An ecological environment monitoring method and system based on ecological function data analysis

Abstract

The present application provides an ecological environment monitoring method and system based on ecological function data analysis, which relates to the field of big data analysis and environmental monitoring technology. The method includes: obtaining multidimensional parameter time series data corresponding to adjacent preset time periods of each monitoring area in the target area; for each multidimensional parameter time series data, processing the multidimensional parameter time series data based on a multiscale adaptive Bayesian network to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale; based on each comprehensive ecological function characteristic, constructing an ecological function graph structure, the ecological function graph structure includes multiple graph nodes and edge connections; updating the node characteristics of each graph node in the ecological function graph structure based on the spatiotemporal graph neural network, and generating an ecological state distribution map according to each updated node characteristic. Thus, a comprehensive assessment of the service functions of various ecosystems in the target area can be achieved, which can effectively support the refined management of ecological environment protection.

Description

Technical Field

[0001] The present application relates to the field of big data analysis and environmental monitoring technology, and in particular to an ecological environment monitoring method and system based on ecological function data analysis. Background Art

[0002] With the intensification of global climate change and the continuous impact of human activities on the natural environment, the health of the ecological environment has become the focus of global attention. Scientific research institutions and environmental protection organizations continue to strengthen the monitoring of the ecological environment in order to timely detect and respond to potential environmental crises.

[0003] Traditional monitoring methods mainly rely on remote sensing data or field sampling data. Although remote sensing data can provide a wide range of environmental information, its spatial and temporal resolution is limited and it cannot capture subtle ecological changes. Although field sampling data has high accuracy, it is limited by sampling frequency and sampling point distribution, and its data coverage is limited, making it difficult to reflect the overall status of the ecosystem.

[0004] In addition, ecosystem service functions are various beneficial services provided by ecosystems to human society, such as air purification, water purification, climate regulation, etc. However, in the current ecological environment monitoring, more attention is paid to the compliance of single physical indicators or chemical monitoring parameters, and there is a lack of monitoring of ecosystem functional indicators, which makes the monitoring results unable to fully reflect the actual functional status of regional ecosystems, resulting in insufficient evaluation of ecosystem service functions.

[0005] In response to the above problems, the industry has not yet proposed a better technical solution. Summary of the invention

[0006] The present application provides an ecological environment monitoring method, system, storage medium, computer program product and electronic device based on ecological function data analysis, which is used to at least solve the problem that the current ecological monitoring system cannot comprehensively monitor the actual functional status of the regional ecosystem.

[0007] In the first aspect, the embodiment of the present application provides an ecological environment monitoring method based on ecological function data analysis, including: obtaining multidimensional parameter time series data corresponding to adjacent preset time periods of each monitoring area in the target area; the multidimensional parameter time series data includes multiple adjacent historical time steps and corresponding multimodal sensor data; the multimodal sensor data includes SAR data, space meteorological parameters, water quality parameters and soil environmental parameters; for each multidimensional parameter time series data, processing the multidimensional parameter time series data based on a multiscale adaptive Bayesian network to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale; the comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity; based on each The comprehensive ecological function characteristics are used to construct an ecological function graph structure; the ecological function graph structure includes multiple graph nodes and edge connections; each of the graph nodes uniquely corresponds to a monitoring area, and the node characteristics of the graph nodes are defined by the corresponding comprehensive ecological function characteristics; the edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than a similarity threshold; based on the spatiotemporal graph neural network, the node characteristics of each graph node in the ecological function graph structure are updated, and an ecological status distribution map is generated according to each updated node characteristic; the ecological status distribution map describes the spatial distribution of ecological service indicators in the target area, and the ecological service indicators include carbon sequestration indicators, soil conservation indicators and water conservation indicators.

[0008] In the second aspect, the embodiment of the present application provides an ecological environment monitoring system based on ecological function data analysis, including: a data acquisition unit, used to acquire multidimensional parameter time series data corresponding to adjacent preset time periods of each monitoring area in the target area; the multidimensional parameter time series data includes multiple adjacent historical time steps and corresponding multimodal sensor data; the multimodal sensor data includes SAR data, space meteorological parameters, water quality parameters and soil environmental parameters; a feature extraction unit, used to process the multidimensional parameter time series data based on a multi-scale adaptive Bayesian network for each multidimensional parameter time series data, so as to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive space-time scale; the comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity; a graph structure construction A construction unit is used to construct an ecological function map structure based on each of the comprehensive ecological function characteristics; the ecological function map structure includes multiple graph nodes and edge connections; each of the graph nodes uniquely corresponds to a monitoring area, and the node characteristics of the graph nodes are defined by the corresponding comprehensive ecological function characteristics; the edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than a similarity threshold; an ecological map generation unit is used to update the node characteristics of each graph node in the ecological function map structure based on a spatiotemporal graph neural network, and generate an ecological status distribution map according to each updated node characteristic; the ecological status distribution map describes the spatial distribution of ecological service indicators in the target area, and the ecological service indicators include carbon sequestration indicators, soil conservation indicators and water conservation indicators.

[0009] According to a third aspect, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of the ecological environment monitoring method based on ecological function data analysis of any embodiment of the present application.

[0010] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, characterized in that when the program is executed by a processor, the steps of the ecological environment monitoring method based on ecological function data analysis of any embodiment of the present application are implemented.

[0011] In the fifth aspect, an embodiment of the present application provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the ecological environment monitoring method based on ecological function data analysis of any embodiment of the present application.

[0012] The ecological environment monitoring method based on ecological function data analysis provided by this application can produce at least the following technical effects:

[0013] (1) By processing multidimensional parameter time series data through a multiscale adaptive Bayesian network, it is possible to adaptively select the spatiotemporal scale and accurately extract comprehensive ecological function characteristics closely related to ecological functions, such as meteorological influencing factors, soil moisture content and net primary productivity. This not only makes up for the low resolution problem of traditional remote sensing data, but also provides more detailed and comprehensive ecological function information in combination with multimodal sensing data, so that the monitoring results can better reflect the actual functional status of the ecosystem. In addition, by processing multidimensional parameter time series data through a multiscale adaptive Bayesian network, it is possible to adaptively adjust the spatiotemporal scale and accurately extract comprehensive ecological function characteristics including meteorological influencing factors, soil moisture content and net primary productivity, so as to capture the subtle ecological changes of regional ecosystems in a more dynamic and comprehensive manner.

