Vegetation and soil coupling effect evaluation method and system based on big data

Abstract

The invention discloses a vegetation and soil coupling effect evaluation method and system based on big data, and the method comprises the steps: collecting experiment data and monitoring data of a preset region, and carrying out the preprocessing of the experiment data and the monitoring data; performing multi-dimensional coupling correlation feature extraction on the experimental data and the monitoring data to obtain a coupling feature set, and performing space-air-ground collaborative coupling evaluation on the monitoring data through the coupling feature set to obtain an original coupling degree; performing disturbance evaluation on the soil data according to a soil dynamic adjustment decision to obtain a disturbance coupling degree, and performing coupling mechanism decomposition based on deep learning according to the original coupling degree and the disturbance coupling degree to obtain a coupling difference mechanism; and constructing a vegetation and soil coupling effect evaluation model according to the disturbance coupling effect and the coupling difference mechanism, inputting to-be-evaluated data into the vegetation and soil coupling effect evaluation model, and outputting an evaluation result.

Description

Technical Field

[0001] This invention relates to the field of vegetation-soil coupling evaluation, and more particularly to a method and system for evaluating the effect of vegetation-soil coupling based on big data. Background Technology

[0002] As core components of terrestrial ecosystems, vegetation and soil directly influence ecosystem stability, nutrient cycling, and service functions, playing a decisive role in key ecological processes such as soil and water conservation, carbon sequestration regulation, and biodiversity maintenance. With the intensification of global climate change and human activities, ecological problems such as vegetation degradation, soil desertification, and nutrient imbalance are becoming increasingly prominent. Accurately assessing the coupling effect between vegetation and soil has become a core requirement for ecological protection, land restoration, and sustainable management.

[0003] Traditional coupled assessment methods often rely on single-dimensional data or local monitoring samples, which have limitations such as narrow data coverage, low spatiotemporal resolution, and insufficient characterization of coupling mechanisms, making it difficult to capture the complex interactions and dynamic changes of vegetation-soil systems. Furthermore, existing methods lack effective integration of multi-source data and fail to fully utilize the massive data resources provided by integrated space-air-ground monitoring technology, resulting in insufficient accuracy, timeliness, and universality of assessment results, and failing to provide scientific and precise support for ecological restoration decision-making.

[0004] Therefore, there is an urgent need to invent a new big data-based evaluation method for vegetation and soil coupling effects. This method should integrate multi-dimensional and multi-scale vegetation, soil, and environmental data using big data technology, and combine deep learning, information theory, and other theoretical methods to mine coupling correlation characteristics, analyze coupling difference mechanisms, and construct an efficient and accurate coupling effect evaluation model. This will become a key path to break through the bottleneck of traditional evaluation and improve the level of ecosystem management. Summary of the Invention

[0005] The purpose of this invention is to provide a method for evaluating the coupling effect of vegetation and soil based on big data.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution: This invention includes the following steps: Experimental and monitoring data were collected from a pre-defined area, and the data were preprocessed. The experimental data included vegetation data, soil data, and environmental data. The monitoring data included vegetation monitoring data and soil monitoring data. The vegetation data included physiological indicators, biomass and structure, and chemical composition. The soil data included physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data included phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data included multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the effects of alternating wet and dry conditions on soil structure. Multi-dimensional coupling correlation feature extraction is performed on the experimental data and the monitoring data to obtain a coupling feature set. The original coupling degree is obtained by performing air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set. Based on the soil dynamic adjustment decision, the soil data is perturbed to obtain the perturbation coupling degree. Based on the original coupling degree and the perturbation coupling degree, the coupling mechanism is decomposed based on deep learning to obtain the coupling difference mechanism. Based on the disturbance coupling effect and the coupling difference mechanism, a vegetation-soil coupling effect evaluation model is constructed. The data to be evaluated is input into the vegetation-soil coupling effect evaluation model, and the evaluation results are output.

[0007] Furthermore, a method for extracting coupled features from the experimental data and the monitoring data in multiple dimensions to obtain a coupled feature set includes: Construct an initial vegetation community feature matrix, where rows represent sampling points and columns represent vegetation feature dimensions; Construct a soil physicochemical property matrix, where rows represent sampling points corresponding to the initial vegetation community feature matrix, and columns represent soil feature dimensions; Calculate carbon storage, water oxygen content, and soil retention. Construct an ecological function parameter matrix based on carbon storage, water oxygen content, and soil retention. Calculate vegetation-soil mutual information. Construct a vegetation-soil mutual information matrix based on vegetation-soil mutual information. Calculate the temporal transfer entropy matrix, construct the information-theoretic correlation strength matrix, construct the vegetation-soil coupling network, the node set includes vegetation feature nodes and soil feature nodes, and the edge set is determined by spatial adjacency and attribute similarity. Obtain node coupling strength indices and extract sub-matrices between vegetation nodes and soil nodes to generate spatial coupling weight matrices; node coupling strength indices include node betweenness centrality, eigenvector centrality, and node coupling strength. The expert knowledge weight matrix is ​​obtained, and the information theory correlation strength matrix, spatial coupling weight matrix and expert knowledge weight matrix are weighted and fused to obtain a multi-source weight matrix. The importance of the features of the multi-source weight matrix is ​​evaluated to obtain the comprehensive importance of the features. To obtain the inflection point of the importance distribution, features with importance higher than the inflection point within the multi-source weight matrix are selected. The ecological function parameters are then projected onto the selected feature space to obtain the coupled feature set, expressed as:

[0008] in For the coupling feature set, For the feature set within the multi-source weight matrix above the inflection point, This is an ecological function parameter matrix.

