Soil salt content dynamic inversion method and device, electronic equipment and storage medium

Abstract

The application provides a soil salt content dynamic inversion method and device, electronic equipment and storage medium, in the method, the direct and indirect causal relationship between the soil salt content, the environment drought and the crop growth is quantified by introducing a structural equation (SEM) model, which overcomes the limitation that the traditional method only pays attention to the correlation and cannot analyze the coupling effect, provides theoretical support for inversion, filters the input features of the CNN model based on the causal relationship between variables analyzed by the SEM model, realizes high-precision inversion of deep fusion of mechanism and data, significantly improves the soil salt content inversion precision, combines multi-temporal remote sensing observation data and spatial interpolation technology, realizes the leap from static evaluation to dynamic monitoring, and the application can accurately capture the spatiotemporal migration trajectory of the soil salt content and the dynamic response of the crop growth, and provides a reliable technical scheme for targeted regulation and scientific intervention of precision agriculture.

Description

Technical Field

[0001] This invention relates to the field of soil salinity monitoring technology, and in particular to a method, apparatus, electronic device, and storage medium for dynamic inversion of soil salinity. Background Technology

[0002] Soil salinization is a major constraint on sustainable agricultural development in arid and semi-arid regions worldwide, severely impacting crop growth, development, and yield. Cotton, a core economic crop in my country's arid regions, is highly sensitive to soil salinity, and the widespread presence of saline-alkali soils in these areas leads to significant yield losses, hindering the high-quality development of the cotton industry.

[0003] Traditional soil salinity monitoring relies on manual sampling and laboratory analysis, which is time-consuming, labor-intensive, has poor spatial representativeness, and cannot capture spatiotemporal dynamic changes. Existing remote sensing inversion methods mostly focus on bare soil salinity estimation, neglecting the coupling effect of soil-plant-environment during the crop growing season, and lack comprehensive quantitative characterization of cotton growth status. Furthermore, current technologies do not effectively integrate causal relationship analysis with high-precision inversion algorithms, making it difficult to achieve dynamic monitoring of saline-alkali soil and failing to provide targeted technical support for precision agriculture. Therefore, there is an urgent need to develop a soil salinity monitoring method that integrates multi-source data, quantifies coupling relationships, and achieves dynamic and accurate inversion.

[0004] In summary, traditional soil salinity monitoring suffers from technical problems such as low accuracy, lack of quantification of coupling relationships, and static nature. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method, apparatus, electronic device and storage medium for dynamic inversion of soil salinity, so as to alleviate the technical problems of low accuracy, lack of quantification of coupling relationship and static nature of traditional soil salinity monitoring.

[0006] In a first aspect, the present invention provides a method for dynamic inversion of soil salinity, comprising: Acquire multi-time series and multi-source measurement data of the target crop area, wherein the multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes: UAV remote sensing data, regional sampling data, and crop growth parameters; The comprehensive crop growth index is calculated based on the crop growth parameters, and the temperature and vegetation drought index is calculated based on the UAV remote sensing data. Based on the regional sampling data and the crop comprehensive growth index, the spatial distribution map of soil salinity and the spatial distribution map of crop growth are obtained by interpolation. The spatiotemporal migration trajectory of soil salinity is determined according to the spatial distribution map of soil salinity in each period, and the dynamic response of crop growth is determined according to the spatial distribution map of crop growth in each period. A structural equation model was constructed and solved to obtain the causal relationships between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, while soil salinity, temperature-vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. Based on the causal relationship between the variables, the observed variables that have a significant causal relationship with the soil salinity are selected from the observed variables as input features, and a convolutional neural network model is trained based on the input features to obtain the soil salinity inversion model; The soil salinity inversion model is used to invert the soil salinity of the target crop area to obtain the soil salinity inversion results of the target crop area.

[0007] Further, the crop growth parameters include: plant height, leaf chlorophyll content, and leaf area index. The comprehensive crop growth index is calculated based on these parameters, including: The plant height, leaf chlorophyll content, and leaf area index were normalized to obtain normalized plant height, normalized leaf chlorophyll content, and normalized leaf area index. The normalized plant height, the normalized leaf chlorophyll content, and the normalized leaf area index are weighted and summed to obtain the crop comprehensive growth index.

[0008] Furthermore, the UAV remote sensing data includes: multispectral data and thermal infrared data. The temperature-vegetation drought index is calculated based on the UAV remote sensing data, including: The multispectral data is preprocessed to obtain preprocessed multispectral data, and the normalized vegetation index is extracted from the preprocessed multispectral data. The thermal infrared data is preprocessed to obtain preprocessed thermal infrared data, and the surface temperature is extracted from the preprocessed thermal infrared data. The minimum surface temperature corresponding to a specific normalized vegetation index value and the maximum surface temperature corresponding to the specific normalized vegetation index value are determined based on the normalized vegetation index and the surface temperature. The temperature-vegetation drought index is calculated based on the lowest and highest surface temperatures.