[0014] (2) By constructing an ecological function map structure based on comprehensive ecological function characteristics and using a spatiotemporal graph neural network to update the characteristics, an accurate ecological status distribution map can be generated. The edge weights can be used to reflect the similarity of ecological function characteristics in different monitoring areas, thereby realizing the spatiotemporal correlation analysis of the ecological status between regions. This can comprehensively evaluate ecological service indicators such as carbon sequestration, soil conservation, and water conservation in each monitoring area, and accurately reflect the spatial distribution of these indicators, thereby achieving a comprehensive evaluation of the ecosystem service function.

[0015] (3) Using spatiotemporal graph neural networks to update the node features of each monitoring area can dynamically capture the spatiotemporal correlation between different monitoring areas, thereby accurately updating the ecological function characteristics of each area. Through the dynamic feature update mechanism, the mutual influence between the ecological functions of each region can be revealed, providing more scientific data support for environmental decision-making.

[0016] Through this technical solution, the comprehensive application of multimodal sensor data, multi-scale adaptive Bayesian networks and spatiotemporal graph neural networks overcomes the problems of limited coverage and low frequency of traditional on-site sampling, realizes full-process intelligent processing from data processing, feature extraction to ecological status assessment, and supports efficient and real-time monitoring of the ecological environment over a large area. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0018] Figure 1A flowchart showing an example of an ecological environment monitoring method based on ecological function data analysis according to an embodiment of the present application is shown;

[0019] Figure 2 A schematic diagram of the structural connection of an example of a multi-scale adaptive Bayesian network according to an embodiment of the present application is shown;

[0020] Figure 3 An operation flow chart of an example of determining a target time scale and a target space scale in an adaptive manner according to an embodiment of the present application is shown;

[0021] Figure 4 A schematic diagram of the structural connection of an example of a spatiotemporal graph neural network according to an embodiment of the present application is shown;

[0022] Figure 5 A structural block diagram of an example of an ecological environment monitoring system based on ecological function data analysis according to an embodiment of the present application is shown;

[0023] Figure 6 It is a schematic structural diagram of an embodiment of an electronic device of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0026] Figure 1 A flowchart of an example of an ecological environment monitoring method based on ecological function data analysis according to an embodiment of the present application is shown.

[0027] Regarding the executor of the method of the embodiment of the present application, it can be any controller or processor with computing or processing capabilities. Through the innovative application of multi-scale adaptive Bayesian networks and spatiotemporal graph neural networks, comprehensive and accurate monitoring of ecosystem functional indicators is achieved, effectively improving the overall level of ecological environment monitoring, and providing more scientific and reliable technical support for responding to global climate change and protecting the ecological environment.

[0028] In some examples, it can be integrated and configured in an electronic device or terminal through software, hardware, or a combination of software and hardware, and the type of the terminal or electronic device can be diverse, such as a mobile phone, a tablet computer, or a desktop computer, etc.

[0029] like Figure 1 As shown, in step S110, multi-dimensional parameter time series data of corresponding adjacent preset time periods of each monitoring area in the target area is obtained.

[0030] Here, the multidimensional parameter time series data includes multiple adjacent historical time steps and corresponding multimodal sensor data, and the multimodal sensor data includes SAR (Synthetic Aperture Radar) data, space meteorological parameters, water quality parameters and soil environmental parameters.

[0031] In some embodiments, a variety of sensing devices are arranged in the target area to ensure that a sufficient spatial range and different environmental parameters are covered in the monitoring area. The sensing devices include synthetic aperture radars for acquiring SAR data, meteorological collectors for collecting space meteorological parameters, water monitoring sensors, and soil monitoring sensors.

[0032] SAR data is used to capture the reflective characteristics of the surface, reflecting macro-ecological information such as vegetation coverage and water body boundaries. Space meteorological parameters include ambient temperature, ambient humidity, precipitation, light intensity and wind speed, thus reflecting the impact of climate on the ecosystem. Water quality parameters include dissolved oxygen, pH value and pollutant concentration, thus reflecting the ecological status of the water body. Soil environmental parameters include soil nutrient content, soil temperature and soil moisture, thus reflecting the health of the soil.

[0033] Specifically, multimodal sensor data is collected according to the preset sampling window. For each monitoring area, the data is arranged on the time axis to generate a data sequence including multiple historical time steps, forming multidimensional parameter time series data, which not only includes the instantaneous value of each time step, but also considers their evolution relationship over time. By coordinating the layout of corresponding sensor networks in different monitoring areas of the target area, more comprehensive and accurate environmental data can be collected.

[0034] In step S120, for each multidimensional parameter time series data, the multidimensional parameter time series data is processed based on a multi-scale adaptive Bayesian network to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale.

[0035] Here, the comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity. Meteorological influencing factors represent the comprehensive characteristics of the ecosystem affected by meteorological factors (such as temperature, humidity, precipitation, wind speed, etc.), which are related to the water cycle and energy flow in the ecosystem, and can be extracted from spatial meteorological parameters. Soil moisture content represents the moisture content in the soil. It can be combined with soil moisture, temperature and precipitation to estimate the dynamic changes in soil moisture. Therefore, this feature will be affected to a certain extent by meteorological influencing factors. Net primary productivity (NPP) is the efficiency of plants fixing and converting photosynthetic products. It represents the productivity of vegetation and is affected by factors such as meteorological influencing factors, soil moisture content and light intensity.

[0036] In some embodiments, for each monitoring area, an adaptive Bayesian network is constructed using a multi-scale analysis method to process multidimensional parameter time series data at different time and space scales. In this way, based on the Multi-Scale Adaptive Bayesian Network (MSABN), the multidimensional parameter time series data is analyzed to extract the comprehensive ecological function characteristics related to the ecosystem function. The MSABN can process data at different time and space scales and adaptively select the most appropriate time and space scale to extract the characteristics that have a key impact on the ecological function, thereby fully considering the time dependence and spatial correlation of the data. Therefore, according to the ecological characteristics of different monitoring areas, the time and space scales of the Bayesian network are dynamically adjusted to ensure that the network can adaptively capture the key characteristics of the ecosystem at each time and space scale, which represents the health status of the ecosystem at different time and space scales.

[0037] In step S130, an ecological function graph structure is constructed based on each comprehensive ecological function feature.