[0009] Furthermore, a method for obtaining the original coupling degree by performing a space-air-ground coordinated coupling assessment on the monitoring data through a coupling feature set includes: Establish a unified geographic reference framework, perform time matching on the coupled feature set and monitoring data, and vertically stratify the coupled feature set after time matching; The root zone soil moisture was retrieved using a stochastic rough surface model, and a vegetation correction term was introduced to correct the root zone soil moisture. Vegetation cover and water stress index were obtained by inverting multi-band vegetation parameters from Landsat. The canopy three-dimensional structure was reconstructed and hyperspectral vegetation biochemical parameters were retrieved based on UAV data. Vertical coupling feature layers were constructed based on root zone soil corrected moisture, water stress index, canopy three-dimensional structure, and leaf area index. The vertical coupling feature layers are canopy-root coupling layer, root-soil coupling layer, and surface-atmosphere coupling layer. Canopy and root coupling layer: Extract the correlation characteristics between canopy physiological activity and root zone water, and calculate the vertical coupling index of canopy water use efficiency; Root-soil coupling layer: Construct a root distribution matching matrix in soil properties and calculate the vertical gradient of effective soil water capacity. Surface-atmosphere coupling layer: Calculate the correlation between evapotranspiration and surface temperature gradient and construct an energy balance closure index. Scale-adaptive weight allocation is performed based on the coupling feature set. The scale-adaptive weight allocation includes expanding the spatial heterogeneity index and calculating the scale weight. The coupling degree of multi-source data is calculated based on the coupling feature set. Computational multi-scale fusion, vertical and horizontal coupling integration, and environmental covariate correction:

[0010]

[0011] in For position Vertical and horizontal integration coupling at the location Here is the total scale number, and here is the location. First Adaptive weights at each scale For the first radius of scale The calculated local coupling degree For position Vertical coupling index at the location. These are the horizontal coupling weighting coefficients. For environmental regulation function, For in position The degree of ecological coupling at the location, For precipitation, For slope, The slope aspect is represented; the output is the original coupling degree based on the ecological coupling degree.

[0012] Furthermore, a method for obtaining the perturbation coupling degree by perturbing the soil data based on soil dynamic adjustment decisions includes: The preset area is divided into grid cells. Each cell contains soil physical parameters, chemical parameters and vegetation parameters. The cell state is defined as a five-dimensional state vector, which includes vegetation biomass, soil organic carbon content, root soil volume water content, root resistance coefficient and coupling coordination degree. A multi-factor logistic probability model is used to update the cell state, and a dynamic neighborhood window is designed. Root competition is considered at the micro scale, and the NDVI spatial autocorrelation coefficient is introduced at the macro scale. The neighborhood influence weight is adjusted according to the time step exponent. The interference rules are designed based on the type of interference, including: For extreme precipitation, when the daily precipitation exceeds 50mm, the soil erosion rule is activated, with the expression:

[0013] in For runoff depth, Let be the soil organic carbon content at time t. Let be the soil organic carbon content at time t+1; For vegetation degradation, a remediation rule is activated when the coupling coordination degree is less than 0.4, where the remediation rule automatically selects root inhibition coefficients and strategy plants; for fertilization remediation, an organic carbon enhancement rule is triggered when the Shapley interpretable value identifies soil organic carbon content as a limiting factor; the expression is:

[0014] in The fertilizer application rate is set; control scenario, restoration scenario, and time step are set; the control scenario has no human intervention, and the cells evolve according to natural rules; the restoration scenario applies intervention rules to the cells in the target area according to the preset decision plan. Extract the cell state coupling coordination degree in the 5th year under the repair scenario, and subtract the coupling coordination degree of the repair scenario from that of the control scenario to obtain the perturbation coupling degree increment.

[0015] Furthermore, the method for obtaining the coupling difference mechanism includes: A heterogeneous graph was constructed based on experimental and monitoring data, where nodes represent vegetation / soil elements and edge weights represent the interaction strength between nodes. A multi-view graph neural network is used to fuse multi-dimensional features including physiological, structural, and chemical features, and a parameterized graph neural network interpreter is used to identify the subgraph structure that contributes the most to the coupling degree. For time-series monitoring data of soil temperature and humidity and vegetation phenology, the spatiotemporal attention mechanism of ST-Transformer is used to capture long-term dependencies; the study area is divided into 3km×3km grids, and the spatial interaction between grids is calculated through self-attention layers to obtain a high-resolution perturbation coupling degree map. The contribution of each element to the difference in coupling degree is calculated by using the SHAP value of the parametric graph neural network interpreter. The distribution of SHAP values ​​under the original and perturbed scenarios is compared to identify key regulatory factors. A dynamic response matrix between elements is constructed based on the attention weights of the spatiotemporal Transformer, and the dynamic response coefficient is calculated based on the dynamic response matrix. The key regulatory factors and dynamic response coefficients are output as the coupling difference mechanism.

[0016] Furthermore, the method for constructing an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism includes: Experimental and monitoring data were standardized and forward-oriented. A leaf optical characteristic radiative transfer model was constructed by combining vegetation chemical composition and hyperspectral red edge band to retrieve leaf nitrogen content. The hysteresis response features of soil temperature and humidity time series data were extracted by long short-term memory network and a dynamic feature set was constructed by combining phenological observations. The entropy weight method was used to determine the index weights of the vegetation subsystem and the soil subsystem respectively, and the weighted summation was used to obtain the plant comprehensive index and the soil comprehensive index; the soil comprehensive index includes physical properties, chemical properties and biological properties; The coupling degree and coordination degree are calculated based on the vegetation subsystem and soil subsystem. The study area is divided into grid cells. Each cell state includes the current coupling degree, soil diffusion rate, and vegetation root resistance coefficient. Soil moisture and nutrients drive dynamic vegetation density, expressed as:

[0017] in For growth rate, For mortality rate, For soil moisture, For nutrients, Let be the vegetation density at time t. Let be the vegetation density at time t-1; The impact of remediation measures on cell states is simulated using U-Net as the basic architecture, embedding Richards' soil moisture movement equation as a physical constraint term, expressed as:

[0018] in For data loss, For physical loss, This refers to the soil volumetric water content. This is the physical loss weighting coefficient. For the total loss function, Based on actual observation data, To predict soil moisture movement state variables, For divergence operators, The water conductivity of unsaturated soil; By decomposing the difference in coupling degree before and after the perturbation using SHAP values, key driving factors are identified, and the coupling effect is graded according to the degree of coordination and output as a coupling effect evaluation result.