[0009] Further, the regional sampling data includes the measured soil salinity content at multiple sampling points within the target crop area. Based on the regional sampling data and the crop comprehensive growth index, a spatial distribution map of soil salinity and a spatial distribution map of crop growth are obtained through interpolation, including: Geostatistical analysis was performed on the soil salinity content and the crop comprehensive growth index to obtain interpolation parameters; Based on the interpolation parameters, the Kriging interpolation algorithm is used to spatially interpolate the soil salinity content and the crop comprehensive growth index to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth.

[0010] Furthermore, the input features include: primary features and secondary features. The primary features include: the temperature-vegetation drought index and the crop comprehensive growth index. The secondary features include: the land surface temperature, the normalized vegetation index, the plant height, the leaf area index, and the leaf chlorophyll content. The convolutional neural network model includes: convolutional layers, pooling layers, and fully connected layers.

[0011] Furthermore, training is performed using a convolutional neural network model based on the input features, including: The convolutional neural network model is trained based on the first-level features and the second-level features respectively, to obtain the model trained by the first-level features and the model trained by the second-level features; The soil salinity inversion model is determined based on the accuracy of the model trained using the first-level features and the accuracy of the model trained using the second-level features.

[0012] Furthermore, the method also includes: Based on the spatial distribution map of soil salinity and the spatial distribution map of crop growth, high-salinity accumulation areas and dynamic change hotspots are identified, and a dynamic monitoring report of saline-alkali soil is generated.

[0013] Secondly, the present invention also provides a soil salinity dynamic inversion device, comprising: The acquisition unit is used to acquire multi-time series and multi-source measurement data of the target crop area, wherein the multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes: UAV remote sensing data, regional sampling data, and crop growth parameters; The calculation unit is used to calculate the comprehensive crop growth index based on the crop growth parameters and to calculate the temperature and vegetation drought index based on the UAV remote sensing data. The interpolation and determination unit is used to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth based on the regional sampling data and the comprehensive crop growth index through interpolation methods, and to determine the spatiotemporal migration trajectory of soil salinity according to the spatial distribution map of soil salinity in each period, and to determine the dynamic response of crop growth according to the spatial distribution map of crop growth in each period. A construction and solution unit is used to construct a structural equation model and solve the structural equation model to obtain the causal relationship between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, and soil salinity, temperature-vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. The screening and training unit is used to screen out the observed variables that have a significant causal relationship with the soil salinity from the observed variables according to the causal relationship between the variables, and to train the soil salinity inversion model based on the input features using a convolutional neural network model. The soil salinity inversion unit is used to perform soil salinity inversion on the target crop area using the soil salinity inversion model, and obtain the soil salinity inversion results of the target crop area.

[0014] Thirdly, the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method described in the first aspect.

[0015] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method described in the first aspect.

[0016] This invention provides a method for dynamic inversion of soil salinity, comprising: acquiring multi-temporal, multi-source measurement data of a target crop area, wherein the multi-temporal data includes key time points of rapid crop growth, and the multi-source measurement data includes: UAV remote sensing data, regional sampling data, and crop growth parameters; calculating a comprehensive crop growth index based on crop growth parameters, and calculating a temperature-vegetation-drought index based on UAV remote sensing data; obtaining a spatial distribution map of soil salinity and a spatial distribution map of crop growth based on regional sampling data and the comprehensive crop growth index using interpolation methods, determining the spatiotemporal migration trajectory of soil salinity based on the spatial distribution maps of soil salinity at each time period, and determining the dynamic response of crop growth based on the spatial distribution maps of crop growth at each time period; and constructing a structure. The structural equation model was developed and solved to obtain the causal relationships between variables. In the structural equation model, soil salinity, environmental drought, and crop growth were latent variables, while soil salinity content, temperature, vegetation drought index, crop comprehensive growth index, surface temperature, normalized difference vegetation index, plant height, leaf chlorophyll content, and leaf area index were observed variables. Based on the causal relationships between variables, observed variables with significant causal relationships with soil salinity were selected as input features. A convolutional neural network model was then trained based on these input features to obtain a soil salinity inversion model. The soil salinity inversion model was then used to invert soil salinity in the target crop area to obtain the soil salinity inversion results for the target crop area. As described above, the soil salinity dynamic inversion method of this invention quantifies the direct and indirect causal relationships between soil salinity, environmental drought, and crop growth by introducing a structural equation model. This overcomes the limitations of traditional methods that only focus on correlations and cannot analyze coupling effects, providing theoretical support for inversion. Based on the causal relationships between variables analyzed by the SEM model, the input features of the CNN model are selected, achieving high-precision inversion with deep integration of mechanism and data, significantly improving the accuracy of soil salinity inversion. Combining multi-time series remote sensing observation data and spatial interpolation technology, this invention achieves a leap from static assessment to dynamic monitoring. This invention can accurately capture the spatiotemporal migration trajectory of soil salinity and the dynamic response of crop growth, providing a reliable technical solution for targeted regulation and scientific intervention in precision agriculture, and alleviating the technical problems of low accuracy, lack of quantification of coupling relationships, and static nature of traditional soil salinity monitoring. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1A flowchart of a method for dynamically inverting soil salinity provided in an embodiment of the present invention; Figure 2 A schematic diagram of the process for dynamic inversion of soil salinity provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the spatial distribution of soil salinity and cotton growth provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the plant-soil water-salt coupling framework structure provided in an embodiment of the present invention; Figure 5 A schematic diagram of a soil salinity dynamic inversion device provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Traditional soil salinity monitoring methods suffer from low accuracy, lack of quantification of coupling relationships, and static nature.