[0038] Here, the ecological function graph structure includes multiple graph nodes and edge connections. Each graph node uniquely corresponds to a monitoring area. The node characteristics of the graph node are defined by the corresponding comprehensive ecological function characteristics. The edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than the similarity threshold.

[0039] Specifically, each monitoring area is mapped to a node in the ecological function graph structure, and the feature vector of each node is defined by the comprehensive ecological function characteristics of the area. By calculating the similarity between each graph node, for example, based on indicators such as Euclidean distance and cosine similarity to measure, the edge connection in the graph is constructed. Exemplarily, if the similarity between the comprehensive ecological function characteristics of two graph nodes exceeds a preset similarity threshold, an edge is added between the two graph nodes, and the weight of the edge is determined by the similarity value. In other words, in the ecological function graph structure, not all graph nodes will have edge connections, but only when the similarity exceeds a certain threshold will there be edge connections between nodes.

[0040] By constructing an ecological function map structure and integrating the ecological function characteristics of each monitoring area into a unified graph model, we can clearly express the similarities and correlations of the ecological functions between the regions. This can not only analyze the ecological functions of a single region, but also identify the mutual influence relationships between regions in the ecosystem.

[0041] In step S140, the node features of each graph node in the ecological function graph structure are updated based on the spatiotemporal graph neural network, and an ecological status distribution map is generated according to each updated node feature.

[0042] Through the Spatio-Temporal Graph Neural Network (ST-GNN), spatial dependency modeling and temporal dynamic modeling are performed on the basis of traditional GNN. In spatial dependency modeling, graph neural networks are used to capture the spatial dependency relationship between nodes in graph structure data, and the node representation is updated by aggregating the information of neighboring nodes to capture the local graph structure characteristics. In temporal dynamic modeling, the time series analysis method is combined to capture the changing trend of graph structure data over time, which can process sequence data and capture the dependency relationship between time steps. Furthermore, by fusing spatial dependency and temporal dynamics to simultaneously capture the complex patterns in the spatiotemporal graph, the spatiotemporal features of each node are fused to further extract high-dimensional spatiotemporal dependency information.

[0043] Furthermore, by analyzing the updated ecological function map structure, the ecological service indicators of each monitoring area are identified, including carbon sequestration indicators, soil conservation indicators and water conservation indicators. The carbon sequestration indicator reflects the role of ecosystems in climate regulation, supports biodiversity conservation, and provides a scientific basis for sustainable management. The soil conservation indicator can prevent land degradation, manage water resources, reduce disaster risks, and maintain the stability and productivity of the ecosystem. The water conservation indicator can ensure water supply, mitigate flood disasters, support diversified ecosystem services, and maintain the overall health of the ecosystem. The generated ecological status distribution map describes the spatial distribution of ecological service indicators in the target area, provides an intuitive spatial decision support tool for ecological protection and management, and enables decision makers to better formulate targeted ecological management strategies.

[0044] In some examples of the embodiments of the present application, the node features after ST-GNN update are the comprehensive ecological function feature vectors of each monitoring area at the real-time time step, which contain data in multiple dimensions, reflecting the spatiotemporal dependencies between nodes and the dynamic changes of ecological function characteristics within the region.

[0045] Specifically, in order to generate an ecological status distribution map, node features need to be mapped to specific ecological service indicators. Each ecological service indicator needs to be extracted from the node features through a set of specific mapping functions. These functions can be linear regression models, nonlinear functions, or algorithms based on empirical rules, and the specific form depends on the relationship between each ecological service indicator and the node characteristics.

[0046] For example, the carbon sequestration index can be obtained by weighted calculation of net primary productivity and meteorological influence factors, the soil conservation index can be obtained by weighted calculation of soil moisture content and vegetation coverage, and the water conservation index can be obtained by weighted calculation of meteorological influence factors and soil moisture content. Furthermore, for each graph node (i.e., each monitoring area), the above-mentioned mapping function is used to calculate the various ecological service indicators of the area. After the calculation is completed, the ecological service indicator values ​​corresponding to all graph nodes are plotted on the geographic space of the target area to generate an ecological status distribution map, and the ecological service indicators of each monitoring area are expressed in the form of colors, layers or contour lines, etc., to provide a visual overview of the ecological status. Furthermore, each ecological service indicator can be weighted and superimposed according to the time step to generate a comprehensive ecological status distribution map.

[0047] Through the embodiments of the present application, the node features updated by ST-GNN can be effectively converted into an ecological service indicator distribution map that can actually be used for ecological management and decision support. By considering the spatiotemporal dynamic changes of node features, the generated ecological status distribution map can more accurately reflect the ecological service functions in the region, especially in areas with climate change and frequent human activities. With the continuous updating of ST-GNN, the ecological status distribution map can dynamically reflect the current state of the ecosystem, and spatial visualization enables managers to intuitively understand the ecological function status in the region.

[0048] Figure 2 A schematic diagram of the structural connection of an example of a multi-scale adaptive Bayesian network according to an embodiment of the present application is shown.

[0049] like Figure 2 As shown, the multi-scale adaptive Bayesian network 200 includes an input layer 210 , an adaptive scale resampling layer 220 , a conditional probability inference layer 230 , and an output layer 240 .

[0050] The input layer 210 is used to receive multi-dimensional parameter time series data.

[0051] The adaptive scale resampling layer 220 is used to determine the corresponding target time scale and target space scale according to the spatiotemporal variation characteristics of the multidimensional parameter time series data, and resample the multidimensional parameter time series data based on the target time scale and target space scale to obtain the corresponding resampled time series data.

[0052] It should be noted that ecological and environmental data usually have different temporal and spatial resolutions. For example, SAR data may have high spatial resolution but low temporal resolution; meteorological data may be the opposite. Therefore, through the adaptive scale selection layer, the most appropriate temporal and spatial scales are dynamically determined according to the temporal and spatial characteristics of the input data, so as to align the data to a unified scale for subsequent Bayesian inference analysis.

[0053] Specifically, the time scale determines the temporal resolution of the data. Smaller time scales represent higher temporal resolution and are suitable for capturing rapidly changing ecological processes, while larger time scales are suitable for long-term trend analysis. The spatial scale determines the spatial resolution of the data. Smaller spatial scales represent higher spatial resolution and are suitable for capturing subtle spatial heterogeneity, while larger spatial scales are suitable for regional trend analysis.