[0019] Secondly, a big data-based evaluation system for the coupling effect between vegetation and soil includes: Data Acquisition and Processing Module: Used to collect experimental and monitoring data from a preset area, and to preprocess the experimental and monitoring data. The experimental data includes vegetation data, soil data, and environmental data. The monitoring data includes vegetation monitoring data and soil monitoring data. The vegetation data includes physiological indicators, biomass and structure, and chemical composition. The soil data includes physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data includes phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data includes multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the impact of alternating wet and dry conditions on soil structure. Feature extraction and coupling evaluation module: used to extract multi-dimensional coupling correlation features from the experimental data and the monitoring data to obtain a coupling feature set, and to perform air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set to obtain the original coupling degree; The perturbation and mechanism decomposition module is used to evaluate the perturbation of the soil data based on the soil dynamic adjustment decision to obtain the perturbation coupling degree, and to perform deep learning-based coupling mechanism decomposition based on the original coupling degree and the perturbation coupling degree to obtain the coupling difference mechanism. Model construction and output module: This module is used to construct an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism, input the data to be evaluated into the evaluation model of vegetation-soil coupling effect, and output the evaluation results.

[0020] The beneficial effects of this invention are: This invention relates to a method and system for evaluating the coupling effect of vegetation and soil based on big data. Compared with existing technologies, this invention has the following technical advantages: This invention improves the accuracy of user intent recognition in multi-dimensional human-computer interaction scenarios through preprocessing, feature extraction, coupling evaluation, perturbation evaluation, coupling mechanism decomposition, and model construction. By extracting multi-dimensional features, conducting space-air-ground collaborative evaluation, and simulating perturbations, it accurately captures the coupling patterns and key regulatory factors of vegetation-soil. The evaluation model integrates physical constraints and dynamic response mechanisms to improve the accuracy and timeliness of results, providing scientific support for ecological restoration and land management, and possessing both universality and practicality. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the steps of the vegetation-soil coupling effect evaluation method based on big data, as described in this invention. Detailed Implementation

[0022] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0023] The present invention provides a method and system for evaluating the coupling effect of vegetation and soil based on big data, comprising the following steps: like Figure 1 As shown, this embodiment includes the following steps: Experimental and monitoring data were collected from a pre-defined area, and the data were preprocessed. The experimental data included vegetation data, soil data, and environmental data. The monitoring data included vegetation monitoring data and soil monitoring data. The vegetation data included physiological indicators, biomass and structure, and chemical composition. The soil data included physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data included phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data included multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the effects of alternating wet and dry conditions on soil structure. In the actual assessment, a certain ecological restoration area on a plateau (geographical coordinates: 106°20′-107°30′E, 35°10′-36°20′N) was taken as the research object. The area of ​​this region is 500 km². 2 The terrain is mainly hilly and gully, the climate is semi-arid continental, the average annual precipitation is 450 mm, the main vegetation types are sea buckthorn, oil pine and white sheep grass, and the soil type is mainly yellow cotton soil. Experimental data: Physiological indicators of vegetation (chlorophyll content 35-60 mg / g, photosynthetic rate 8-22 μmol / (m²)) were collected from 2020 to 2024. 2 •s)), biomass (aboveground biomass 2-15 t / hm) 2 Chemical composition (leaf nitrogen content 1.2%-2.8%); soil physical properties (bulk density 1.1-1.5 g / cm³). 3 Porosity 40%-55%, chemical properties (pH 7.5-8.5, organic carbon content 0.5%-1.8%), biological properties (microbial biomass carbon 50-200 mg / kg), and hydrological parameters (field holding capacity 20%-30%); environmental data (average annual temperature 8.5℃, annual sunshine hours 2600h); Monitoring data: Normalized Difference Vegetation Index (NDVI 0.3-0.8) and Enhanced Vegetation Index (EVI 0.2-0.7) were acquired via satellite remote sensing; UAV monitoring was used to measure the three-dimensional structure of the canopy (average height 2-6m) and hyperspectral data (700-750nm red-edge band); Ground monitoring was conducted on multiple soil layers: temperature and humidity (0-20cm soil layer 10%-25%, 20-60cm soil layer 15%-28%), soil electrical conductivity (50-200μS / cm), pH value, and CO2 flux (1-5μmol / (m²)). 2•s), as well as rainfall, evapotranspiration (300-450 mm / year), surface runoff (5-20 mm / year), and erosion (500-2000 t / (km²)). 2 ·Year)); Missing data were supplemented using linear interpolation, outliers were removed using the 3σ principle, and all data were standardized (mean 0, variance 1). The WGS-84 coordinate system and UTM projection were used uniformly, with a reference resolution of 30m. Multi-dimensional coupling correlation feature extraction is performed on the experimental data and the monitoring data to obtain a coupling feature set. The original coupling degree is obtained by performing air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set. Based on the soil dynamic adjustment decision, the soil data is perturbed to obtain the perturbation coupling degree. Based on the original coupling degree and the perturbation coupling degree, the coupling mechanism is decomposed based on deep learning to obtain the coupling difference mechanism. Based on the disturbance coupling effect and the coupling difference mechanism, a vegetation-soil coupling effect evaluation model is constructed. The data to be evaluated is input into the vegetation-soil coupling effect evaluation model, and the evaluation results are output.

