10 cm.

[0039] For example, when determining the soil nitrogen mineralization rate label corresponding to each observation point based on the soil samples of the monitoring area samples, the soil samples of the monitoring area samples can be first pretreated, such as air-drying, impurity removal, screening, etc., and the pretreated soil samples can be subjected to soil agrochemical analysis to determine the content of soil ammonium nitrogen and soil nitrate nitrogen in the soil samples of each observation point.

[0040] For each viewpoint, after determining the content of soil ammonium nitrogen and soil nitrate nitrogen in the soil sample, the soil nitrogen mineralization rate label corresponding to the observation point can be determined according to the content of soil ammonium nitrogen and soil nitrate nitrogen in the soil sample of the observation point. For details, please refer to the following formula 1.

[0041] Formula 1 in, Indicates the soil nitrogen mineralization rate label corresponding to the observation point, Indicates the content of soil ammonium nitrogen in the soil sample, in mg kg -1 , Indicates the content of soil nitrate nitrogen, in mg kg -1 .

[0042] Through the above formula 1, the soil nitrogen mineralization rate label corresponding to each observation point can be calculated. After obtaining the soil nitrogen mineralization rate label corresponding to each observation point, the remote sensing image sample and the soil nitrogen mineralization rate label corresponding to each observation point can be combined to train the soil nitrogen mineralization rate prediction model, that is, execute the following S103 and S104.

[0043] S103. Input the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extract multiple vegetation index features corresponding to each observation point from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and select the target vegetation index feature from the multiple vegetation index features based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point.

[0044] For example, multiple vegetation index features may include Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Ratio Vegetation Index (RVI), Difference Vegetation Index (DI), and Enhanced Vegetation Index (EVI), which can be set according to actual needs.

[0045] Among them, the normalized difference vegetation index NDVI = (NIR-RED) / (NIR+RED), NIR represents the reflectivity of the near-infrared band, and RED represents the reflectivity of the red light band.

[0046] Green Normalized Vegetation Index GNDVI = (NIR-GREEN) / (NIR+GREEN), NIR represents the reflectance of the near-infrared band, and GREEN represents the reflectance of the green light band.

[0047] Ratio vegetation index RVI=RVI= NIR / RED, NIR represents the reflectivity of the near-infrared band, and RED represents the reflectivity of the red light band.

[0048] The difference vegetation index DI = NIR-RED, NIR represents the reflectivity of the near-infrared band, and RED represents the reflectivity of the red light band.

[0049] The Enhanced Vegetation Index (EVI) includes EVI1 and EVI2.

[0050] in, , , NIR represents the reflectivity of the near-infrared band, RED represents the reflectivity of the red light band, and BLUE represents the reflectivity of the blue light band.

[0051] After extracting multiple vegetation index features corresponding to each observation point from the remote sensing image samples, the target vegetation index feature can be selected from the multiple vegetation index features based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point. The target vegetation index feature can be understood as the vegetation index feature with the highest correlation with the soil nitrogen mineralization rate, and can be better used to determine the soil nitrogen mineralization rate.

[0052] S104. Based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each observation point, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train a soil nitrogen mineralization rate prediction model, which is used to predict the soil nitrogen mineralization rate.

[0053] For example, the initial soil nitrogen mineralization rate prediction model can be a deep learning-based network model or a non-deep learning network model, and can be set according to actual needs.

[0054] Among them, the network model based on deep learning can be a Convolutional Neural Networks (CNN) model, or a Recurrent Neural Networks (RNN) model, etc., which can be set according to actual needs.

[0055] For example, if the target vegetation index characteristic is the normalized difference vegetation index NDVI, the model parameters of the initial soil nitrogen mineralization rate prediction model can be updated based on the normalized difference vegetation index NDVI and the soil nitrogen mineralization rate label corresponding to each observation point; if the target vegetation index characteristic is the green normalized vegetation index GNDVI, the model parameters of the initial soil nitrogen mineralization rate prediction model can be updated based on the green normalized vegetation index GNDVI and the soil nitrogen mineralization rate label corresponding to each observation point. The specific settings can be made according to actual needs.