[0021] Based on this, the soil salinity dynamic inversion method of this invention introduces a structural equation model to quantify the direct and indirect causal relationships between soil salinity, environmental drought, and crop growth. This overcomes the limitations of traditional methods that only focus on correlations and cannot analyze coupling effects, providing theoretical support for inversion. Based on the causal relationships between variables analyzed by the SEM model, the input features of the CNN model are selected, achieving high-precision inversion with deep integration of mechanism and data, significantly improving the accuracy of soil salinity inversion. Combined with multi-time series remote sensing observation data and spatial interpolation technology, it realizes the leap from static assessment to dynamic monitoring, accurately capturing the spatiotemporal migration trajectory of soil salinity and the dynamic response of crop growth, providing a reliable technical solution for targeted regulation and scientific intervention in precision agriculture.

[0022] To facilitate understanding of this embodiment, a method for dynamic inversion of soil salinity disclosed in this embodiment of the invention will first be described in detail.

[0023] Example 1: According to an embodiment of the present invention, an embodiment of a method for dynamic inversion of soil salinity is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] Figure 1 This is a flowchart of a method for dynamically inverting soil salinity according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps: Step S102: Obtain multi-time series and multi-source measurement data of the target crop area. The multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes UAV remote sensing data, regional sampling data, and crop growth parameters. In this embodiment of the invention, cotton is used as an example for illustration. Of course, other crops can also be used. This embodiment of the invention does not impose specific limitations on the above-mentioned crops.

[0025] When the crop is cotton, the above multi-time series includes five key time points for rapid cotton growth, the above regional sampling data specifically refers to cotton field sampling data, the above crop growth parameters specifically refer to cotton growth parameters, and the above target crop region can specifically refer to a cotton-growing saline-alkali area in an arid region of my country.

[0026] Drone remote sensing data: From June 19 to July 17, 2024, multispectral images were acquired using the DJI Mavic 3M multispectral sensor every week, and thermal infrared images were acquired using the DJI Matrice 350 RTK equipped with the H20T thermal imaging sensor. Data was collected at 2 pm every day, for a total of 5 periods. Sampling data from cotton fields: 23 uniform sampling points were set up in a 22m×40m experimental field. The coordinates were recorded by handheld GPS. Soil samples were collected at a depth of 0-20cm at each period. The extract was prepared at a water-to-soil ratio of 5:1. The soil salinity was determined by evaporation, drying, and weighing. The determination range was 6-13g / kg. Cotton growth parameters: Plant height (pH, distance from cotyledon node to plant top, reflecting cotton structural information) was measured manually at each stage. Chlorophyll was extracted with 95% ethanol and the leaf chlorophyll content (LCC, reflecting cotton physiological state) was measured using a spectrophotometer. Leaf area index (LAI, reflecting cotton light energy utilization efficiency) was measured using a LAI 2200C plant canopy analyzer. The measured values ​​of each parameter were recorded.

[0027] Step S104: Calculate the comprehensive crop growth index based on crop growth parameters, and calculate the temperature and vegetation drought index based on UAV remote sensing data. Specifically, the aforementioned temperature and vegetation drought index is used to reflect soil dryness, and the specific process is as follows: Figure 2 As shown below, the process of calculating the comprehensive crop growth index and the temperature-vegetation-drought index will be described in detail below.

[0028] Step S106: Based on regional sampling data and crop comprehensive growth index, soil salinity spatial distribution map and crop growth spatial distribution map are obtained by interpolation method. Soil salinity spatiotemporal migration trajectory is determined according to soil salinity spatial distribution map of each period, and crop growth dynamic response is determined according to crop growth spatial distribution map of each period. Specifically, such as Figure 3 The diagram shows the spatial distribution of soil salinity and cotton growth.