[0024] In this embodiment, the method for extracting coupled features from the experimental data and the monitoring data in multiple dimensions to obtain a coupled feature set includes: An initial vegetation community feature matrix was constructed, with rows representing sampling points and columns representing vegetation feature dimensions. The initial vegetation community feature matrix integrates vegetation physiological indicators, biomass and structure, and chemical composition from experimental data. It also integrates normalized vegetation index, enhanced vegetation index, leaf area index, and photosynthetically active radiation absorption ratio from monitoring data. Dimensionality reduction was achieved through principal component analysis, and principal components with cumulative contribution rates greater than or equal to the contribution threshold were retained to obtain an optimized vegetation community feature matrix. A soil physicochemical property matrix was constructed, with rows representing sampling points corresponding to the initial vegetation community characteristic matrix and columns representing soil characteristic dimensions. The soil physicochemical property matrix integrates physical, chemical, biological properties and hydrological parameters from experimental data. The soil physicochemical property matrix also integrates multi-layer soil temperature, humidity, electrical conductivity, pH and carbon dioxide flux from monitoring data. Standardization was used to eliminate dimensions to obtain a standard soil physicochemical property matrix. Calculate carbon storage, water oxygen content, and soil retention:

[0025]

[0026]

[0027] in For aboveground biomass, Soil organic matter, Field holding capacity The wilting coefficient, As an intrusive rain erosion factor, As a soil erosion factor, For slope length erosion factor, For vegetation cover factor, For soil retention, Oxygen content of the water source For carbon reserves, The carbon conversion factor of aboveground biomass. The soil organic matter carbon conversion coefficient; An ecological function parameter matrix is ​​constructed based on carbon storage, water oxygen content, and soil retention. Vegetation-soil mutual information is then calculated, and a vegetation-soil mutual information matrix is ​​constructed based on this mutual information. The expression is as follows:

[0028] in To optimize the i-th vegetation feature in the vegetation community feature matrix, Let j be the j-th soil feature in the soil physicochemical property matrix. Vegetation characteristics and soil characteristics The joint probability distribution, Vegetation characteristics The marginal probability, Soil characteristics The marginal probability, Vegetation characteristics and soil characteristics Mutual information; Calculate the time-series transfer entropy matrix:

[0029] in For the t-th time... Embedded dimension vegetation features For the t-th time... One embedded dimension of soil features, Let the vegetation characteristics be at time t+1. The transfer entropy of soil parameters to vegetation parameters. Construct the information-theoretic correlation strength matrix, expressed as:

[0030] in Let be the information theory correlation strength between the i-th vegetation feature and the j-th soil feature. Let be the transfer entropy of the i-th soil parameter to the j-th vegetation parameter. To maximize the propagation entropy, For maximum mutual information, Mutual information weighting coefficients For the transfer entropy weighting coefficient; A vegetation-soil coupled network is constructed, with the node set including vegetation feature nodes and soil feature nodes. The edge set is determined by spatial adjacency and attribute similarity, expressed as:

[0031] in Let be the spatial distance between the i-th node and the j-th node. For bandwidth parameters, For feature similarity function, For the i-th node, For the j-th node, The maximum distance threshold, Let the connection weights be the connection weights between the i-th node and the j-th node. This is the balance coefficient; Obtain node coupling strength indices, extract sub-matrices between vegetation nodes and soil nodes to generate a spatial coupling weight matrix; node coupling strength indices include node betweenness centrality, eigenvector centrality, and node coupling strength; calculate node betweenness centrality, eigenvector centrality, and node coupling strength:

[0032]

[0033]

[0034] in For nodes betweenness centrality, For nodes eigenvector centrality, For nodes The overall coupling strength, The centrality balance coefficient, For nodes The set of neighboring nodes, For nodes and nodes Connection weights between them Obtain the expert knowledge weight matrix, and then perform a weighted fusion of the information theory correlation strength matrix, spatial coupling weight matrix, and expert knowledge weight matrix to obtain a multi-source weight matrix. Finally, evaluate the importance of the features in the multi-source weight matrix to obtain the comprehensive feature importance, expressed as:

[0035]

[0036]

[0037] in For the z-th soil correlation strength, Let z be the vegetation association strength. Let be the fusion weight between the i-th node and the j-th node. The z-th feature of the multi-source weight matrix, For vegetation feature nodes, These are soil characteristic nodes. Features Overall importance; To obtain the inflection point of the importance distribution, features with importance higher than the inflection point within the multi-source weight matrix are selected. The ecological function parameters are then projected onto the selected feature space to obtain the coupled feature set, expressed as:

[0038] in For the coupling feature set, For the feature set within the multi-source weight matrix above the inflection point, This is an ecological function parameter matrix; In the actual assessment, the contribution threshold was 0.85, and an initial vegetation community feature matrix (100 sampling points × 15 feature dimensions) and soil physicochemical property matrix (100 sampling points × 12 feature dimensions) were constructed. Calculate ecological function parameters: carbon storage 25-80 t / hm 2 Water source oxygen content 10%-20%, soil retention capacity 1500-3000 t / (km²) 2 • (year), construct an ecological function parameter matrix; By using mutual information calculation and temporal transfer entropy analysis, an information-theoretic correlation strength matrix is ​​constructed. Combined with spatial adjacency relationships, a vegetation-soil coupling network is constructed. After integrating the expert knowledge weight matrix, 20 key features are selected to generate a coupling feature set.

[0039] In this embodiment, the method for obtaining the original coupling degree by performing a space-air-ground coordinated coupling assessment on the monitoring data through a coupling feature set includes: A unified geographic reference framework is established to perform time matching on the coupled feature set and monitoring data, and the time-matched coupled feature set is vertically layered; the WGS-84 coordinate system and UTM projection are adopted, with a reference resolution of 30m, to construct a multi-scale pyramid data structure. ,in Corresponding to the baseline resolution; temporal matching includes satellite data temporal resampling, UAV data spatial aggregation, and ground data spatial interpolation; vertical stratification includes vegetation canopy layer, root active layer, and deep soil layer, decomposing P-band radar data; The root zone soil moisture is inverted using a stochastic rough surface model, and a vegetation correction term is introduced to correct the root zone soil moisture. The expression is as follows:

[0040] in To adjust soil moisture in the root zone, The angle of incidence is the angle between the radar beam and the vertical normal to the ground surface. for Polarization backscattering coefficient, the intensity of radar echo transmitted and received horizontally; for Polarization backscattering coefficient, the intensity of radar echo transmitted vertically and received vertically; for Polarization backscattering coefficient, cross-polarization echo intensity when transmitted horizontally and received vertically; Let be the decay function caused by vegetation moisture content. For the inverted root zone soil moisture, For a random rough surface model, Leaf area index, Normalized Difference Vegetation Index; Vegetation cover and water stress index were obtained by inverting multi-band vegetation parameters from Landsat satellites, and the expressions are as follows:

[0041]

[0042]

[0043] in The reflectance of the TM image in band 4. The reflectance of the TM image in band 3. This represents the maximum reflectivity of the TM4 band. This represents the minimum reflectivity in the TM3 band. For parameters related to atmospheric transmittance, This is the brightness temperature value for the TM6 band. , , For atmospheric parameters, Here, is the surface emissivity, and is the surface temperature of a pixel. This represents the minimum surface temperature in the study area. This represents the maximum surface temperature in the study area. This is the water stress index. Vegetation coverage; Based on UAV data, the canopy 3D structure reconstruction and hyperspectral vegetation biochemical parameter inversion were performed, expressed as follows:

[0044]

[0045] in For the first One regression coefficient, For the first Reflectivity at a specific wavelength For cell position Canopy height at that location, For digital surface models, For digital terrain models, For the regression intercept term, For the residual term, For the first One center wavelength; Vertical coupling feature layers were constructed based on root zone soil corrected moisture, water stress index, canopy three-dimensional structure, and leaf area index. The vertical coupling feature layers are canopy-root coupling layer, root-soil coupling layer, and surface-atmosphere coupling layer. Canopy and root coupling layer: The correlation characteristics between canopy physiological activity and root zone water use were extracted, and the vertical coupling index of canopy water use efficiency was calculated. The expression is as follows:

[0046]

[0047] in The vertical coupling index for canopy water use efficiency. Total primary production flux, Evapotranspiration flux This is the soil moisture gradient adjustment coefficient. This refers to surface soil moisture. For deep soil moisture, Soil moisture gradient, To enhance the vegetation index, Photochemical index; Root-soil coupling layer: Construct a root distribution matching matrix with soil properties, and calculate the vertical gradient of effective soil water capacity, expressed as:

[0048]

[0049] in For soil depth, Root density distribution Soil depth Soil organic matter content, Soil depth Soil bulk density, Soil depth The distribution of gap size, For tensor product, Soil depth Root system soil property matching matrix, This is the lower limit of soil depth. This represents the upper limit of soil depth. The vertical gradient of soil effective water capacity; Surface-Atmosphere Coupling Layer: The correlation between evapotranspiration and surface temperature gradient is calculated, and an energy balance closure index is constructed, expressed as:

[0050]

[0051] in For sensible heat flux, For latent heat flux, Net radiation, Evapotranspiration flux For soil heat flux, Let V be the variance of the surface temperature gradient. This represents the covariance between evapotranspiration flux and surface temperature gradient. Scale-adaptive weight allocation is performed based on the coupled feature set. This allocation includes the spatial heterogeneity index and the calculation of scale weights, expressed as follows:

[0052]

[0053] in For Center, radius The neighborhood, The number of pixels in the neighborhood. For the first A scale radius For temperature parameters, For position Location, radius Spatial heterogeneity index For pixels The coupling eigenvalues, Center pixel The coupling eigenvalues, The radius of the spatial dimension. For position First Adaptive weights at each scale The number of scales; Calculating the coupling degree of multi-source data based on coupling feature sets:

[0054]

[0055] in For position Local coupling degree, For in position At the i-th vegetation feature, For in position At the i-th soil feature, For the feature pair total, The weight of the i-th feature is... The vertical coupling index. This is a characteristic of canopy-root coupling. This is a characteristic of root system coupling with soil. This is a characteristic of the coupling between the Earth's surface and the atmosphere. The correlation coefficients for each coupling layer; Computational multi-scale fusion, vertical and horizontal coupling integration, and environmental covariate correction:

[0056]

[0057] in For position Vertical and horizontal integration coupling at the location Here is the total scale number, and here is the location. First Adaptive weights at each scale For the first radius of scale The calculated local coupling degree For position Vertical coupling index at the location. These are the horizontal coupling weighting coefficients. For environmental regulation function, For in position The degree of ecological coupling at the location, For precipitation, For slope, Slope direction; The output is the original coupling degree based on the ecological coupling degree; In actual assessments, the vegetation canopy layer is 0-2m, the root active layer is 0-60cm, and the deep soil layer is 60-120cm. Key parameters retrieved: root zone soil moisture 12%-26%, vegetation cover 40%-85%, water stress index 0.3-0.7, reconstructed canopy three-dimensional structure, and retrieved leaf area index 1.5-4.0; Calculate the vertical coupling index: vertical coupling index of canopy water use efficiency 0.4-0.8, vertical gradient of soil effective water capacity 0.05-0.15, and energy balance closure index 0.6-0.9; After multi-scale fusion and environmental covariate correction, the original coupling degree of the region was obtained as 0.35-0.78.

[0058] In this embodiment, the method for obtaining the perturbation coupling degree by perturbing the soil data based on soil dynamic adjustment decisions includes: The preset area is divided into grid cells. Each cell contains soil physical parameters, chemical parameters and vegetation parameters. The cell state is defined as a five-dimensional state vector, which includes vegetation biomass, soil organic carbon content, root soil volume water content, root resistance coefficient and coupling coordination degree. The cell state is updated using a multi-factor logistic probability model, expressed as follows:

[0059] in The wilting coefficient, The organic carbon threshold, The threshold value for the root resistance coefficient. This is the root resistance coefficient. For random interference terms, The root feedback coefficient is... This refers to the volumetric water content of the root zone soil. Soil organic carbon content, This is the organic carbon weighting coefficient. This is the soil moisture content weighting coefficient; The design incorporates a dynamic neighborhood window, considering root competition at the microscale and introducing the NDVI spatial autocorrelation coefficient at the macroscale. The neighborhood influence weights are adjusted based on the time step exponent, expressed as follows:

[0060] in The adjusted neighborhood influence weight at time t. As the initial neighborhood influence weight, The time decay coefficient, For time step; The interference rules are designed based on the type of interference, including: For extreme precipitation, when the daily precipitation exceeds 50mm, the soil erosion rule is activated, with the expression:

[0061] in For runoff depth, Let be the soil organic carbon content at time t. Let be the soil organic carbon content at time t+1; For vegetation degradation, a remediation rule is activated when the coupling coordination degree is less than 0.4, where the remediation rule automatically selects root inhibition coefficients and strategy plants; for fertilization remediation, an organic carbon enhancement rule is triggered when the Shapley interpretable value identifies soil organic carbon content as a limiting factor; the expression is:

[0062] in The fertilizer application rate is set; control scenario, restoration scenario, and time step are set; the control scenario has no human intervention, and the cells evolve according to natural rules; the restoration scenario applies intervention rules to the cells in the target area according to the preset decision plan. Extract the cell state coupling coordination degree in the 5th year under the repair scenario, and subtract the coupling coordination degree of the repair scenario from that of the control scenario to obtain the perturbation coupling degree increment. In the actual evaluation, the grid cell is 10m×10m, and the coupling coordination degree is the original coupling degree multiplied by the development level index and the square root. The neighborhood window at the micro scale is 3×3 and the neighborhood window at the macro scale is 11×11. When the NDVI spatial autocorrelation coefficient is greater than 0.6, the cluster growth rule is triggered. Vegetation planting: When planting deep-rooted plants, such as sea buckthorn, the soil's resistance to erosion increases by 8% for every 10cm increase in the depth of the cell roots, and the corresponding coupling coordination value increases by 0.05. Disturbance scenarios: extreme precipitation scenario (daily precipitation 60 mm, runoff depth 8 mm), vegetation degradation restoration scenario (coupling coordination degree 0.35, planting sea buckthorn), fertilization restoration scenario (fertilizer application rate 500 kg / hm²). 2 The increase in perturbation coupling degree is as follows: it decreases by 0.08-0.15 under extreme precipitation scenarios, increases by 0.12-0.20 under vegetation restoration scenarios, and increases by 0.09-0.16 under fertilization restoration scenarios.

[0063] In this embodiment, the method for obtaining the coupling difference mechanism includes: A heterogeneous graph was constructed based on experimental and monitoring data, where nodes represent vegetation / soil elements and edge weights represent the interaction strength between nodes. A multi-view graph neural network is used to fuse multi-dimensional features including physiological, structural, and chemical features, and a parameterized graph neural network interpreter is used to identify the subgraph structure that contributes the most to the coupling degree. For time-series monitoring data of soil temperature and humidity and vegetation phenology, the spatiotemporal attention mechanism of ST-Transformer is used to capture long-term dependencies; the study area is divided into 3km×3km grids, and the spatial interaction between grids is calculated through self-attention layers to obtain a high-resolution perturbation coupling degree map. The contribution of each element to the difference in coupling degree is calculated by using the SHAP value of the parametric graph neural network interpreter. The distribution of SHAP values ​​under the original and perturbed scenarios is compared to identify key regulatory factors. A dynamic response matrix between elements is constructed based on the attention weights of the spatiotemporal Transformer, and the dynamic response coefficient is calculated based on the dynamic response matrix. The key regulatory factors and dynamic response coefficients are output as the coupling difference mechanism; In the actual assessment, the coupling difference mechanism identified soil organic carbon content, root zone soil moisture, and vegetation cover as key regulatory factors, with dynamic response coefficients of 0.32, 0.28, and 0.25, respectively.

[0064] In this embodiment, the method for constructing an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism includes: Experimental and monitoring data were standardized and forward-oriented. A leaf optical characteristic radiative transfer model was constructed by combining vegetation chemical composition and hyperspectral red edge band to retrieve leaf nitrogen content. The hysteresis response features of soil temperature and humidity time series data were extracted by long short-term memory network and a dynamic feature set was constructed by combining phenological observations. The entropy weight method was used to determine the index weights of the vegetation subsystem and the soil subsystem respectively, and the weighted summation was used to obtain the plant comprehensive index and the soil comprehensive index; the soil comprehensive index includes physical properties, chemical properties and biological properties; The coupling degree and coordination degree are calculated based on the vegetation subsystem and soil subsystem. The study area is divided into grid cells. Each cell state includes the current coupling degree, soil diffusion rate, and vegetation root resistance coefficient. Soil moisture and nutrients drive dynamic vegetation density, expressed as:

[0065] in For growth rate, For mortality rate, For soil moisture, For nutrients, Let be the vegetation density at time t. Let be the vegetation density at time t-1; The impact of remediation measures on cell states is simulated using U-Net as the basic architecture, embedding Richards' soil moisture movement equation as a physical constraint term, expressed as:

[0066] in For data loss, For physical loss, This refers to the soil volumetric water content. This is the physical loss weighting coefficient. For the total loss function, Based on actual observation data, To predict soil moisture movement state variables, For divergence operators, The water conductivity of unsaturated soil; By decomposing the difference in coupling degree before and after the perturbation using SHAP values, key driving factors are identified, and the coupling effect is graded according to the degree of coordination and output as a coupling effect evaluation result. In actual assessment, the red-edge band of the spectrum is 700-750nm. Phenological observation, namely LSI at the beginning of the growing season, is based on U-Net architecture, embedding the Richards soil moisture movement equation, and combining long short-term memory network to extract temporal features.

[0067] Model validation: Training set accuracy 92%, validation set accuracy 89%, Kappa coefficient 0.85.

[0068] Evaluation results: The coupling effect in the study area was graded as poor (0.35-0.45, accounting for 15%), medium (0.45-0.60, accounting for 45%), good (0.60-0.75, accounting for 30%), and excellent (0.75-0.78, accounting for 10%). Among them, the coupling effect of vegetation restoration and fertilization areas was significantly improved.