[0054] The conditional probability reasoning layer 230 is used to calculate the posterior probability distribution corresponding to each Bayesian node in the Bayesian network based on the resampled time series data and prior knowledge.

[0055] The Bayesian network includes multiple Bayesian nodes and multiple directed edges, each Bayesian node is used to represent the corresponding ecological function characteristic variable, and the directed edge represents the conditional dependency between the ecological function characteristic variables indicated by the connected Bayesian nodes. Exemplarily, corresponding graph nodes are set for meteorological influencing factors, soil moisture content, and net primary productivity, respectively, and the conditional dependency relationship can be pre-set according to the natural association relationship between the ecological function characteristic variables, such as based on scientific understanding of the ecosystem and expert knowledge, so as to reflect the causal relationship or correlation between different ecological variables. For example, if it is known that meteorological conditions (such as precipitation) will affect soil moisture content, then a directed edge from the "meteorological influencing factor" node to the "soil moisture content" node will be set in the Bayesian network. By modeling and reasoning these conditional dependencies, the system can more accurately predict key ecological function characteristics (such as soil moisture content and NPP), thereby improving the accuracy of the overall monitoring system.

[0056] In the multi-scale adaptive Bayesian network, the main task of the conditional probability inference layer is to calculate the posterior probability of specific ecological function characteristics based on the input data. Through known observation data and prior knowledge, the posterior probability distribution of intermediate characteristics representing the health status of the ecosystem can be extracted from the input multi-dimensional spatiotemporal data. Specifically, the posterior probability of each ecological function characteristic is calculated layer by layer through the Bayesian formula to reflect the most likely state of each ecological function characteristic (meteorological influencing factor, soil moisture content and net primary productivity) under the current spatiotemporal conditions.

[0057] The output layer 240 is used to calculate the expected values ​​of the corresponding ecological function characteristic variables according to the posterior probability distribution corresponding to each Bayesian node, and obtain the comprehensive ecological function characteristics through combination.

[0058] Here, the posterior probability distribution output by the Bayesian network actually represents the possibility of a certain feature taking a certain value under given observation data. Then, by calculating the mathematical expectation of the probability distribution as a quantitative representation of the feature, the corresponding ecological functional characteristics are obtained.

[0059] Figure 3 An operational flowchart of an example of determining a target time scale and a target spatial scale in an adaptive manner according to an embodiment of the present application is shown.

[0060] like Figure 3 As shown, in step S310, the multi-scale characteristics of the multi-dimensional parameter time series data are analyzed based on the wavelet transform method to determine the corresponding target time scale.

[0061] Here, a wavelet transform-based method is used to analyze the multi-scale characteristics of time series data, which can capture the changes of time series data at different frequencies, thereby flexibly selecting the time scale and adapting to the multi-scale characteristics of the data.

[0062] Specifically, the target time scale is calculated as follows:

[0063] , Formula (1)

[0064] , Formula (2)

[0065] In the formula, are the coefficients of the wavelet transform; t is the time parameter, which indicates the position of the wavelet function on the time axis; a is the scale parameter, which is used to control the degree of expansion and contraction of the wavelet function; represents the scale parameter used to find the expression that maximizes a ; To represent the integral time variable, it is used to integrate the signal at any time on the time axis. It is a multidimensional parameter time series data in The value of the moment; is the complex conjugate of the wavelet function; is the time series length corresponding to the multidimensional parameter time series data, It is the target time scale adaptively selected based on wavelet energy maximization.

[0066] Here, by performing wavelet transform on the input time series data, we can The changes in the data are observed, and the target time scale is determined based on the wavelet energy maximization criterion, which can determine the time scale that can best reflect the characteristics of time series data.

[0067] In step S320, the spatial distribution density of the data points corresponding to the multi-dimensional parameter time series data is calculated, and the corresponding target spatial scale is determined according to the spatial distribution density.

[0068] Regarding the selection of spatial scale, a variational adaptive grid generation method is adopted. By minimizing the spatial gradient changes of the data, the optimal spatial scale is determined, and a spatial grid that is more in line with the terrain and ecological characteristics can be adaptively generated.

[0069] Specifically, the target spatial scale is calculated by the following formula:

[0070] , Formula (3)

[0071] , Formula (4)

[0072] In the formula, It's location The spatial distribution density of It's location The neighborhood set of nearby data points, is the position of any data point in the neighborhood set, is a neighborhood set The number of data points in ; is the spatial influence radius; is the target spatial scale, which is used to indicate the corresponding sampling density; Represents the scaling constant.

[0073] Here, spatial distribution density is used to describe the distribution of spatial data points. By calculating the spatial distribution density of data points in their neighborhood, the concentration of data points in space can be identified. Adaptively adjust the target spatial scale according to the spatial distribution density, and apply higher spatial resolution (smaller) in dense data areas. ), apply lower spatial resolution (larger ) to optimize the spatial sampling of the data.

[0074] In the embodiment of the present application, through wavelet transform, the system can analyze ecological data at different time scales, making the monitoring of long-term trends and short-term fluctuations more accurate, which helps to fully understand the dynamic changes of the ecosystem. In addition, the spatial distribution in the ecological environment is often heterogeneous. Through adaptive analysis of the spatial distribution density, higher spatial resolution can be applied in high-density areas (such as cities, lakes and other complex environments), thereby improving the accuracy of data sampling, and reducing the resolution in low-density areas to reduce unnecessary computational overhead. Therefore, under different environmental conditions and data characteristics, the analysis scale is adaptively adjusted. By optimizing the time and space scales, the system can reduce unnecessary computing resource consumption while ensuring accuracy, ensure the stability and robustness of the system, and support effective application in various complex ecological monitoring scenarios.

[0075] In some examples of the embodiments of the present application, resampling is performed based on the target time scale in the following manner to obtain corresponding time resampled data:

[0076] Get the sensing sampling time scale and detect whether the target time scale is an integer multiple of the sensing sampling time scale. If so, downsample the multidimensional parameter time series data directly according to the target time scale; and, if not, use the Gaussian kernel weight interpolation method to time resample the multidimensional parameter time series data:

[0077] , Formula (5)

[0078] , Formula (6)

[0079] In the formula, At the interpolation time point The time-resampled data at is the first time point with known data points, N is the total number of known data points in the multidimensional parameter time series data, At a known time The multidimensional parameter value at ; It is known data point pairs The interpolation weight of is the bandwidth parameter, which is used to control the distribution range of weights; is the normalization factor, which represents the sum of the Gaussian kernel function values ​​for all known data points in the multidimensional parameter time series data.