[0029] Step S108: Construct a structural equation model and solve the structural equation model to obtain the causal relationship between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, while soil salinity content, temperature vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. Specifically, based on structural equation modeling (SEM), a plant-soil water-salt coupling framework is constructed (e.g., Figure 4 As shown in the figure, a schematic diagram of the plant-soil water-salt coupling framework is presented. The causal relationship between soil salinity, environmental drought and cotton growth is analyzed through structural equation modeling.

[0030] In the structural equation model, soil salinity, environmental drought, and crop growth are latent variables, while soil salinity, temperature-vegetation drought index, crop composite growth index, surface temperature, normalized difference vegetation index, plant height, leaf chlorophyll content, and leaf area index are observed variables. A SEM model (structural equation model) is constructed using the lavaan package in R. Model fitting and validation: The model fit is validated using Chisq / df, GFI, CFI, SRMR, and RMSEA metrics, and the model meets the fitting validation standards.

[0031] Solving the model (using R language to solve for model parameters), the standardized path coefficients were calculated (a key indicator in structural equation modeling (SEM) used to quantify the strength and direction of causal relationships between variables): the direct negative effect of soil salinity on cotton growth was std. all = -1.23, the direct negative effect of environmental drought on cotton growth was std. all = -0.41, and the indirect positive effect of environmental drought on soil salinity was std. all = 0.14. It is clear that the direct negative effect of soil salinity on cotton growth is stronger than that of environmental drought, and the indirect positive effect of environmental drought on soil salinity is stronger.

[0032] Step S110: Based on the causal relationship between variables, select the observed variables that have a significant causal relationship with soil salinity from the observed variables as input features, and train the convolutional neural network model based on the input features to obtain the soil salinity inversion model. Step S112: Use the soil salinity inversion model to perform soil salinity inversion on the target crop area and obtain the soil salinity inversion results for the target crop area.

[0033] Specifically, by inputting the input features corresponding to the soil salinity inversion model for the target crop area into the soil salinity inversion model, the soil salinity inversion results can be obtained. By inputting the corresponding input features for multiple periods into the soil salinity inversion model, the soil salinity inversion results for the target crop area at each period can be obtained, thereby realizing dynamic monitoring of saline-alkali soil.

[0034] This invention provides a method for dynamic inversion of soil salinity. By introducing a structural equation model, it quantifies the direct and indirect causal relationships between soil salinity, environmental drought, and crop growth. Based on the causal relationships between variables analyzed by the SEM model, it selects input features for the CNN model, achieving high-precision inversion through deep fusion of mechanism and data. This significantly improves the accuracy of soil salinity inversion. Combined with multi-time-series remote sensing observation data and spatial interpolation technology, it achieves a leap from static assessment to dynamic monitoring. This invention can accurately capture the spatiotemporal migration trajectory of soil salinity and the dynamic response of crop growth, providing a reliable technical solution for targeted regulation and scientific intervention in precision agriculture. It alleviates the technical problems of low accuracy, lack of quantification of coupling relationships, and static nature of traditional soil salinity monitoring.

[0035] The above provides a brief overview of the soil salinity dynamic inversion method of the present invention. The specific details involved are described in detail below.

[0036] In an optional embodiment of the present invention, crop growth parameters include: plant height, leaf chlorophyll content, and leaf area index. The calculation of the comprehensive crop growth index based on these parameters specifically includes the following steps: (1) The plant height, leaf chlorophyll content and leaf area index were normalized to obtain the normalized plant height, normalized leaf chlorophyll content and normalized leaf area index. Specifically, the plant height, leaf chlorophyll content, and leaf area index at each stage were normalized. The normalization formula is as follows:

[0037] Where i represents the category of the calculated index (plant height, leaf chlorophyll content, and leaf area index), and P i X represents the standardized indicator. i X represents the specific value of each indicator.min X represents the minimum value of this indicator during the same period. max This indicates the maximum value of the indicator during the same period.

[0038] (2) The normalized plant height, normalized leaf chlorophyll content and normalized leaf area index are weighted and summed to obtain the crop comprehensive growth index.

[0039] Specifically, considering that the three parameters are equally important in representing crop growth, an equal-weighted summation method was used to sum the normalized plant height, normalized leaf chlorophyll content, and normalized leaf area index to obtain the comprehensive crop growth index.

[0040] The following formula is used:

[0041] Where CCGI represents the comprehensive growth index of crops (such as cotton), i represents the category of the calculated index (plant height, leaf chlorophyll content, and leaf area index), and P... i The variable n represents a standardized indicator. In this invention, the variable n is set to 3, corresponding to three different indicators.