[0069] Secondly, a big data-based evaluation system for the coupling effect between vegetation and soil includes: Data Acquisition and Processing Module: Used to collect experimental and monitoring data from a preset area, and to preprocess the experimental and monitoring data. The experimental data includes vegetation data, soil data, and environmental data. The monitoring data includes vegetation monitoring data and soil monitoring data. The vegetation data includes physiological indicators, biomass and structure, and chemical composition. The soil data includes physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data includes phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data includes multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the impact of alternating wet and dry conditions on soil structure. Feature extraction and coupling evaluation module: used to extract multi-dimensional coupling correlation features from the experimental data and the monitoring data to obtain a coupling feature set, and to perform air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set to obtain the original coupling degree; The perturbation and mechanism decomposition module is used to evaluate the perturbation of the soil data based on the soil dynamic adjustment decision to obtain the perturbation coupling degree, and to perform deep learning-based coupling mechanism decomposition based on the original coupling degree and the perturbation coupling degree to obtain the coupling difference mechanism. Model construction and output module: This module is used to construct an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism, input the data to be evaluated into the evaluation model of vegetation-soil coupling effect, and output the evaluation results.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the coupling effect of vegetation and soil based on big data, characterized in that, Includes the following steps: Experimental and monitoring data were collected from a pre-defined area, and the data were preprocessed. The experimental data included vegetation data, soil data, and environmental data. The monitoring data included vegetation monitoring data and soil monitoring data. The vegetation data included physiological indicators, biomass and structure, and chemical composition. The soil data included physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data included phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data included multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the effects of alternating wet and dry conditions on soil structure. Multi-dimensional coupling correlation feature extraction is performed on the experimental data and the monitoring data to obtain a coupling feature set. The original coupling degree is obtained by performing air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set. Based on the soil dynamic adjustment decision, the soil data is perturbed to obtain the perturbation coupling degree. Based on the original coupling degree and the perturbation coupling degree, the coupling mechanism is decomposed based on deep learning to obtain the coupling difference mechanism. Based on the disturbance coupling effect and the coupling difference mechanism, a vegetation-soil coupling effect evaluation model is constructed. The data to be evaluated is input into the vegetation-soil coupling effect evaluation model, and the evaluation results are output.

2. The method for evaluating the coupling effect of vegetation and soil based on big data according to claim 1, characterized in that, A method for extracting coupled feature sets from the experimental data and the monitoring data using multi-dimensional coupled correlation features includes: Construct an initial vegetation community feature matrix, where rows represent sampling points and columns represent vegetation feature dimensions; Construct a soil physicochemical property matrix, where rows represent sampling points corresponding to the initial vegetation community feature matrix, and columns represent soil feature dimensions; Calculate carbon storage, water oxygen content, and soil retention. Construct an ecological function parameter matrix based on carbon storage, water oxygen content, and soil retention. Calculate vegetation-soil mutual information. Construct a vegetation-soil mutual information matrix based on vegetation-soil mutual information. Calculate the temporal transfer entropy matrix, construct the information-theoretic correlation strength matrix, construct the vegetation-soil coupling network, the node set includes vegetation feature nodes and soil feature nodes, and the edge set is determined by spatial adjacency and attribute similarity. Obtain node coupling strength indices and extract sub-matrices between vegetation nodes and soil nodes to generate spatial coupling weight matrices; node coupling strength indices include node betweenness centrality, eigenvector centrality, and node coupling strength. The expert knowledge weight matrix is ​​obtained, and the information theory correlation strength matrix, spatial coupling weight matrix and expert knowledge weight matrix are weighted and fused to obtain a multi-source weight matrix. The importance of the features of the multi-source weight matrix is ​​evaluated to obtain the comprehensive importance of the features. To obtain the inflection point of the importance distribution, features with importance higher than the inflection point within the multi-source weight matrix are selected. The ecological function parameters are then projected onto the selected feature space to obtain the coupled feature set, expressed as: ； in For the coupling feature set, For the feature set within the multi-source weight matrix above the inflection point, This is an ecological function parameter matrix.

3. The method for evaluating the coupling effect of vegetation and soil based on big data according to claim 1, characterized in that, A method for obtaining the original coupling degree by performing a space-air-ground coordinated coupling assessment on the monitoring data using a coupling feature set includes: Establish a unified geographic reference framework, perform time matching on the coupled feature set and monitoring data, and vertically stratify the coupled feature set after time matching; The root zone soil moisture was retrieved using a stochastic rough surface model, and a vegetation correction term was introduced to correct the root zone soil moisture. Vegetation cover and water stress index were obtained by inverting multi-band vegetation parameters from Landsat. The canopy three-dimensional structure was reconstructed and hyperspectral vegetation biochemical parameters were retrieved based on UAV data. Vertical coupling feature layers were constructed based on root zone soil corrected moisture, water stress index, canopy three-dimensional structure, and leaf area index. The vertical coupling feature layers are canopy-root coupling layer, root-soil coupling layer, and surface-atmosphere coupling layer. Canopy and root coupling layer: Extract the correlation characteristics between canopy physiological activity and root zone water, and calculate the vertical coupling index of canopy water use efficiency; Root-soil coupling layer: Construct a root distribution matching matrix in soil properties and calculate the vertical gradient of effective soil water capacity. Surface-atmosphere coupling layer: Calculate the correlation between evapotranspiration and surface temperature gradient and construct an energy balance closure index. Scale-adaptive weight allocation is performed based on the coupling feature set. The scale-adaptive weight allocation includes expanding the spatial heterogeneity index and calculating the scale weight. The coupling degree of multi-source data is calculated based on the coupling feature set. Computational multi-scale fusion, vertical and horizontal coupling integration, and environmental covariate correction: ； ； in For position Vertical and horizontal integration coupling at the location Here is the total scale number, and here is the location. First Adaptive weights for each scale For the first radius of scale The calculated local coupling degree For position Vertical coupling index at the location. These are the horizontal coupling weighting coefficients. For environmental regulation function, For in position The degree of ecological coupling at the location, For precipitation, For slope, The slope aspect is represented; the output is the original coupling degree based on the ecological coupling degree.

