Method for monitoring and checking earth surface ecological diversity through satellite remote sensing

Abstract

The invention discloses a method for monitoring and checking surface ecological diversity through satellite remote sensing, which belongs to the technical field of image processing and comprises the steps of data collection and preprocessing, ecological parameter calculation, interpretation and classification, dynamic change analysis and result output. The technical problems of inconsistent data processing, insufficient dynamic change analysis and low classification precision in earth surface ecological diversity monitoring are solved, the consistency and high quality of data are ensured, the dynamic characteristics of an ecological system can be comprehensively reflected, the change trend of the ecological system is quantified, accurate reference basis is provided for ecological protection, and the method is suitable for popularization and application. And the classification accuracy of the complex ecosystem is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a method for satellite remote sensing monitoring and verifying surface ecological diversity. Background Art

[0002] Biodiversity is the basis of ecosystem productivity, the cornerstone of the stable operation of the earth's ecosystem, and the basis for maintaining the ecological balance of the earth, ensuring human well-being and the sustainable development of nature. Biodiversity covers the richness and variability of all life forms on the earth. Diverse organisms are interdependent and mutually restrictive, and together maintain the stability and function of the ecosystem.

[0003] Although traditional ground survey methods have a certain degree of accuracy at the microscopic scale, they are inefficient and have limited coverage, making it difficult to meet the needs of large-scale, multi-temporal dynamic monitoring. In recent years, ecological monitoring technology based on remote sensing images has been widely used. However, existing technologies still have shortcomings in the following aspects: Insufficient data processing: Traditional remote sensing image analysis methods rely on a single data source and find it difficult to integrate spatiotemporal information from multiple sources (such as meteorological and land use data), which limits the accuracy of ecosystem dynamics analysis.

[0004] The calculation of ecological parameters is single: At present, existing technologies only focus on the Normalized Difference Vegetation Index (NDVI) or simple ecological parameters, while the joint calculation and analysis capabilities of key ecological indicators (such as net primary productivity, aboveground biomass, vegetation coverage, etc.) are weak, making it difficult to fully reflect the status of ecological diversity.

[0005] Limited analysis of dynamic changes: Current dynamic analyses are mostly limited to simple trend assessments and lack quantitative analysis of ecosystem spatial autocorrelation and habitat fragmentation, and are unable to accurately capture the complex characteristics of ecosystem change.

[0006] Insufficient classification accuracy: Traditional interpretation and classification methods (such as pixel-based classification or expert interpretation rules) perform poorly in diverse and complex ecosystems, with low classification accuracy, making it difficult to meet the needs of ecological protection policy making. Summary of the invention

[0007] The purpose of the present invention is to provide a method for satellite remote sensing monitoring and verification of surface ecological diversity, which solves the technical problems of inconsistent data processing, insufficient dynamic change analysis and low classification accuracy in surface ecological diversity monitoring.

[0008] To achieve the above object, the present invention adopts the following technical solution: A method for satellite remote sensing monitoring and verification of surface ecological diversity, comprising the following steps: Step 1: Establish a data collection module. The data collection module collects remote sensing images and auxiliary data from the Internet, pre-processes the remote sensing images and auxiliary data respectively, and obtains the original data set; The preprocessing of remote sensing images includes radiation correction, atmospheric correction, geometric correction and cropping of the study area; the preprocessing of auxiliary data includes unification into raster data and meteorological data standardization; Step 2: The data modeling module performs mathematical modeling on ecological data and calculates ecological parameters, including normalized vegetation index, net primary productivity, vegetation coverage and aboveground biomass; Step 3: The interpretation and classification module builds an interpretation symbol library to classify the ecosystems displayed by the ecological data; Step 4: The dynamic change analysis module dynamically analyzes ecological data, analyzes the spatial autocorrelation of ecosystem changes, and quantifies the ecosystem change process. The dynamic analysis includes trend analysis calculations and habitat fragmentation calculations; Step 5: The result output module outputs the final results, including ecosystem classification maps, dynamic change reports, and protection strategy reports.