[0056] It can be seen that in the embodiment of the present application, remote sensing image samples of monitoring area samples and soil samples of monitoring area samples are obtained; wherein the monitoring area samples include multiple observation points; and based on the soil samples of the monitoring area samples, the soil nitrogen mineralization rate label corresponding to each observation point is determined; the remote sensing image samples of the monitoring area samples are input into the initial soil nitrogen mineralization rate prediction model, and the initial soil nitrogen mineralization rate prediction model is used to extract multiple vegetation index features corresponding to each observation point from the remote sensing image samples, and based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, a target vegetation index feature is selected from the multiple vegetation index features; and then based on the target vegetation index feature corresponding to each observation point and the soil nitrogen mineralization rate label, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train and obtain a soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate. In this way, remote sensing technology is used to obtain remote sensing image samples of the monitored area, and the soil nitrogen mineralization rate prediction model is trained by combining the remote sensing image samples and the soil nitrogen mineralization rate labels corresponding to multiple observation points in the monitored area samples. This makes it possible to dynamically predict the soil nitrogen mineralization rate through the soil nitrogen mineralization rate prediction model, and achieves the goal of effectively improving not only the calculation efficiency of the soil nitrogen mineralization rate but also the calculation accuracy of the soil nitrogen mineralization rate without being restricted by the area of ​​the monitored area.

[0057] Based on the above Figure 1 In the embodiment shown, for example, in the above S103, based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, when selecting the target vegetation index feature from the multiple vegetation index features, at least two possible implementations may be included: In a possible implementation, the Pearson correlation coefficient method may be used to select a target vegetation index feature from a plurality of vegetation index features, specifically including: For each vegetation index feature, the correlation coefficient corresponding to the vegetation index feature can be determined based on the vegetation index feature corresponding to each observation point, the average value of the vegetation index features corresponding to all observation points, the soil nitrogen mineralization rate label corresponding to each observation point, and the average value of the soil nitrogen mineralization rate labels corresponding to all observation points; and based on the correlation coefficients corresponding to each vegetation index feature, the vegetation index feature corresponding to the maximum correlation coefficient is determined as the target vegetation index feature.

[0058] For example, in an embodiment of the present application, based on the vegetation index feature corresponding to each observation point, the average value of the vegetation index feature corresponding to all observation points, the soil nitrogen mineralization rate label corresponding to each observation point, and the average value of the soil nitrogen mineralization rate label corresponding to all observation points, the correlation coefficient corresponding to the vegetation index feature is determined, including: according to Determine the correlation coefficient corresponding to the vegetation index characteristics; in, represents the correlation coefficient corresponding to the vegetation index feature, i represents the i-th observation point, N represents the total number of observation points, represents the vegetation index characteristics corresponding to the i-th observation point, Represents the average value of vegetation index characteristics corresponding to all observation points, represents the soil nitrogen mineralization rate label corresponding to the i-th observation point, Represents the average value of the soil nitrogen mineralization rate labels corresponding to all observation points.

[0059] Taking the normalized difference vegetation index NDVI as an example, when determining the correlation coefficient corresponding to the normalized difference vegetation index NDVI, the normalized difference vegetation index NDVI corresponding to each observation point, the average value of the normalized difference vegetation index NDVI corresponding to all observation points, the soil nitrogen mineralization rate label corresponding to each observation point, and the average value of the soil nitrogen mineralization rate label corresponding to all observation points can be brought into the above correlation coefficient calculation formula to obtain the correlation coefficient corresponding to the normalized difference vegetation index NDVI. Similarly, other vegetation indices can be calculated through the above correlation coefficient calculation formula, for example, the green normalized vegetation index GNDVI, the ratio vegetation index RVI, the difference vegetation index DI, and the enhanced vegetation index EVI. Correlation coefficients corresponding to each of them.