[0080] Here, the weights are calculated based on the Gaussian kernel function. , the weight decays exponentially with the square of the time distance. If the interpolation time point (or target time point) Close to a known data point ,but is small, so in the inverse distance weighting method, In the Gaussian kernel weighting method, the weights will also increase accordingly. The data is weighted summed to obtain the interpolated value.

[0081] It should be noted that ecological environment monitoring data usually have smoothness, continuity and noise anomalies. Environmental data usually change relatively slowly and continuously in a short period of time, so the interpolation method needs to be able to smoothly reflect this change. In addition, there may be some noise or outliers in the ecological environment monitoring data, so the interpolation method needs to have a certain degree of noise resistance.

[0082] In view of this, the Gaussian kernel weight method can process data more smoothly by exponentially decaying the square of the distance, avoiding sudden changes in interpolation results caused by noise or individual abnormal data points. , it can ensure that only the closer data points have a significant impact on the interpolation results, while the farther data points have a smaller impact, which is consistent with the characteristics of ecological data and achieves relatively stable changes in interpolation data within a short time range. In addition, since the impact of data far away from the target point on the interpolation results will be significantly weakened, the Gaussian kernel weight method is more robust when dealing with noisy data.

[0083] Through the embodiments of the present application, the Gaussian kernel weight interpolation method is used in the time resampling process, which can accurately adjust the time resolution without destroying the smoothness and continuity of the data. The Gaussian kernel weight method can flexibly deal with the problem of uneven distribution of data points when processing time series data, ensuring the consistency of data on the time axis.

[0084] In addition, resampling is performed based on the target spatial scale in the following way to obtain the corresponding spatial resampled data:

[0085] Get the sensor sampling space scale and detect whether the target space scale is an integer multiple of the sensor sampling space scale. If so, downsample the multidimensional parameter time series data directly according to the target space scale; and, if not, use the graph Laplace interpolation method to spatially resample the multidimensional parameter time series data:

[0086] , Formula (7)

[0087] In the formula, is the position in the interpolation space The spatial resampled data at is the graph Laplacian matrix eigenvalues, U is the total number of eigenvalues ​​in the graph Laplacian matrix; is the eigenvalue corresponding to In the interpolation space The feature vector of .

[0088] Here, by constructing a graph structure of spatial data, a graph Laplacian matrix is ​​generated to capture the topological relationship of spatial data. Then, the spatial data is smoothed and interpolated by the graph Laplacian operator to generate resampled data at the target spatial scale. The graph Laplacian interpolation method can reduce the errors caused by sparse or irregular distribution of spatial data in spatial interpolation by considering spatial relationships, thereby improving the accuracy of the overall spatial data.

[0089] Through the embodiments of the present application, the smoothing of spatial data by the graph Laplace operator can better preserve the structural characteristics of spatial data, while achieving efficient spatial interpolation and resampling. In the process of spatial resampling, the use of the graph Laplace interpolation method can maintain the structural characteristics of spatial data, especially for areas with complex geographical features or ecological structures, and can effectively maintain the integrity and rationality of spatial distribution.

[0090] The time resampled data and the space resampled data are fused to obtain the corresponding resampled time series data:

[0091] , Formula (8)

[0092] In the formula, Indicates the resampled time series data at the target time and space position The value at and Represent the temporal resampling weight and spatial resampling weight respectively.

[0093] Through reasonable time and space resampling, it is possible to reduce unnecessary computational overhead while ensuring data accuracy. Especially when the target time scale and space scale are not integer multiples of the original data scale, resampling through Gaussian kernel weighting method and graph Laplace interpolation method can avoid processing a large amount of redundant data, which not only improves the computational efficiency of the system, but also reduces the energy and resource consumption of the system, making the ecological environment monitoring system more efficient in long-term operation.

[0094] In one example of an embodiment of the present application, the similarity threshold selected for edge connections in constructing an ecological function graph structure may be pre-set. It should be noted that ecosystems often exhibit a high degree of heterogeneity, and there may be significant differences between different regions. In another example of an embodiment of the present application, an adaptive adjustment mechanism may be adopted. Through an adaptive similarity threshold, the system is allowed to flexibly adjust the connection criteria according to the environmental characteristics and ecological function characteristics of a specific area, so that the system can flexibly adjust the connection strength between nodes according to the dynamic changes of ecological characteristics and the environmental characteristics of different regions, thereby improving the accuracy of the graph structure and the accuracy of ecological monitoring.

[0095] More specifically, it can be determined by:

[0096] , Formula (9)

[0097] In the formula, Graph Node and The characteristic distance between and The nodes are and In the The value of the feature dimension, M is the dimension of the feature vector.

[0098] , Formula (10)

[0099] , formula (11)

[0100] In the formula, Q is the total number of graph nodes in the ecological function graph structure, and They represent the mean and standard deviation of the characteristic distances between all node pairs in the ecological function graph structure. Express satisfaction for all i < j The distance between each pair of graph nodes is calculated only once, thus avoiding repeated calculations, and the characteristic distances of all non-duplicate graph nodes are summed.

[0101] Calculate the similarity threshold corresponding to each pair of graph nodes in the ecological function graph structure:

[0102] , Formula (12)

[0103] In the formula, Representing a graph node and The similarity threshold between and is the similarity adjustment factor.

[0104] Here, the similarity threshold The calculation depends on the node and By combining the statistical characteristics of the characteristic distance between nodes and the mean and standard deviation of the characteristic distance, the adaptive similarity threshold can better balance the problem of connecting too many or too few nodes, thereby enhancing the robustness and representativeness of the graph structure. In addition, as the ecosystem changes dynamically, the adaptive threshold mechanism can continuously adjust the threshold setting to maintain the real-time update and accuracy of the graph structure, providing support for the continuous monitoring of the ecological status.