[0009] Preferably, when executing step 1, the remote sensing image is Landsat and MODIS image data obtained from a public platform, covering multi-temporal data within a preset year; Auxiliary data are land use data and meteorological data obtained from third-party platforms; The preprocessing of remote sensing images includes: Radiation correction processing, specifically using Landsat LED APS or MODIS correction tools to perform radiation correction and standardize the spectral reflectance of different image sources; Atmospheric correction processing, specifically using FLAASH or 6S model to remove atmospheric effects so that the image truly reflects the surface features; Geometric correction processing, specifically, making all remote sensing images consistent with the ground through geometric division; Crop the research area, specifically crop the remote sensing image according to the preset research scope; The preprocessing of land use data and meteorological data includes: The format of land use data is unified into raster data, and the spatial resolution is aligned with LandsatLEDAPS; The meteorological data are standardized and interpolated to the annual time scale on a monthly basis.

[0010] Preferably, when executing step 2, the mathematical models for calculating the normalized vegetation index, the net primary productivity, the vegetation coverage and the aboveground biomass are as follows: Normalized Difference Vegetation Index NDVI calculation model: ; in, is the reflectivity in the near-infrared band, is the reflectivity of red light band; The Normalized Difference Vegetation Index (NDVI) calculation model finally outputs a multi-phase NDVI time series data grid map within the preset year; Net primary productivity NPP calculation model, namely CASA model: ; in, represents the net primary productivity of pixel x in month t; represents the photosynthetically active radiation absorbed by pixel x in month t; It represents the actual light energy utilization rate of pixel x in month t; Represents the vegetation growing season distribution curve factor, with a value in the range of [0,1]; Absorbed photosynthetically active radiation The calculation formula is as follows: ; in, It represents the total solar radiation at pixel x in month t; 0.5 is a constant, which represents the proportion of effective solar radiation that can be used by vegetation to the total solar radiation; Indicates the absorption ratio of the vegetation layer to the incident photosynthetically active radiation PAR: ; in, represents the NDVI value at pixel x in month t, and They represent the maximum and minimum NDVI values ​​corresponding to the i-th vegetation type; and The values ​​are 0.01 and 0.95 respectively; Dynamic adjustment based on sample statistics and : ; ; in, is the average value of NDVI at pixel x, obtained by statistically analyzing sample NDVI data; is the standard deviation of NDVI at pixel x, which is used to measure the data dispersion of NDVI; and It is an empirical parameter, and its value is 1.5; Actual light energy utilization The calculation formula is as follows: ; in, and They respectively represent the effects of plants on light energy conversion efficiency under different temperature conditions; It indicates the effect of plants on light energy conversion rate under different water conditions; Indicates the maximum light energy utilization under ideal conditions; ; in, It represents the average temperature of the month when the NDVI reaches the highest in the area where the pixel x is located, that is, the optimum temperature, which represents the optimum temperature of point x in month t; ; in, Represents the actual temperature at point x in month t; ; in, represents the actual evapotranspiration at point x in month t, It represents the regional potential evapotranspiration at point x in month t; The specific model for calculating vegetation coverage is as follows: ; in, Represents vegetation coverage, Indicates the maximum NDVI value of pure vegetation pixels, Indicates the minimum NDVI value of pure non-vegetated pixels; The aboveground biomass calculation model is as follows: ; Among them, B represents biomass, NDVI represents vegetation index, represents latitude, α, β and γ represent regression coefficients; Vegetation growing season distribution curve factor The calculation formula is as follows: ; in, represents the NDVI value at pixel x in month t, It is used to reflect the dynamic distribution of vegetation growing season, which depends on the change of NDVI over time.