[0060] After respectively determining the correlation coefficients corresponding to the multiple vegetation indices, the vegetation index feature corresponding to the maximum correlation coefficient can be determined as the target vegetation index feature, thereby determining the vegetation index feature with the highest correlation with the soil nitrogen mineralization rate. For example, if the vegetation index feature corresponding to the maximum correlation coefficient is the normalized vegetation index NDVI, the normalized vegetation index NDVI can be determined as the target vegetation index feature; if the vegetation index feature corresponding to the maximum correlation coefficient is the green normalized vegetation index GNDVI, the green normalized vegetation index GNDVI can be determined as the target vegetation index feature, and the specific setting can be made according to actual needs.

[0061] In another possible implementation, a principal component analysis (PCA) method may be used to select a target vegetation index feature from a plurality of vegetation index features, specifically including: Based on multiple vegetation index features corresponding to each observation point, the covariance matrix between the multiple vegetation index features is determined, and the covariance matrix is ​​used to characterize the correlation between the multiple vegetation index features; and based on the eigenvalues ​​and corresponding eigenvectors of the covariance matrix, multiple principal component vegetation index features are determined from the multiple vegetation index features; and based on the correlation between the multiple principal component vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, the target vegetation index feature is selected from the multiple principal component vegetation index features.

[0062] For example, when selecting a target vegetation index feature from a plurality of vegetation index features, the plurality of vegetation index features may be preprocessed first, including missing value processing, outlier detection and data standardization, to ensure the accuracy of PCA analysis.

[0063] For example, based on the correlation between multiple principal component vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, when selecting a target vegetation index feature from multiple principal component vegetation index features, the correlation between each principal component vegetation index feature and the soil nitrogen mineralization rate label corresponding to each observation point can be calculated respectively, and the principal component vegetation index feature corresponding to the maximum correlation is determined as the target vegetation index feature, thereby determining the vegetation index feature with the highest correlation with the soil nitrogen mineralization rate.

[0064] Combining the above two possible implementation methods, after selecting the vegetation index feature with the highest correlation with the soil nitrogen mineralization rate from multiple vegetation index features, the above S104 can be executed. Based on the target vegetation index feature and the soil nitrogen mineralization rate label corresponding to each observation point, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train a soil nitrogen mineralization rate prediction model for predicting the soil nitrogen mineralization rate, enabling the dynamic prediction of the soil nitrogen mineralization rate through the soil nitrogen mineralization rate prediction model. Without being restricted by the area of the monitored region, it not only effectively improves the calculation efficiency of the soil nitrogen mineralization rate but also improves the calculation accuracy of the soil nitrogen mineralization rate.

[0065] It can be understood that after training a soil nitrogen mineralization rate prediction model for predicting the soil nitrogen mineralization rate, the soil nitrogen mineralization rate prediction model can also be evaluated. For example, the evaluation indicators can include the correlation coefficient R2, the root mean squared error (RMSE), the rate of percentage difference (RPD), and the ratio of performance to inter - quartile (RPIQ), etc.

[0066] Generally, if RPD > 3.0, it indicates that the soil nitrogen mineralization rate prediction model has excellent prediction ability and can be used for quantitative analysis; if 2.5 < RPD ≤ 3.0, it indicates that the soil nitrogen mineralization rate prediction model has good prediction ability and is suitable for most quantitative analyses; if 2.0 < RPD ≤ 2.5, it indicates that the soil nitrogen mineralization rate prediction model has good prediction ability and needs further optimization; if 1.7 < RPD ≤ 2.0, it indicates that the soil nitrogen mineralization rate prediction model has average prediction ability and is only suitable for qualitative analysis or rough estimation; if RPD ≤ 1.7, it indicates that the soil nitrogen mineralization rate prediction model has poor prediction ability and is not recommended for quantitative analysis.