[0042] In an optional embodiment of the present invention, the UAV remote sensing data includes: multispectral data and thermal infrared data. The calculation of the temperature-vegetation drought index based on the UAV remote sensing data specifically includes the following steps: (1) Preprocess the multispectral data to obtain preprocessed multispectral data, and extract the normalized vegetation index from the preprocessed multispectral data. Specifically, the above preprocessing includes: reflectance calibration, image registration, and digital orthophoto generation. The specific process is as follows: reflectance calibration, image registration, camera optimization, and digital orthophoto generation are completed using Agisoft Metashape v2.1.0 software, from which the normalized vegetation index is extracted.

[0043] (2) Preprocess the thermal infrared data to obtain preprocessed thermal infrared data, and extract the surface temperature from the preprocessed thermal infrared data; Specifically, the raw thermal infrared images (i.e., thermal infrared data) are processed in batches using the DJI Thermal SDK Module, converted into single-band temperature images, and then the surface temperature is obtained after registration and correction.

[0044] (3) Determine the minimum surface temperature and the maximum surface temperature corresponding to a specific normalized vegetation index value based on the normalized vegetation index and the surface temperature. Specifically, an LST (land surface temperature)-NDVI (normalized vegetation index) feature space is constructed based on land surface temperature and normalized vegetation index, and dry and wet edge equations are fitted: NDVI NDVI Where LSTmin is the lowest surface temperature in the image corresponding to a specific normalized NDVI value, and a1 and b1 represent the coefficients of the TVDI (Temperature Vegetation Aridity Index) wet edge equation. LST max α is the highest surface temperature in the image corresponding to the same normalized NDVI value, while a2 and b2 represent the coefficients of the TVDI (Temperature Vegetation Drought Index) dry edge equation.

[0045] For example, the dry-side equation on June 19th is: LST max =-7.21*NDVI+41.16, the wet edge equation is LST min =-0.69*NDVI+22.17.

[0046] (4) Calculate the temperature vegetation drought index based on the lowest and highest surface temperatures.

[0047] Specifically, the calculation formula is as follows: TVDI =

[0048] In an optional embodiment of the present invention, the regional sampling data includes the soil salinity content of multiple sampling points within the target crop area as determined by measurement. Based on the regional sampling data and the comprehensive crop growth index, a spatial distribution map of soil salinity and a spatial distribution map of crop growth are obtained through interpolation methods, specifically including the following steps: (1) Geostatistical analysis was performed on soil salinity content and crop comprehensive growth index to obtain interpolation parameters; Specifically, the geostatistical analysis process involves: conducting normality tests and spatial correlation analyses on soil salinity content and crop (e.g., cotton) comprehensive growth index data; using a spherical variogram model to fit the experimental variogram function; and determining the interpolation parameters.

[0049] (2) Based on the interpolation parameters, the Kriging interpolation algorithm is used to perform spatial interpolation on the soil salinity content and the comprehensive crop growth index to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth.

[0050] Specifically, the Kriging interpolation algorithm is used to spatially interpolate discrete soil salinity sampling data and cotton comprehensive growth index data to generate high-resolution spatial distribution maps of soil salinity and cotton growth, clearly presenting the spatial pattern of soil salinity distribution and achieving optimal spatial distribution estimation.

[0051] In an optional embodiment of the present invention, the input features include: primary features and secondary features. The primary features include: temperature-vegetation-drought index and crop comprehensive growth index. The secondary features include: surface temperature, normalized vegetation index, temperature-vegetation-drought index, plant height, leaf area index, and leaf chlorophyll content. Convolutional neural network models include: convolutional layers, pooling layers, and fully connected layers.

[0052] Specifically, the input features include primary and secondary features, with soil salinity content as the output variable, divided into training and test sets in a 7:3 ratio. The convolutional neural network model includes convolutional layers, pooling layers, and fully connected layers. The convolutional layers use a 5×1 filter with ReLU activation, the pooling layers use a 2×1 max-pooling filter with a stride of 2, and the fully connected layers have an output dimension of 32. K-fold cross-validation (k=5) is used to optimize the CNN model, and the model is trained using measured salinity content as the output label.

[0053] In an optional embodiment of the present invention, training is performed using a convolutional neural network model based on input features, including: (1) Train convolutional neural network models based on first-level features and second-level features respectively to obtain models trained by first-level features and models trained by second-level features; (2) Determine the soil salinity inversion model based on the accuracy of the model trained by the first-level features and the accuracy of the model trained by the second-level features.

[0054] Specifically, by comparing the accuracy of the output models of the first-level features and the second-level features, a soil salinity inversion model is obtained.