[0011] Preferably, when executing step 3, the following steps are specifically included: Step 3-1: Collect characteristic samples of different ecosystem types and build an interpretation sign library, including collecting direct signs and indirect signs. Direct signs include shape, size, shadow, hue, color, texture, pattern, location, and layout, while indirect signs include water system, landform, and vegetation. Construct ecosystem change raster data and calculate the change value of each pixel; Step 3-2: Construct a spatial weight matrix based on Euclidean distance or adjacency relationship , the Moran's I index model was used to evaluate the spatial autocorrelation of ecosystem changes, and the specific formula is as follows: ; in, It represents the Moran's I index, which reflects spatial autocorrelation. Its value range is [-1,1]. Its positive value indicates positive correlation, and its negative value indicates negative correlation. Indicates the number of sample points; , represent the ecological change values ​​of the ith and jth locations, respectively; represents the average value of all samples; represents the weight matrix, which represents the spatial weight between position i and position j (specifically the inverse of the distance); Represents the normalization factor of the weight matrix: ; Step 3-3: Use deep learning methods to classify ecosystems. Specifically, use the convolutional neural network (CNN) model to classify ecosystems. The input data of the convolutional neural network (CNN) model is remote sensing images and raster data. The specific model formula is as follows: ; in, represents the kth feature map in the lth layer; Represents the convolution kernel weight; represents the bias term; Represents the activation function ReLU; The pooling layer of the convolutional neural network (CNN) model reduces the dimension of the feature map, retains the key features, performs classification probability prediction in the fully connected layer, and outputs the ecological type classification results; Step 3-4: Use the confusion matrix to evaluate the classification accuracy, calculate the overall accuracy OA and Kappa coefficient, and the goal is to achieve an overall accuracy OA of 85%.

[0012] Preferably, when executing step 4, the specific models for trend analysis calculation and habitat fragmentation calculation are as follows: The model for trend analysis calculation is the Sen+MK model, and the specific formula is as follows: Sen slope formula: ; in, It represents the slope of time series change; Use the MK significance test to calculate the significance p-value and determine whether the trend is significant; The model formula for calculating habitat fragmentation is as follows: ; in, is the fragmentation degree of the i-th ecosystem, is the total area, is the number of plaques.

[0013] The method for satellite remote sensing monitoring and verification of surface ecological diversity described in the present invention solves the technical problems of inconsistent data processing, insufficient dynamic change analysis and low classification accuracy in surface ecological diversity monitoring. The present invention integrates Landsat and MODIS remote sensing images, meteorological data and land use data, and ensures data consistency and high quality through preprocessing processes such as radiation correction and atmospheric correction. It introduces a multidimensional calculation model of normalized difference vegetation index (NDVI), net primary productivity (NPP), vegetation coverage and aboveground biomass, which can comprehensively reflect the dynamic characteristics of the ecosystem, quantify the change trend of the ecosystem through trend analysis (Sen+MK model) and habitat fragmentation calculation, evaluate the spatial autocorrelation of ecological changes through the Moran's I index model, provide an accurate reference basis for ecological protection, adopt convolutional neural network (CNN) to classify the ecosystem, improve the accuracy of complex ecosystem classification by optimizing the convolution layer and the classification model, and use confusion matrix to evaluate the classification effect to ensure that the overall accuracy (OA) reaches more than 85%. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a schematic diagram of the system architecture of the present invention; Figure 2 It is the main flow chart of the present invention; Figure 3 is a flow chart of step 2 of the present invention; Figure 4 It is a flow chart of step 4 of the present invention. DETAILED DESCRIPTION

[0015] Depend on Figure 1-Figure 4 A method for satellite remote sensing monitoring and verification of surface ecological diversity is shown, comprising the following steps: Step 1: Establish a data collection module. The data collection module collects remote sensing images and auxiliary data from the Internet, pre-processes the remote sensing images and auxiliary data respectively, and obtains the original data set; In this embodiment, in actual application, the data collection module is deployed in the data input layer, which mainly includes a remote sensing image acquisition module, an auxiliary data acquisition module and a preprocessing module. The remote sensing image acquisition module is used to obtain public remote sensing images such as Landsat and MODIS, covering multi-temporal image data; the auxiliary data acquisition module is used to obtain land use data (such as rasterized land use classification data), and also obtains meteorological data (monthly meteorological observation data); the preprocessing module is used to preprocess remote sensing images and auxiliary data.