[0067] Generally, if RPIQ > 2.5, it indicates that the soil nitrogen mineralization rate prediction model has excellent prediction ability; if 2.5 > RPIQ > 2.0, it indicates that the soil nitrogen mineralization rate prediction model has very good prediction ability; if 2.0 > RPIQ > 1.7, it indicates that the soil nitrogen mineralization rate prediction model has good prediction ability; if 1.7 > RPIQ > 1.4, it indicates that the soil nitrogen mineralization rate prediction model has average prediction ability; if RPIQ < 1.4, it indicates that the soil nitrogen mineralization rate prediction model has poor prediction ability.

[0068] In an embodiment of the present application, in order to evaluate the trained soil nitrogen mineralization rate prediction model, four different types of soils, such as sandy soil, ordinary soil, silt soil and silt soil, can be used to verify the prediction results of the soil nitrogen mineralization rate prediction model, which can be combined with the following Table 1.

[0069] Table 1

[0070] Combined with Table 1, it can be seen that for sandy soil, the correlation coefficient R2 between the prediction results of the soil nitrogen mineralization rate prediction model and the soil nitrogen mineralization rate label is 0.81, the root mean square error RMSE is 0.04, the percentage difference ratio RPD is 3.43, and the performance and interquartile range ratio RPIQ is 3.22; for ordinary soil, the correlation coefficient R2 between the prediction results of the soil nitrogen mineralization rate prediction model and the soil nitrogen mineralization rate label is 0.79, the root mean square error RMSE is 0.05, the percentage difference ratio RPD is 2.16, and the performance and interquartile range ratio RPIQ is 2.83; for silt soil, the correlation coefficient R2 between the prediction results predicted by the soil nitrogen mineralization rate prediction model and the soil nitrogen mineralization rate label was 0.45, the root mean square error RMSE was 0.11, the percentage difference ratio RPD was 1.44, and the performance and interquartile range ratio RPIQ was 1.41; for silt soil, the correlation coefficient R2 between the prediction results predicted by the soil nitrogen mineralization rate prediction model and the soil nitrogen mineralization rate label was 0.62, the root mean square error RMSE was 0.09, the percentage difference ratio RPD was 1.71, and the performance and interquartile range ratio RPIQ was 2.10.

[0071] It can be seen from Table 1 above that for different types of soil, the soil nitrogen mineralization rate prediction model trained by the embodiment of the present application has excellent prediction ability, and can dynamically predict the soil nitrogen mineralization rate, without being restricted by the area of ​​the monitoring area. It can not only effectively improve the calculation efficiency of the soil nitrogen mineralization rate, but also improve the calculation accuracy of the soil nitrogen mineralization rate, which helps agricultural managers to dynamically understand the soil nutrient supply and take reasonable management measures.

[0072] Based on any of the above embodiments, after the soil nitrogen mineralization rate prediction model is trained, the soil nitrogen mineralization rate can also be predicted by the soil nitrogen mineralization rate prediction model. For example, see the following Figure 3 The embodiment shown.

[0073] Figure 3 A schematic diagram of a method for predicting soil nitrogen mineralization rate provided in an embodiment of the present application, for example, see Figure 3As shown, the method for predicting the soil nitrogen mineralization rate may include: S301, obtaining a remote sensing image of a target in a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points.

[0074] For example, the target remote sensing image may be a multispectral remote sensing image of different resolutions, or a hyperspectral remote sensing image, etc., which may be specifically set according to actual needs.

[0075] For example, when obtaining the target remote sensing image of the monitoring area to be predicted, the satellite image of the monitoring area to be predicted can be collected by satellite and used as the target remote sensing image; the satellite image of the monitoring area to be predicted can also be collected by drone and used as the target remote sensing image; of course, the satellite image and drone image of the monitoring area to be predicted can also be collected by satellite and drone respectively, and the collected satellite image and drone image can be fused to obtain the target remote sensing image of the monitoring area to be predicted. The specific settings can be made according to actual needs.

[0076] S302, inputting the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predicting the soil nitrogen mineralization rate corresponding to each target observation point through the soil nitrogen mineralization rate prediction model.