[0055] For example, the model accuracy obtained using first-level features is R²=0.95, RMSE=0.41, and MAE=0.31, while the model accuracy obtained using second-level features is R²=0.97, RMSE=0.35, and MAE=0.24; the soil salinity inversion model based on second-level features is the optimal model.

[0056] Dynamic analysis based on data from five periods shows that the overall soil salinity in the region is decreasing, and high-salinity areas (≥10g / kg) are gradually migrating from the central and southern regions to the south. Combined with the dynamic changes in CCGI, targeted irrigation programs (increasing irrigation frequency in the high-salinity areas in the south) and soil improvement measures can be developed for the region, which can effectively improve the growth status of cotton.

[0057] In an optional embodiment of the present invention, the method further includes the following steps: Based on the spatial distribution maps of soil salinity and crop growth, high-salinity accumulation areas and dynamic change hotspots are identified, and a dynamic monitoring report of saline-alkali soil is generated.

[0058] Specifically, by analyzing spatiotemporal changes, high-salinity clusters and dynamic hotspots are identified, generating dynamic monitoring reports on saline-alkali soil to provide a basis for precision agricultural interventions. In other words, identifying high-salinity clusters and dynamic hotspots and generating dynamic monitoring reports on saline-alkali soil provides targeted support for agricultural interventions such as precision irrigation and soil improvement.

[0059] The method of this invention solves the problems of low accuracy, static nature, and lack of quantification of coupling relationships in existing soil salinity monitoring technologies, and realizes accurate dynamic inversion of soil salinity, providing technical support for precision agriculture.

[0060] This study innovatively constructs a plant-soil water-salt coupling framework, integrating structural equation modeling causal analysis with high-precision CNN inversion. For the first time, it quantifies the coupling relationship between soil salinity, environmental drought, and cotton growth, providing mechanistic support for salinity inversion. Based on pH (plant height), LCC (leaf chlorophyll content), and LAI (leaf area index), a CCGI (cotton comprehensive growth index) is constructed to comprehensively characterize the structural, physiological, and light energy utilization characteristics of cotton, solving the problem of incomplete characterization by a single growth index. Combining Kriging interpolation and multi-period data, it achieves optimal estimation and dynamic monitoring of soil salinity spatial distribution, overcoming the limitations of traditional static assessment. The CNN model achieves an inversion accuracy of R²=0.97, superior to traditional algorithms, and requires no extensive manual sampling, reducing monitoring costs and providing an efficient and accurate technical solution for precision agriculture.

[0061] Example 2: This invention also provides a soil salinity dynamic inversion device, which is mainly used to execute the soil salinity dynamic inversion method provided in Embodiment 1 of this invention. The following is a detailed description of the soil salinity dynamic inversion device provided in this invention.

[0062] Figure 5 This is a schematic diagram of a soil salinity dynamic inversion device according to an embodiment of the present invention, as shown below. Figure 5 As shown, the device mainly includes: an acquisition unit 10, a calculation unit 20, an interpolation and determination unit 30, a construction and solution unit 40, a screening and training unit 50, and a soil salinity inversion unit 60, wherein: The acquisition unit is used to acquire multi-time series and multi-source measurement data of the target crop area. The multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes UAV remote sensing data, regional sampling data, and crop growth parameters. The calculation unit is used to calculate the comprehensive crop growth index based on crop growth parameters and the temperature and vegetation drought index based on UAV remote sensing data. The interpolation and determination unit is used to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth based on regional sampling data and crop comprehensive growth index through interpolation methods, and to determine the spatiotemporal migration trajectory of soil salinity based on the spatial distribution map of soil salinity at each period, and to determine the dynamic response of crop growth based on the spatial distribution map of crop growth at each period. The construction and solution unit is used to construct the structural equation model and solve the structural equation model to obtain the causal relationship between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, while soil salinity, temperature and vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. The screening and training unit is used to select observed variables that have a significant causal relationship with soil salinity from the observed variables as input features based on the causal relationship between variables, and to train a convolutional neural network model based on the input features to obtain a soil salinity inversion model. The soil salinity inversion unit is used to invert soil salinity in a target crop area using a soil salinity inversion model, and obtain the soil salinity inversion results for the target crop area.

[0063] In the soil salinity dynamic inversion device of this invention, a structural equation model is introduced to quantify the direct and indirect causal relationships between soil salinity, environmental drought, and crop growth. This overcomes the limitations of traditional methods that only focus on correlations and cannot analyze coupling effects, providing theoretical support for inversion. Based on the causal relationships between variables analyzed by the structural equation model, the input features of the CNN model are selected, achieving high-precision inversion with deep integration of mechanism and data, significantly improving the accuracy of soil salinity inversion. Combined with multi-time series remote sensing observation data and spatial interpolation technology, a leap from static assessment to dynamic monitoring is achieved. This invention can accurately capture the spatiotemporal migration trajectory of soil salinity and the dynamic response of crop growth, providing a reliable technical solution for targeted regulation and scientific intervention in precision agriculture, and alleviating the technical problems of low accuracy, lack of quantification of coupling relationships, and static nature of traditional soil salinity monitoring.