[0016] The preprocessing of remote sensing images includes radiation correction, atmospheric correction, geometric correction and cropping of the study area; the preprocessing of auxiliary data includes unification into raster data and meteorological data standardization; When executing step 1, the remote sensing images are Landsat and MODIS image data obtained from the public platform, covering multi-temporal data within the preset year; Auxiliary data are land use data and meteorological data obtained from third-party platforms; In this embodiment, image data is obtained from the official data release platform of Landsat and MODIS satellites. The sources of remote sensing images are as follows: Sentinel remote sensing data: The Sentinel remote sensing data used is from the European Space Agency website. Sentinel-2 is a high-resolution multispectral imaging satellite carrying a multispectral imager (MSI) for land monitoring. It can provide images of vegetation, soil and water cover, inland waterways and coastal areas. It is divided into two satellites, 2A and 2B. Sentinel-2B is the same group as the Sentinel-2A satellite launched in June 2015. The Sentinel-2 satellite is 786km high, covering 13 spectral bands with a width of 290 kilometers. The ground resolutions are 10m, 20m and 60m respectively. The revisit period of one satellite is 10 days, and the two satellites complement each other with a revisit period of 5 days. From visible light and near-infrared to short-wave infrared, it has different spatial resolutions. Among optical data, Sentinel-2 data is the only one that contains three bands in the red edge range, which is very effective for monitoring vegetation health information.

[0017] Source of MODIS satellite remote sensing data: The MODIS remote sensing satellite data used comes from NASA's MODIS download website. MODIS NDVI (Normalized Difference Vegetation Index) and NOAA Advanced Very High Resolution Radiometer (AVHRR) NDVI products can be used together to show the time series changes of vegetation index. MODIS NDVI products are calculated by bidirectional atmospheric correction surface reflectance processed by water, cloud, heavy aerosol, and cloud shadow mask. The global MYD13Q1 data is a 16-day synthetic 250-meter L3 data product, projected as a sinusoidal curve projection.

[0018] LANDSAT satellite remote sensing data source: Landsat satellite remote sensing data mainly includes data from two satellites, Landsat-8 and Landsat-9. The data comes from the NASA data download website in the United States.

[0019] The high-resolution remote sensing dataset mainly comes from Google images, and its main purpose is to assist in ecosystem sample delineation and verify ecosystem classification data.

[0020] The preprocessing of remote sensing images includes: Radiation correction processing, specifically using Landsat LED APS or MODIS correction tools to perform radiation correction and standardize the spectral reflectance of different image sources; Atmospheric correction processing, specifically using FLAASH or 6S model to remove atmospheric effects so that the image truly reflects the surface features; Geometric correction processing, specifically, making all remote sensing images consistent with the ground through geometric division; Crop the research area, specifically crop the remote sensing image according to the preset research scope; The preprocessing of land use data and meteorological data includes: The format of land use data is unified into raster data, and the spatial resolution is aligned with LandsatLEDAPS; The meteorological data are standardized and interpolated to the annual time scale on a monthly basis.

[0021] Step 2: The data modeling module performs mathematical modeling on ecological data and calculates ecological parameters, including normalized vegetation index, net primary productivity, vegetation coverage and aboveground biomass; In this embodiment, the data modeling module is deployed in the data processing layer, and is mainly used for ecological parameter calculation, including normalized difference vegetation index (NDVI) calculation, net primary productivity (NPP) calculation based on CASA model, vegetation coverage calculation and aboveground biomass calculation.

[0022] When executing step 2, the mathematical models for calculating the normalized vegetation index, net primary productivity, vegetation coverage, and aboveground biomass are as follows: Normalized Difference Vegetation Index NDVI calculation model: ; in, is the reflectivity in the near-infrared band, is the reflectivity of red light band; The Normalized Difference Vegetation Index (NDVI) calculation model finally outputs a multi-phase NDVI time series data grid map within the preset year; In this example, NDVI characterizes vegetation growth conditions by the difference in reflectance between the red and near-infrared (NIR) bands. Vegetation absorbs red light for photosynthesis and strongly reflects near-infrared light. The NDVI value range is [−1,1], usually: NDVI>0.2: indicates vegetation coverage, and the higher the value, the healthier the vegetation; NDVI≤0.2: usually soil, rock or water; The Normalized Difference Vegetation Index (NDVI) provides the vegetation growth status of pixels in remote sensing images and provides input for the calculation of photosynthetically active radiation (FPAR) and vegetation cover (FC).

[0023] Net primary productivity NPP calculation model, namely CASA model: ; in, represents the net primary productivity of pixel x in month t; represents the photosynthetically active radiation absorbed by pixel x in month t; It represents the actual light energy utilization rate of pixel x in month t; Represents the vegetation growing season distribution curve factor, with a value in the range of [0,1]; In this embodiment, It is the net primary productivity of pixel x in period t and is an important indicator for measuring the amount of carbon fixed by vegetation through photosynthesis. NPP is the core output of the entire model and is used to study the carbon cycle of the ecosystem.