[0077] The soil nitrogen mineralization rate prediction model is a soil nitrogen mineralization rate prediction model obtained by training according to any one of the above embodiments.

[0078] In this way, the soil nitrogen mineralization rate prediction model can be used to dynamically predict the soil nitrogen mineralization rate corresponding to each target observation point in the predicted monitoring area, which not only effectively improves the calculation efficiency of the soil nitrogen mineralization rate, but also improves the calculation accuracy of the soil nitrogen mineralization rate without being restricted by the area of ​​the monitoring area.

[0079] For example, after the soil nitrogen mineralization rate corresponding to each target observation point in the monitoring area to be predicted is predicted by the soil nitrogen mineralization rate prediction model, the soil nitrogen mineralization rate corresponding to each target observation point in the monitoring area to be predicted can be further visualized to generate a soil nitrogen mineralization rate thematic monitoring map, and the soil nitrogen mineralization rate thematic monitoring map can be intuitively displayed to the user, thereby facilitating the refined fertilization management of agricultural production, improving fertilizer utilization, and reducing environmental pollution caused by fertilization.

[0080] It is understood that in the embodiments of the present application, the training of the soil nitrogen mineralization rate prediction model and the application of the soil nitrogen mineralization rate prediction model can be achieved by a soil nitrogen mineralization rate monitoring system. For example, see Figure 4 As shown, Figure 4A schematic diagram of a soil nitrogen mineralization rate monitoring system provided for an embodiment of the present application, the soil nitrogen mineralization rate monitoring system may include a data acquisition module, a data preprocessing module, a model building module and a prediction output module. Among them, the data acquisition module can be used to collect remote sensing image samples of monitoring area samples and soil samples of monitoring area samples; the data preprocessing module is used to collect remote sensing image samples of monitoring area samples and soil samples of monitoring area samples for preprocessing; the model building module is used to extract multiple vegetation index features corresponding to each observation point from the remote sensing image samples, and based on the multiple vegetation index features corresponding to each observation point and the soil nitrogen mineralization rate label corresponding to each observation point, select the target vegetation index feature from the multiple vegetation index features; based on the target vegetation index feature and soil nitrogen mineralization rate label corresponding to each observation point, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train and obtain the soil nitrogen mineralization rate prediction model; the prediction output module is used to predict the soil nitrogen mineralization rate corresponding to the observation point based on the soil nitrogen mineralization rate prediction model, and visualize the soil nitrogen mineralization rate corresponding to the observation point.

[0081] The training device of the soil nitrogen mineralization rate prediction model and the prediction device of the soil nitrogen mineralization rate provided in the present application are described below. The training device of the soil nitrogen mineralization rate prediction model described below and the training method of the soil nitrogen mineralization rate prediction model described above can be referenced to each other, and the prediction device of the soil nitrogen mineralization rate and the prediction method of the soil nitrogen mineralization rate described above can be referenced to each other.

[0082] Figure 5 A schematic diagram of a training device for a soil nitrogen mineralization rate prediction model provided in an embodiment of the present application, for example, see Figure 5 As shown, the training device 50 of the soil nitrogen mineralization rate prediction model may include: The acquisition unit 501 is used to acquire a remote sensing image sample of a monitoring area sample and a soil sample of the monitoring area sample; wherein the monitoring area sample includes a plurality of observation points; The first processing unit 502 is used to determine the soil nitrogen mineralization rate label corresponding to each of the observation points based on the soil samples of the monitoring area samples; The second processing unit 503 is used to input the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extract a plurality of vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and select a target vegetation index feature from the plurality of vegetation index features based on the plurality of vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; The updating unit 504 is used to update the model parameters of the initial soil nitrogen mineralization rate prediction model based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each of the observation points, so as to train a soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

[0083] For example, in the embodiment of the present application, the second processing unit 503 is used to select a target vegetation index feature from the multiple vegetation index features based on the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points, including: For each of the vegetation index characteristics, based on the vegetation index characteristics corresponding to each of the observation points, the average value of the vegetation index characteristics corresponding to all the observation points, the soil nitrogen mineralization rate label corresponding to each of the observation points, and the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points, determine the correlation coefficient corresponding to the vegetation index characteristics; Based on the correlation coefficients corresponding to the vegetation index features, the vegetation index feature corresponding to the maximum correlation coefficient is determined as the target vegetation index feature.