[0105] Through the embodiments of the present application, the adaptive similarity threshold dynamically adjusts the threshold according to the ecological function characteristics of different monitoring areas (such as meteorological influencing factors, soil moisture content, net primary productivity, etc.). With the dynamic changes of the ecosystem (such as seasonal changes, extreme weather events, etc.), the distribution of ecological function characteristics may change significantly. Therefore, through the adaptive similarity threshold mechanism, the connection standards between nodes can be adjusted in real time according to the latest monitoring data, so that the graph structure always reflects the current ecological status, thereby providing more reliable support for real-time ecological monitoring.

[0106] Figure 4 A schematic diagram of the structural connection of an example of a spatiotemporal graph neural network according to an embodiment of the present application is shown.

[0107] like Figure 4 As shown, the spatiotemporal graph neural network 400 includes an input layer 410, a multi-scale temporal attention module 420, a spatial graph convolution layer 430, a spatiotemporal fusion layer 440 and an output layer 450.

[0108] The input layer 410 is used to receive the updated node features corresponding to each graph node to determine the corresponding graph node feature matrix:

[0109] , Formula (13)

[0110] In the formula, Representing a graph node At time step The characteristic vector of Indicates that at time step The feature matrix of all graph nodes.

[0111] The multi-scale temporal attention module 420 is used to extract multi-scale features of the graph node feature matrix corresponding to the time dimension to determine the corresponding temporal attention weights:

[0112] , Formula (14)

[0113] , Formula (15)

[0114] , Formula (16)

[0115] In the formula, Indicates that at time step Corresponding to The temporal attention weights of time scales, Indicates that at time step Corresponding to The query vector at each time scale is for The vector transpose of ; Indicates that at time step Corresponding to The key vector at each time scale is Indicates The weight matrix of the query vector at each time scale, Indicates The weight matrix of the key vector at each time scale, Indicates the dimension of the key vector.

[0116] It should be noted that the importance of ecological function characteristics (such as meteorological factors, soil moisture content, etc.) may vary at different time scales. This enables the network to focus on more relevant time scales.

[0117] The spatial graph convolution layer 430 is used to extract the features of the spatial dimension to determine the corresponding spatial graph convolution result:

[0118] , Formula (17)

[0119] In the formula, Is a graph node The set of nodes in the neighborhood graph of is any graph node in the set of neighborhood graph nodes at time step The characteristic vector of is the weight matrix in the graph convolution operation, Is a graph node At time step Spatial features after graph convolution.

[0120] Here, the multi-scale temporal attention mechanism allows the network to dynamically adjust the attention weights according to changes at different time scales, while the spatial graph convolution layer effectively integrates the spatial neighborhood information, enabling ST-GNN to adaptively cope with the non-stationarity and complexity in the ecosystem, and enhancing the robustness of the model when dealing with data noise or incomplete data.

[0121] Through the embodiments of the present application, by integrating the multi-scale temporal attention mechanism and the spatial graph convolution layer, ST-GNN can dynamically adjust the attention to node features at different time scales, while integrating the spatial dependencies between nodes, so that the network can more accurately capture the spatiotemporal changes of ecological functional characteristics in the monitoring area, thereby effectively reflecting the complex dynamics of the ecosystem.

[0122] The spatiotemporal fusion layer 440 is used to fuse the temporal attention weights and the spatial graph convolution results to generate spatiotemporal fusion features:

[0123] , Formula (18)

[0124] In the formula, Representing a graph node At time step The updated feature vector, Represents the weight matrix for linear transformation of node features. L Represents the total number of time scales.

[0125] In the embodiment of the present application, through the multi-scale time attention mechanism, ST-GNN can weightedly integrate node features at different time scales, so that the model can capture the changing characteristics of the ecosystem at different time scales, such as short-term meteorological fluctuations and long-term ecological succession, etc. It can not only reflect the node status at the current time step, but also consider the comprehensive impact of the past period of time. The features of not only contain their own information, but also integrate the features of their neighborhood graph nodes, enhancing their expressiveness in the spatial dimension and helping to describe the mutual influence between different regions in the ecosystem. The node features after spatiotemporal fusion can more accurately describe the status of each monitoring area in the ecosystem, not only considering the spatial relationship between graph nodes, but also integrating the dynamic changes in the time dimension.

[0126] The output layer 450 is used to update the feature matrix of the output graph nodes according to each generated spatiotemporal fusion feature:

[0127] , Formula (19)

[0128] In the formula, Indicates that at time step The graph nodes corresponding to the ecological function graph structure update the feature matrix.

[0129] Through the embodiments of the present application, using the multi-scale time attention mechanism, ST-GNN can extract subtle changes in ecological characteristics at different time scales, while spatial graph convolution extracts the mutual influence between nodes in different spatial ranges, which can effectively integrate large-scale spatiotemporal data. Based on multi-level feature extraction, the network can deeply understand the operating mechanism of the ecosystem from both global and local levels. As a result, the network's adaptive adjustment ability in the spatiotemporal dimension improves the efficiency of data processing, speeds up data processing and analysis, enables the ecological monitoring system to respond to environmental changes more quickly, and provides a basis for the final real-time and accurate generation of ecological status distribution maps for the target area.

[0130] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of actions combined, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0131] Figure 5 A structural block diagram of an example of an ecological environment monitoring system based on ecological function data analysis according to an embodiment of the present application is shown.

[0132] like Figure 5As shown, the ecological environment monitoring system 500 based on ecological function data analysis includes a data acquisition unit 510, a feature extraction unit 520, a graph structure construction unit 530 and an ecological graph generation unit 540.

[0133] The data acquisition unit 510 is used to obtain multidimensional parameter time series data corresponding to adjacent preset time periods of each monitoring area in the target area; the multidimensional parameter time series data includes multiple adjacent historical time steps and corresponding multimodal sensing data; the multimodal sensing data includes SAR data, space meteorological parameters, water quality parameters and soil environmental parameters.

[0134] The feature extraction unit 520 is used to process each multidimensional parameter time series data based on a multi-scale adaptive Bayesian network to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale; the comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity.

[0135] The graph structure construction unit 530 is used to construct an ecological function graph structure based on each of the comprehensive ecological function characteristics; the ecological function graph structure includes multiple graph nodes and edge connections; each of the graph nodes uniquely corresponds to a monitoring area, and the node characteristics of the graph node are defined by the corresponding comprehensive ecological function characteristics; the edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than a similarity threshold.