[0024] Absorbed photosynthetically active radiation The calculation formula is as follows: ; in, It represents the total solar radiation at pixel x in month t; 0.5 is a constant, which represents the proportion of effective solar radiation that can be used by vegetation to the total solar radiation; Indicates the absorption ratio of the vegetation layer to the incident photosynthetically active radiation PAR: ; in, represents the NDVI value at pixel x in month t, and They represent the maximum and minimum NDVI values ​​corresponding to the i-th vegetation type, respectively, and determine the normalized range of FPAR; and The values ​​are 0.01 and 0.95 respectively; In this embodiment, FPAR is a key part of APAR. It quantifies the efficiency of plants in absorbing light energy and is closely related to the coverage and health of vegetation.

[0025] Dynamic adjustment based on sample statistics and : ; ; in, is the average value of NDVI at pixel x, obtained by statistically analyzing sample NDVI data; is the standard deviation of NDVI at pixel x, which is used to measure the data dispersion of NDVI; and It is an empirical parameter, and its value is 1.5; Actual light energy utilization The calculation formula is as follows: ; in, and They respectively represent the effects of plants on light energy conversion efficiency under different temperature conditions; It indicates the effect of plants on light energy conversion rate under different water conditions; Indicates the maximum light energy utilization under ideal conditions; In this embodiment, The energy conversion efficiency used for APAR reflects the impact of environmental constraints on photosynthesis.

[0026] ; in, It represents the average temperature of the month when the NDVI reaches the highest in the area where the pixel x is located, that is, the optimum temperature, which represents the optimum temperature of point x in month t; ; in, Represents the actual temperature at point x in month t; ; in, represents the actual evapotranspiration at point x in month t, It represents the regional potential evapotranspiration at point x in month t; The specific model for calculating vegetation coverage is as follows: ; in, Represents vegetation coverage, Indicates the maximum NDVI value of pure vegetation pixels, Indicates the minimum NDVI value of pure non-vegetated pixels; In this embodiment, the vegetation coverage It reflects the vegetation coverage level of pixels in the study area, provides support for ecosystem evaluation and classification, and is highly correlated with changes in NDVI.

[0027] The aboveground biomass calculation model is as follows: ; Among them, B represents biomass, NDVI represents vegetation index, represents latitude, α, β and γ represent regression coefficients; Vegetation growing season distribution curve factor The calculation formula is as follows: ; in, represents the NDVI value at pixel x in month t, It is used to reflect the dynamic distribution of vegetation growing season, which depends on the change of NDVI over time.

[0028] Step 3: The interpretation and classification module builds an interpretation symbol library to classify the ecosystems displayed by the ecological data; When executing step 3, the specific steps include: Step 3-1: Collect characteristic samples of different ecosystem types and build an interpretation sign library, including collecting direct signs and indirect signs. Direct signs include shape, size, shadow, hue, color, texture, pattern, location, and layout, while indirect signs include water system, landform, and vegetation. Construct ecosystem change raster data and calculate the change value of each pixel; Step 3-2: Construct a spatial weight matrix based on Euclidean distance or adjacency relationship , the Moran's I index model was used to evaluate the spatial autocorrelation of ecosystem changes, and the specific formula is as follows: ; in, It represents the Moran's I index, which reflects spatial autocorrelation. Its value range is [-1,1]. Its positive value indicates positive correlation, and its negative value indicates negative correlation. Indicates the number of sample points; , represent the ecological change values ​​of the ith and jth locations, respectively; represents the average value of all samples; represents the weight matrix, which represents the spatial weight between position i and position j (specifically the inverse of the distance); Represents the normalization factor of the weight matrix: ; Step 3-3: Use deep learning methods to classify ecosystems. Specifically, use the convolutional neural network (CNN) model to classify ecosystems. The input data of the convolutional neural network (CNN) model is remote sensing images and raster data. The specific model formula is as follows: ; in, represents the kth feature map in the lth layer; Represents the convolution kernel weight; represents the bias term; Represents the activation function ReLU; The pooling layer of the convolutional neural network (CNN) model reduces the dimension of the feature map, retains the key features, performs classification probability prediction in the fully connected layer, and outputs the ecological type classification results; The training data for the convolutional neural network (CNN) model is labeled ecological type sample data; The loss function is cross entropy loss: ; in, is the true value of sample i belonging to category c, Represents the probability value predicted by the model; Step 3-4: Use the confusion matrix to evaluate the classification accuracy, calculate the overall accuracy OA and Kappa coefficient, and the goal is to achieve an overall accuracy OA of 85%.