[0084] For example, in an embodiment of the present application, the second processing unit 503 is used to determine the correlation coefficient corresponding to the vegetation index feature based on the vegetation index feature corresponding to each of the observation points, the average value of the vegetation index feature corresponding to all the observation points, the soil nitrogen mineralization rate label corresponding to each of the observation points, and the average value of the soil nitrogen mineralization rate label corresponding to all the observation points, including: according to Determining a correlation coefficient corresponding to the vegetation index feature; in, represents the correlation coefficient corresponding to the vegetation index feature, i represents the i-th observation point, N represents the total number of observation points, represents the vegetation index feature corresponding to the i-th observation point, represents the average value of the vegetation index characteristics corresponding to all observation points, represents the soil nitrogen mineralization rate label corresponding to the i-th observation point, It represents the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points.

[0085] For example, in the embodiment of the present application, the second processing unit 503 is used to select a target vegetation index feature from the multiple vegetation index features based on the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points, including: Based on the multiple vegetation index features corresponding to each of the observation points, determining a covariance matrix between the multiple vegetation index features, wherein the covariance matrix is ​​used to characterize the correlation between the multiple vegetation index features; Determining a plurality of principal component vegetation index features from the plurality of vegetation index features based on the eigenvalues ​​and corresponding eigenvectors of the covariance matrix; Based on the correlation between the multiple principal component vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate labels corresponding to each of the observation points, the target vegetation index feature is selected from the multiple principal component vegetation index features.

[0086] The training device 50 for the soil nitrogen mineralization rate prediction model provided in the embodiment of the present application can execute the technical solution of the training method for the soil nitrogen mineralization rate prediction model in any of the above-mentioned embodiments. Its implementation principle and beneficial effects are similar to the implementation principle and beneficial effects of the training method for the soil nitrogen mineralization rate prediction model. Please refer to the implementation principle and beneficial effects of the training method for the soil nitrogen mineralization rate prediction model, which will not be repeated here.

[0087] Figure 6 A schematic diagram of a process flow of a soil nitrogen mineralization rate prediction device provided in an embodiment of the present application, for example, see Figure 6 As shown, the soil nitrogen mineralization rate prediction device 60 may include: The acquisition unit 601 is used to acquire a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; The prediction unit 602 is used to input the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predict the soil nitrogen mineralization rate corresponding to each target observation point through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is the soil nitrogen mineralization rate prediction model obtained by training as described in any of the above embodiments.

[0088] The soil nitrogen mineralization rate prediction device 60 provided in the embodiment of the present application can execute the technical solution of the soil nitrogen mineralization rate prediction method in any of the above-mentioned embodiments. Its implementation principle and beneficial effects are similar to the implementation principle and beneficial effects of the soil nitrogen mineralization rate prediction method. Please refer to the implementation principle and beneficial effects of the soil nitrogen mineralization rate prediction method, which will not be repeated here.

[0089] Figure 7 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application, such as Figure 7As shown, the electronic device may include: a processor (processor) 710 , a communication interface (Communications Interface) 720 , a memory (memory) 730 and a communication bus 740 , wherein the processor 710 , the communication interface 720 , and the memory 730 communicate with each other through the communication bus 740 . The processor 710 can call the logic instructions in the memory 730 to execute the training method of the soil nitrogen mineralization rate prediction model, which method includes: obtaining remote sensing image samples of the monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include multiple observation points; based on the soil samples of the monitoring area samples, determining the soil nitrogen mineralization rate label corresponding to each of the observation points; inputting the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extracting multiple vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and based on the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points, selecting a target vegetation index feature from the multiple vegetation index features; based on the target vegetation index feature corresponding to each of the observation points and the soil nitrogen mineralization rate label, updating the model parameters of the initial soil nitrogen mineralization rate prediction model to train and obtain a soil nitrogen mineralization rate prediction model, wherein the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

[0090] or, A method for predicting soil nitrogen mineralization rate, the method comprising: obtaining a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; inputting the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predicting the soil nitrogen mineralization rate corresponding to each of the target observation points through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is a soil nitrogen mineralization rate prediction model obtained by training according to any of the above embodiments.