[0136] The ecological map generation unit 540 is used to update the node characteristics of each graph node in the ecological function map structure based on the spatiotemporal graph neural network, and generate an ecological status distribution map according to each updated node characteristic; the ecological status distribution map describes the spatial distribution of ecological service indicators in the target area, and the ecological service indicators include carbon sequestration indicators, soil conservation indicators and water conservation indicators.

[0137] In some embodiments, an embodiment of the present application provides a non-volatile computer-readable storage medium, which stores one or more programs including execution instructions, and the execution instructions can be read and executed by an electronic device (including but not limited to a computer, a server, or a network device, etc.) to execute the steps of any of the above-mentioned ecological environment monitoring methods based on ecological function data analysis in the present application.

[0138] In some embodiments, the embodiments of the present application also provide a computer program product, which includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes any step of the above-mentioned ecological environment monitoring method based on ecological function data analysis.

[0139] In some embodiments, the embodiments of the present application also provide an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the steps of an ecological environment monitoring method based on ecological function data analysis.

[0140] Figure 6 is a hardware structure diagram of an electronic device for executing an ecological environment monitoring method based on ecological function data analysis provided by another embodiment of the present application, such as Figure 6 As shown, the device includes:

[0141] One or more processors 610 and memory 620, Figure 6 A processor 610 is taken as an example.

[0142] The device for executing the ecological environment monitoring method based on ecological function data analysis may further include: an input device 630 and an output device 640 .

[0143] The processor 610, the memory 620, the input device 630 and the output device 640 may be connected via a bus or other means. Figure 6 The example of connecting through bus is taken in the following.

[0144] The memory 620, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the program instructions / modules corresponding to the ecological environment monitoring method based on ecological function data analysis in the embodiment of the present application. The processor 610 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 620, that is, the ecological environment monitoring method based on ecological function data analysis in the above method embodiment is realized.

[0145] The memory 620 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 620 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 620 may optionally include a memory remotely arranged relative to the processor 610, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0146] The input device 630 can receive input digital or character information and generate signals related to user settings and function control of the electronic device. The output device 640 can include a display device such as a display screen.

[0147] The one or more modules are stored in the memory 620, and when executed by the one or more processors 610, the ecological environment monitoring method based on ecological function data analysis in any of the above method embodiments is executed.

[0148] The above-mentioned product can execute the method provided in the embodiment of the present application, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the method provided in the embodiment of the present application.

[0149] The electronic devices of the embodiments of the present application exist in various forms, including but not limited to:

[0150] (1) Mobile communication equipment: This type of equipment is characterized by having mobile communication functions and its main purpose is to provide voice and data communications. This type of terminal includes: smart phones, multimedia phones, functional phones, and low-end phones.

[0151] (2) Ultra-mobile personal computer devices: These devices belong to the category of personal computers, have computing and processing functions, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, etc.

[0152] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players, handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0153] (4) Other onboard electronic devices with data interaction functions, such as on-board devices installed in vehicles.

[0154] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0155] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An ecological environment monitoring method based on ecological function data analysis, comprising: Obtain multi-dimensional parameter time series data of corresponding adjacent preset time periods of each monitoring area in the target area; The multidimensional parameter time series data includes a plurality of adjacent historical time steps and corresponding multimodal sensing data; the multimodal sensing data includes SAR data, space meteorological parameters, water quality parameters and soil environmental parameters; For each multidimensional parameter time series data, the multidimensional parameter time series data is processed based on a multiscale adaptive Bayesian network to extract the comprehensive ecological function characteristics corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale; The comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity; Based on each of the comprehensive ecological function characteristics, an ecological function graph structure is constructed; the ecological function graph structure includes a plurality of graph nodes and edge connections; each of the graph nodes uniquely corresponds to a monitoring area, and the node characteristics of the graph nodes are defined by the corresponding comprehensive ecological function characteristics; the edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than a similarity threshold; Based on the spatiotemporal graph neural network, the node features of each graph node in the ecological function graph structure are updated, and an ecological state distribution map is generated according to each updated node feature; the ecological state distribution map describes the spatial distribution of ecological service indicators in the target area, and the ecological service indicators include carbon sequestration indicators, soil conservation indicators and water conservation indicators; Wherein, the multi-scale adaptive Bayesian network includes an input layer, an adaptive scale resampling layer, a conditional probability inference layer and an output layer; The input layer is used to receive multi-dimensional parameter time series data; The adaptive scale resampling layer is used to determine the corresponding target time scale and target space scale according to the spatiotemporal variation characteristics of the multidimensional parameter time series data, and resample the multidimensional parameter time series data based on the target time scale and the target space scale to obtain corresponding resampled time series data; The conditional probability reasoning layer is used to calculate the posterior probability distribution corresponding to each Bayesian node in the Bayesian network according to the resampled time series data and prior knowledge; the Bayesian network includes multiple Bayesian nodes and multiple directed edges; each Bayesian node is used to represent the corresponding ecological function characteristic variable, and the directed edge represents the conditional dependency relationship between the ecological function characteristic variables indicated by the connected Bayesian nodes; The output layer is used to calculate the expected values ​​of the corresponding ecological function characteristic variables according to the posterior probability distribution corresponding to each Bayesian node, and obtain the comprehensive ecological function characteristics through combination; Wherein, determining the corresponding target time scale and target space scale according to the spatiotemporal variation characteristics of the multidimensional parameter time series data includes: The multi-scale characteristics of the multi-dimensional parameter time series data are analyzed based on the wavelet transform method to determine the corresponding target time scale: ， ， In the formula, are the coefficients of the wavelet transform; t is the time parameter, which indicates the position of the wavelet function on the time axis; a is the scale parameter, which is used to control the degree of expansion and contraction of the wavelet function; represents the scale parameter used to find the expression that maximizes a ; To represent the integral time variable, it is used to integrate the signal at any time on the time axis. It is a multidimensional parameter time series data. The value of the moment; is the complex conjugate of the wavelet function; is the time series length corresponding to the multidimensional parameter time series data, It is the target time scale adaptively selected based on wavelet energy maximization; Calculate the spatial distribution density of the data points corresponding to the multidimensional parameter time series data, and determine the corresponding target spatial scale according to the spatial distribution density: ， ， In the formula, It's location The spatial distribution density of It's location The neighborhood set of nearby data points, is the position of any data point in the neighborhood set, is a neighborhood set The number of data points in ; is the spatial influence radius; is the target spatial scale, which is used to indicate the corresponding sampling density; Represents the scaling constant.