[0064] Optionally, crop growth parameters include: plant height, leaf chlorophyll content, and leaf area index. The calculation unit is also used to: normalize the plant height, leaf chlorophyll content, and leaf area index to obtain normalized plant height, normalized leaf chlorophyll content, and normalized leaf area index; and perform weighted summation of the normalized plant height, normalized leaf chlorophyll content, and normalized leaf area index to obtain the comprehensive crop growth index.

[0065] Optionally, the UAV remote sensing data includes multispectral data and thermal infrared data. The computing unit is also used to: preprocess the multispectral data to obtain preprocessed multispectral data, and extract the normalized vegetation index (NVI) from the preprocessed multispectral data; preprocess the thermal infrared data to obtain preprocessed thermal infrared data, and extract the land surface temperature from the preprocessed thermal infrared data; determine the minimum surface temperature corresponding to a specific NVI value and the maximum surface temperature corresponding to a specific NVI value based on the NVI and the surface temperature; and calculate the temperature-vegetation drought index based on the minimum and maximum surface temperatures.

[0066] Optionally, the regional sampling data includes the soil salinity content of multiple sampling points within the target crop area as determined. The interpolation and determination unit is also used to: perform geostatistical analysis on the soil salinity content and the crop comprehensive growth index to obtain interpolation parameters; and use the Kriging interpolation algorithm to spatially interpolate the soil salinity content and the crop comprehensive growth index based on the interpolation parameters to obtain a spatial distribution map of soil salinity and a spatial distribution map of crop growth.

[0067] Optionally, the input features include: primary features and secondary features. Primary features include: temperature, vegetation drought index and crop comprehensive growth index. Secondary features include: surface temperature, normalized vegetation index, plant height, leaf area index and leaf chlorophyll content. The convolutional neural network model includes: convolutional layer, pooling layer and fully connected layer.

[0068] Optionally, the screening and training unit is also used to: train convolutional neural network models based on first-level features and second-level features respectively to obtain models trained by first-level features and models trained by second-level features; and determine the soil salinity inversion model from the models trained by first-level features and models trained by second-level features based on the accuracy of the models trained by first-level features and the accuracy of the models trained by second-level features.

[0069] Optionally, the device is also used to: identify high-salt accumulation areas and dynamic change hotspots based on soil salinity spatial distribution maps and crop growth spatial distribution maps, and generate dynamic monitoring reports on saline-alkali soil.

[0070] The device provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0071] like Figure 6As shown in the embodiment of this application, an electronic device 600 includes a processor 601, a memory 602, and a bus. The memory 602 stores machine-readable instructions that can be executed by the processor 601. When the electronic device is running, the processor 601 communicates with the memory 602 via the bus. The processor 601 executes the machine-readable instructions to perform the steps of the above-described dynamic soil salinity inversion method.

[0072] Specifically, the memory 602 and processor 601 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 601 runs the computer program stored in the memory 602, it can execute the above-mentioned dynamic inversion method of soil salinity.

[0073] The processor 601 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 601 or by instructions in software form. The processor 601 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 602, and processor 601 reads the information from memory 602 and, in conjunction with its hardware, completes the steps of the above method.

[0074] Corresponding to the above-described method for dynamic inversion of soil salinity, this application also provides a computer-readable storage medium storing machine-executable instructions. When these machine-executable instructions are invoked and executed by a processor, they cause the processor to perform the steps of the above-described method for dynamic inversion of soil salinity.

[0075] The soil salinity dynamic inversion device provided in this application embodiment can be specific hardware on a device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this application embodiment are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0076] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0077] For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0080] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the soil salinity dynamic inversion method described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0082] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for dynamic inversion of soil salinity, characterized in that, include: Acquire multi-time series and multi-source measurement data of the target crop area, wherein the multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes: UAV remote sensing data, regional sampling data, and crop growth parameters; The comprehensive crop growth index is calculated based on the crop growth parameters, and the temperature and vegetation drought index is calculated based on the UAV remote sensing data. Based on the regional sampling data and the crop comprehensive growth index, the spatial distribution map of soil salinity and the spatial distribution map of crop growth are obtained by interpolation. The spatiotemporal migration trajectory of soil salinity is determined according to the spatial distribution map of soil salinity in each period, and the dynamic response of crop growth is determined according to the spatial distribution map of crop growth in each period. A structural equation model was constructed and solved to obtain the causal relationships between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, while soil salinity, temperature-vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. Based on the causal relationship between the variables, the observed variables that have a significant causal relationship with the soil salinity are selected from the observed variables as input features, and a convolutional neural network model is trained based on the input features to obtain the soil salinity inversion model; The soil salinity inversion model is used to invert the soil salinity of the target crop area to obtain the soil salinity inversion results of the target crop area.