[0091] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0092] On the other hand, the present application also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the training method of the soil nitrogen mineralization rate prediction model provided by the above methods, the method comprising: obtaining remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include multiple observation points; based on the soil samples of the monitoring area samples, determining the soil nitrogen mineralization rate label corresponding to each of the observation points; inputting the remote sensing image samples of the monitoring area samples into In the initial soil nitrogen mineralization rate prediction model, multiple vegetation index features corresponding to each of the observation points are extracted from the remote sensing image sample through the initial soil nitrogen mineralization rate prediction model, and based on the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate labels corresponding to each of the observation points, a target vegetation index feature is selected from the multiple vegetation index features; based on the target vegetation index features and the soil nitrogen mineralization rate labels corresponding to each of the observation points, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train and obtain a soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

[0093] or, A method for predicting soil nitrogen mineralization rate, the method comprising: obtaining a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; inputting the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predicting the soil nitrogen mineralization rate corresponding to each of the target observation points through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is a soil nitrogen mineralization rate prediction model obtained by training according to any of the above embodiments.

[0094] On the other hand, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a training method for executing the soil nitrogen mineralization rate prediction model provided by the above-mentioned methods, the method comprising: obtaining remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include a plurality of observation points; based on the soil samples of the monitoring area samples, determining the soil nitrogen mineralization rate label corresponding to each of the observation points; inputting the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model , extracting multiple vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and selecting a target vegetation index feature from the multiple vegetation index features based on the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; based on the target vegetation index feature and the soil nitrogen mineralization rate label corresponding to each of the observation points, updating the model parameters of the initial soil nitrogen mineralization rate prediction model to train and obtain a soil nitrogen mineralization rate prediction model, wherein the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

[0095] or, A method for predicting soil nitrogen mineralization rate, the method comprising: obtaining a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; inputting the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predicting the soil nitrogen mineralization rate corresponding to each of the target observation points through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is a soil nitrogen mineralization rate prediction model obtained by training according to any of the above embodiments.

[0096] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0097] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A training method for a soil nitrogen mineralization rate prediction model, characterized in that: include: Acquire remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include multiple observation points; Based on the soil samples of the monitoring area samples, determining the soil nitrogen mineralization rate label corresponding to each of the observation points; Inputting the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extracting a plurality of vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and selecting a target vegetation index feature from the plurality of vegetation index features based on the plurality of vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; Based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each of the observation points, the model parameters of the initial soil nitrogen mineralization rate prediction model are updated to train a soil nitrogen mineralization rate prediction model, which is used to predict the soil nitrogen mineralization rate.

2. The method for training a soil nitrogen mineralization rate prediction model according to claim 1, characterized in that: The selecting a target vegetation index feature from the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points comprises: For each of the vegetation index characteristics, based on the vegetation index characteristics corresponding to each of the observation points, the average value of the vegetation index characteristics corresponding to all the observation points, the soil nitrogen mineralization rate label corresponding to each of the observation points, and the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points, determine the correlation coefficient corresponding to the vegetation index characteristics; Based on the correlation coefficients corresponding to the vegetation index features, the vegetation index feature corresponding to the maximum correlation coefficient is determined as the target vegetation index feature.