2. The method according to claim 1, wherein: The resampling of the multidimensional parameter time series data based on the target time scale and the target spatial scale to obtain corresponding resampled time series data includes: Resampling is performed based on the target time scale in the following manner to obtain corresponding time resampled data: Acquire a sensing sampling time scale, and detect whether the target time scale is an integer multiple of the sensing sampling time scale; If so, directly down-sample the multidimensional parameter time series data according to the target time scale; and If not, the Gaussian kernel weight interpolation method is used to perform time resampling on the multidimensional parameter time series data: ， ， In the formula, At the interpolation time point The time-resampled data at is the first time point with known data points, N is the total number of known data points in the multidimensional parameter time series data, At a known time Multidimensional parameter values ​​at ; It is known data point pairs The interpolation weight of is the bandwidth parameter, which is used to control the distribution range of weights; is the normalization factor, which represents the sum of the Gaussian kernel function values ​​for all known data points in the multidimensional parameter time series data; Based on the target spatial scale, resampling is performed in the following manner to obtain corresponding spatial resampled data: Acquire the sensing sampling space scale, and detect whether the target space scale is an integer multiple of the sensing sampling space scale; If yes, downsampling the multidimensional parameter time series data directly according to the target spatial scale; and If not, the multidimensional parameter time series data is spatially resampled using the graph Laplace interpolation method: ， In the formula, is the position in the interpolation space The spatial resampled data at is the graph Laplacian matrix eigenvalues, U is the total number of eigenvalues ​​in the graph Laplacian matrix; is the eigenvalue corresponding to In the interpolation space The eigenvector of The time resampled data and the space resampled data are fused to obtain the corresponding resampled time series data: ， In the formula, Indicates the resampled time series data at the target time and space position The value at and Represent the temporal resampling weight and spatial resampling weight respectively.

3. The method according to claim 1, wherein: The similarity threshold adopts an adaptive adjustment mechanism and is determined in the following manner: ， In the formula, Graph Node and The characteristic distance between and The nodes are and In the The value of the feature dimension, M is the dimension of the feature vector; ， ， In the formula, Q is the total number of graph nodes in the ecological function graph structure, and They represent the mean and standard deviation of the characteristic distances between all pairs of nodes in the ecological function graph structure; Satisfaction with all i < j Sum the graph node pairs of ; Calculate the similarity threshold corresponding to each pair of graph nodes in the ecological function graph structure: ， In the formula, Representing a graph node and The similarity threshold between and is the similarity adjustment factor.

4. The method according to claim 3, wherein: The spatiotemporal graph neural network comprises an input layer, a multi-scale temporal attention module, a spatial graph convolution layer, a spatiotemporal fusion layer and an output layer; The input layer is used to receive the updated node features corresponding to each graph node to determine the corresponding graph node feature matrix: ， In the formula, Representing a graph node At time step The characteristic vector of Indicates that at time step The feature matrix of all graph nodes; The multi-scale temporal attention module is used to extract the multi-scale features of the graph node feature matrix corresponding to the time dimension to determine the corresponding temporal attention weights: ， ， ， In the formula, Indicates that at time step Corresponding to The temporal attention weights of the time scale, Indicates that at time step Corresponding to The query vector at each time scale is for The vector transpose of ; Indicates that at time step Corresponding to The key vector at each time scale is Indicates The weight matrix of the query vector at each time scale, Indicates The weight matrix of the key vector at each time scale, represents the dimension of the key vector; The spatial graph convolution layer is used to extract the features of the spatial dimension to determine the corresponding spatial graph convolution result: ， In the formula, Is a graph node The set of nodes in the neighborhood graph of is any graph node in the set of neighborhood graph nodes at time step The characteristic vector of is the weight matrix in the graph convolution operation, Is a graph node At time step Spatial features after graph convolution; The spatiotemporal fusion layer is used to fuse the temporal attention weights and the spatial graph convolution results to generate spatiotemporal fusion features: ， In the formula, Representing a graph node At time step The updated feature vector, Represents the weight matrix for linear transformation of node features. L represents the total number of time scales; The output layer is used to update the feature matrix of the output graph nodes according to each generated spatiotemporal fusion feature: ， In the formula, Indicates that at time step The graph nodes corresponding to the ecological function graph structure update the feature matrix.

5. The method according to any one of claims 1 to 4, wherein: The space meteorological parameters include ambient temperature, ambient humidity, precipitation, light intensity and wind speed; the water quality parameters include dissolved oxygen, pH value and pollutant concentration; the soil environmental parameters include soil nutrient content, soil temperature and soil humidity.

6. An ecological environment monitoring system based on ecological function data analysis, used to implement the ecological environment monitoring method based on ecological function data analysis as described in any one of claims 1 to 5, the system comprising: A data acquisition unit, used to acquire multi-dimensional parameter time series data corresponding to adjacent preset time periods of each monitoring area in the target area; The multidimensional parameter time series data includes a plurality of adjacent historical time steps and corresponding multimodal sensing data; the multimodal sensing data includes SAR data, space meteorological parameters, water quality parameters and soil environmental parameters; A feature extraction unit, for processing each multidimensional parameter time series data based on a multiscale adaptive Bayesian network, so as to extract the comprehensive ecological function features corresponding to the multidimensional parameter time series data according to the adaptive spatiotemporal scale; The comprehensive ecological function characteristics include meteorological influencing factors, soil moisture content and net primary productivity; A graph structure construction unit, for constructing an ecological function graph structure based on each of the comprehensive ecological function characteristics; the ecological function graph structure comprises a plurality of graph nodes and edge connections; each of the graph nodes uniquely corresponds to a monitoring area, and the node characteristics of the graph nodes are defined by the corresponding comprehensive ecological function characteristics; the edge weight of each edge connection is determined by the similarity between the corresponding comprehensive ecological function characteristics of the connected graph nodes that is greater than a similarity threshold; An ecological map generation unit is used to update the node characteristics of each graph node in the ecological function map structure based on the spatiotemporal graph neural network, and generate an ecological status distribution map according to each updated node characteristic; the ecological status distribution map describes the spatial distribution of ecological service indicators in the target area, and the ecological service indicators include carbon sequestration indicators, soil conservation indicators and water conservation indicators.

Images