2. The method according to claim 1, characterized in that, The crop growth parameters include: plant height, leaf chlorophyll content, and leaf area index. A comprehensive crop growth index is calculated based on these parameters, including: The plant height, leaf chlorophyll content, and leaf area index were normalized to obtain normalized plant height, normalized leaf chlorophyll content, and normalized leaf area index. The normalized plant height, the normalized leaf chlorophyll content, and the normalized leaf area index are weighted and summed to obtain the crop comprehensive growth index.

3. The method according to claim 1, characterized in that, The UAV remote sensing data includes: multispectral data and thermal infrared data. The temperature and vegetation drought index is calculated based on the UAV remote sensing data, including: The multispectral data is preprocessed to obtain preprocessed multispectral data, and the normalized vegetation index is extracted from the preprocessed multispectral data. The thermal infrared data is preprocessed to obtain preprocessed thermal infrared data, and the surface temperature is extracted from the preprocessed thermal infrared data. The minimum surface temperature corresponding to a specific normalized vegetation index value and the maximum surface temperature corresponding to the specific normalized vegetation index value are determined based on the normalized vegetation index and the surface temperature. The temperature-vegetation drought index is calculated based on the lowest and highest surface temperatures.

4. The method according to claim 1, characterized in that, The regional sampling data includes the soil salinity content measured at multiple sampling points within the target crop area. Based on the regional sampling data and the crop comprehensive growth index, a spatial distribution map of soil salinity and a spatial distribution map of crop growth are obtained through interpolation, including: Geostatistical analysis was performed on the soil salinity content and the crop comprehensive growth index to obtain interpolation parameters; Based on the interpolation parameters, the Kriging interpolation algorithm is used to spatially interpolate the soil salinity content and the crop comprehensive growth index to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth.

5. The method according to claim 1, characterized in that, The input features include: primary features and secondary features. The primary features include: the temperature-vegetation drought index and the crop comprehensive growth index. The secondary features include: the land surface temperature, the normalized vegetation index, the plant height, the leaf area index, and the leaf chlorophyll content. The convolutional neural network model includes: convolutional layers, pooling layers, and fully connected layers.

6. The method according to claim 5, characterized in that, Training a convolutional neural network model based on the input features includes: The convolutional neural network model is trained based on the first-level features and the second-level features respectively, to obtain the model trained by the first-level features and the model trained by the second-level features; The soil salinity inversion model is determined based on the accuracy of the model trained using the first-level features and the accuracy of the model trained using the second-level features.

7. The method according to claim 1, characterized in that, The method further includes: Based on the spatial distribution map of soil salinity and the spatial distribution map of crop growth, high-salinity accumulation areas and dynamic change hotspots are identified, and a dynamic monitoring report of saline-alkali soil is generated.

8. A soil salinity dynamic inversion device, characterized in that, include: The acquisition unit is used to acquire multi-time series and multi-source measurement data of the target crop area, wherein the multi-time series includes key time points of rapid crop growth, and the multi-source measurement data includes: UAV remote sensing data, regional sampling data, and crop growth parameters; The calculation unit is used to calculate the comprehensive crop growth index based on the crop growth parameters and to calculate the temperature and vegetation drought index based on the UAV remote sensing data. The interpolation and determination unit is used to obtain the spatial distribution map of soil salinity and the spatial distribution map of crop growth based on the regional sampling data and the comprehensive crop growth index through interpolation methods, and to determine the spatiotemporal migration trajectory of soil salinity according to the spatial distribution map of soil salinity in each period, and to determine the dynamic response of crop growth according to the spatial distribution map of crop growth in each period. A construction and solution unit is used to construct a structural equation model and solve the structural equation model to obtain the causal relationship between variables. In the structural equation model, soil salinity, environmental drought and crop growth are latent variables, and soil salinity, temperature-vegetation drought index, crop comprehensive growth index, surface temperature, normalized vegetation index, plant height, leaf chlorophyll content and leaf area index are observed variables. The screening and training unit is used to screen out the observed variables that have a significant causal relationship with the soil salinity from the observed variables according to the causal relationship between the variables, and to train the soil salinity inversion model based on the input features using a convolutional neural network model. The soil salinity inversion unit is used to perform soil salinity inversion on the target crop area using the soil salinity inversion model, and obtain the soil salinity inversion results of the target crop area.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program is executed by the processor to perform the method of any one of claims 1 to 7.

Images