4. The method for evaluating the coupling effect of vegetation and soil based on big data according to claim 1, characterized in that, A method for obtaining the perturbation coupling degree by perturbation assessment of the soil data based on soil dynamic adjustment decisions includes: The preset area is divided into grid cells. Each cell contains soil physical parameters, chemical parameters and vegetation parameters. The cell state is defined as a five-dimensional state vector, which includes vegetation biomass, soil organic carbon content, root soil volume water content, root resistance coefficient and coupling coordination degree. A multi-factor logistic probability model is used to update the cell state, and a dynamic neighborhood window is designed. Root competition is considered at the micro scale, and the NDVI spatial autocorrelation coefficient is introduced at the macro scale. The neighborhood influence weight is adjusted according to the time step exponent. The interference rules are designed based on the type of interference, including: For extreme precipitation, when the daily precipitation exceeds 50mm, the soil erosion rule is activated, with the expression: ； in For runoff depth, Let be the soil organic carbon content at time t. Let be the soil organic carbon content at time t+1; For vegetation degradation, a remediation rule is activated when the coupling coordination degree is less than 0.4, where the remediation rule automatically selects root inhibition coefficients and strategy plants; for fertilization remediation, an organic carbon enhancement rule is triggered when the Shapley interpretable value identifies soil organic carbon content as a limiting factor; the expression is: ； in The fertilizer application rate is set; control scenario, restoration scenario, and time step are set; the control scenario has no human intervention, and the cells evolve according to natural rules; the restoration scenario applies intervention rules to the cells in the target area according to the preset decision plan. Extract the cell state coupling coordination degree in the 5th year under the repair scenario, and subtract the coupling coordination degree of the repair scenario from that of the control scenario to obtain the perturbation coupling degree increment.

5. The method for evaluating the coupling effect of vegetation and soil based on big data according to claim 1, characterized in that, The method for obtaining the coupling difference mechanism includes: A heterogeneous graph was constructed based on experimental and monitoring data, where nodes represent vegetation / soil elements and edge weights represent the interaction strength between nodes. A multi-view graph neural network is used to fuse multi-dimensional features including physiological, structural, and chemical aspects, and a parameterized graph neural network interpreter is used to identify the subgraph structure that contributes the most to the coupling degree. For time-series monitoring data of soil temperature and humidity and vegetation phenology, the spatiotemporal attention mechanism of ST-Transformer is used to capture long-term dependencies; the study area is divided into 3km×3km grids, and the spatial interaction between grids is calculated through self-attention layers to obtain a high-resolution perturbation coupling degree map. The contribution of each element to the difference in coupling degree is calculated by using the SHAP value of the parametric graph neural network interpreter. The distribution of SHAP values ​​under the original and perturbed scenarios is compared to identify key regulatory factors. A dynamic response matrix between elements is constructed based on the attention weights of the spatiotemporal Transformer, and the dynamic response coefficient is calculated based on the dynamic response matrix. The key regulatory factors and dynamic response coefficients are output as the coupling difference mechanism.

6. The method for evaluating the coupling effect of vegetation and soil based on big data according to claim 1, characterized in that, A method for constructing an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism includes: Experimental and monitoring data were standardized and forward-oriented. A leaf optical characteristic radiative transfer model was constructed by combining vegetation chemical composition and hyperspectral red edge band to retrieve leaf nitrogen content. The hysteresis response features of soil temperature and humidity time series data were extracted by long short-term memory network and a dynamic feature set was constructed by combining phenological observations. The entropy weight method was used to determine the index weights of the vegetation subsystem and the soil subsystem respectively, and the weighted summation was used to obtain the plant comprehensive index and the soil comprehensive index; the soil comprehensive index includes physical properties, chemical properties and biological properties; The coupling degree and coordination degree are calculated based on the vegetation subsystem and soil subsystem. The study area is divided into grid cells. Each cell state includes the current coupling degree, soil diffusion rate, and vegetation root resistance coefficient. Soil moisture and nutrients drive dynamic vegetation density, expressed as: ； in For growth rate, For mortality rate, For soil moisture, For nutrients, Let be the vegetation density at time t. Let be the vegetation density at time t-1; The impact of remediation measures on cell states is simulated using U-Net as the basic architecture, embedding Richards' soil moisture movement equation as a physical constraint term, expressed as: ； in For data loss, For physical loss, This refers to the soil volumetric water content. This is the physical loss weighting coefficient. For the total loss function, Based on actual observation data, To predict soil moisture movement state variables, For divergence operators, The water conductivity of unsaturated soil; By decomposing the difference in coupling degree before and after the perturbation using SHAP values, key driving factors are identified, and the coupling effect is graded according to the degree of coordination and output as a coupling effect evaluation result.

7. A vegetation-soil coupling effect evaluation system based on big data, used to perform the method according to any one of claims 1-6, characterized in that, include: Data Acquisition and Processing Module: Used to collect experimental and monitoring data from a preset area, and to preprocess the experimental and monitoring data. The experimental data includes vegetation data, soil data, and environmental data. The monitoring data includes vegetation monitoring data and soil monitoring data. The vegetation data includes physiological indicators, biomass and structure, and chemical composition. The soil data includes physical properties, chemical properties, biological properties, and hydrological parameters. The vegetation monitoring data includes phenological observations, vegetation cover dynamics, quadrat species diversity index, normalized difference vegetation index, enhanced vegetation index, leaf area index, photosynthetically active radiation absorption ratio, vegetation cover, vegetation productivity, and hyperspectral data. The soil monitoring data includes multi-layer soil temperature and humidity, soil electrical conductivity, pH, soil CO2 flux, rainfall, air temperature, relative humidity, solar radiation, evapotranspiration, surface runoff, erosion, seasonal soil samples, freeze-thaw processes, and records of the impact of alternating wet and dry conditions on soil structure. Feature extraction and coupling evaluation module: used to extract multi-dimensional coupling correlation features from the experimental data and the monitoring data to obtain a coupling feature set, and to perform air-space-ground collaborative coupling evaluation on the monitoring data through the coupling feature set to obtain the original coupling degree; The perturbation and mechanism decomposition module is used to evaluate the perturbation of the soil data based on the soil dynamic adjustment decision to obtain the perturbation coupling degree, and to perform deep learning-based coupling mechanism decomposition based on the original coupling degree and the perturbation coupling degree to obtain the coupling difference mechanism. Model construction and output module: This module is used to construct an evaluation model of vegetation-soil coupling effect based on the disturbance coupling effect and the coupling difference mechanism, input the data to be evaluated into the evaluation model of vegetation-soil coupling effect, and output the evaluation results.

Images