3. The training method of the soil nitrogen mineralization rate prediction model according to claim 2, characterized in that: The determining of the correlation coefficient corresponding to the vegetation index feature based on the vegetation index feature corresponding to each of the observation points, the average value of the vegetation index feature corresponding to all the observation points, the soil nitrogen mineralization rate label corresponding to each of the observation points, and the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points includes: according to Determining a correlation coefficient corresponding to the vegetation index feature; in, represents the correlation coefficient corresponding to the vegetation index feature, i represents the i-th observation point, N represents the total number of observation points, represents the vegetation index feature corresponding to the i-th observation point, represents the average value of the vegetation index characteristics corresponding to all observation points, represents the soil nitrogen mineralization rate label corresponding to the i-th observation point, It represents the average value of the soil nitrogen mineralization rate labels corresponding to all the observation points.

4. The training method of the soil nitrogen mineralization rate prediction model according to claim 1, characterized in that: The selecting a target vegetation index feature from the multiple vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points comprises: Based on the multiple vegetation index features corresponding to each of the observation points, determining a covariance matrix between the multiple vegetation index features, wherein the covariance matrix is ​​used to characterize the correlation between the multiple vegetation index features; Determining a plurality of principal component vegetation index features from the plurality of vegetation index features based on the eigenvalues ​​and corresponding eigenvectors of the covariance matrix; Based on the correlation between the multiple principal component vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate labels corresponding to each of the observation points, the target vegetation index feature is selected from the multiple principal component vegetation index features.

5. A method for predicting soil nitrogen mineralization rate, characterized in that: include: Acquire a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; The target remote sensing image is input into a soil nitrogen mineralization rate prediction model, and the soil nitrogen mineralization rate corresponding to each target observation point is predicted by the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is the soil nitrogen mineralization rate prediction model obtained by training as described in any one of claims 1 to 4 above.

6. A training device for a soil nitrogen mineralization rate prediction model, characterized in that: include: An acquisition unit, used to acquire remote sensing image samples of monitoring area samples and soil samples of the monitoring area samples; wherein the monitoring area samples include a plurality of observation points; A first processing unit is used to determine the soil nitrogen mineralization rate label corresponding to each of the observation points based on the soil samples of the monitoring area samples; A second processing unit is used to input the remote sensing image samples of the monitoring area samples into the initial soil nitrogen mineralization rate prediction model, extract a plurality of vegetation index features corresponding to each of the observation points from the remote sensing image samples through the initial soil nitrogen mineralization rate prediction model, and select a target vegetation index feature from the plurality of vegetation index features based on the plurality of vegetation index features corresponding to each of the observation points and the soil nitrogen mineralization rate label corresponding to each of the observation points; An updating unit is used to update the model parameters of the initial soil nitrogen mineralization rate prediction model based on the target vegetation index characteristics and soil nitrogen mineralization rate labels corresponding to each of the observation points, so as to train a soil nitrogen mineralization rate prediction model, wherein the soil nitrogen mineralization rate prediction model is used to predict the soil nitrogen mineralization rate.

7. A device for predicting soil nitrogen mineralization rate, characterized in that: include: An acquisition unit, used for acquiring a target remote sensing image of a monitoring area to be predicted; wherein the monitoring area includes a plurality of target observation points; A prediction unit is used to input the target remote sensing image into a soil nitrogen mineralization rate prediction model, and predict the soil nitrogen mineralization rate corresponding to each of the target observation points through the soil nitrogen mineralization rate prediction model; wherein the soil nitrogen mineralization rate prediction model is the soil nitrogen mineralization rate prediction model obtained by training according to any one of claims 1 to 4 above.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the training method of the soil nitrogen mineralization rate prediction model as described in any one of claims 1 to 4, or implements the soil nitrogen mineralization rate prediction method as described in claim 5.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the training method of the soil nitrogen mineralization rate prediction model as described in any one of claims 1 to 4, or implements the soil nitrogen mineralization rate prediction method as described in claim 5.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, it implements the training method of the soil nitrogen mineralization rate prediction model as described in any one of claims 1 to 4, or implements the soil nitrogen mineralization rate prediction method as described in claim 5.

Images