[0029] In this embodiment, the confusion matrix, as an existing technology, has a mature application basis and is widely used in the accuracy assessment of classification problems, such as image recognition, ecological monitoring and remote sensing interpretation. In this embodiment, it is necessary to accurately evaluate the ecological type classification results output by the CNN model. The confusion matrix can provide multi-dimensional evaluation indicators such as precision, recall and overall accuracy (OA), which is suitable for such multi-classification tasks. The evaluation indicators generated by the confusion matrix (such as Kappa coefficient and overall accuracy) can intuitively quantify the performance of the classification model and track and optimize the classification error.

[0030] When used specifically, the specific operations of using the confusion matrix to evaluate classification accuracy are as follows: The input of the confusion matrix is: the classification results predicted by the CNN model + the real ecosystem labeled samples; The confusion matrix calculates the following key metrics and is used to evaluate the performance of a classification model: Overall Accuracy (OA): The ratio of correctly classified samples to the total number of samples; Kappa coefficient: evaluates the random consistency of classification results and is used for more rigorous performance evaluation; Target requirements: OA ≥ 85% to ensure high reliability of ecological classification results; The Kappa coefficient was significantly higher than 0.8, indicating that the classification results were highly consistent.

[0031] Step 4: The dynamic change analysis module dynamically analyzes ecological data, analyzes the spatial autocorrelation of ecosystem changes, and quantifies the ecosystem change process. The dynamic analysis includes trend analysis calculations and habitat fragmentation calculations; In this embodiment, the interpretation and classification module and the dynamic change analysis module are deployed in the analysis and modeling layer, which are mainly used to interpret the construction of the sign library, build a convolutional neural network (CNN) classification model, quantify the changing trend of ecological parameters based on the Sen + MK model, analyze spatial autocorrelation based on the Moran's I index model, and evaluate the degree of ecosystem fragmentation.

[0032] When executing step 4, the specific models for trend analysis and habitat fragmentation calculation are as follows: The model for trend analysis calculation is the Sen+MK model, and the specific formula is as follows: Sen slope formula: ; in, It represents the slope of time series change; Use the MK significance test to calculate the significance p-value and determine whether the trend is significant; The model formula for calculating habitat fragmentation is as follows: ; in, is the fragmentation degree of the i-th ecosystem, is the total area, is the number of plaques.

[0033] Step 5: The result output module outputs the final results, including ecosystem classification maps, dynamic change reports, and protection strategy reports.

[0034] In this embodiment, the result output module is deployed in the result output layer and is responsible for outputting analysis results, dynamic change reports and protection strategy recommendations.

[0035] In this embodiment, the contents displayed in the ecosystem classification map are as follows: Standardized remote sensing images, land use data and meteorological data, which mainly provide basic input data for classification; The generated ecological parameters (such as NDVI, NPP, vegetation cover and aboveground biomass), which are also mainly used as feature input data to support classification models; After the remote sensing images and ecological parameters are classified by convolutional neural network (CNN), the classification results are output.

[0036] The dynamic change report shows the following: Provide multi-temporal NDVI data, NPP, vegetation coverage and other temporal ecological parameters, which are also used for dynamic analysis; Quantify dynamic changes in ecosystems, such as: Results of trend analysis (Sen + MK model) to assess trends and significance of ecosystem changes; The results of spatial autocorrelation analysis (Moran's I index model) were used to analyze the spatial distribution of ecological changes; The results of habitat fragmentation calculation are used to quantify the degree of ecosystem fragmentation.

[0037] The content of the protection strategy report includes: Provide ecological parameters (such as NPP, vegetation cover, and aboveground biomass) to reflect the health status of the ecosystem; Provide ecosystem classification results and clarify the distribution of ecological types; Display indicators such as trend analysis and habitat fragmentation to show the direction of ecosystem change and potential threats.

[0038] The method for satellite remote sensing monitoring and verification of surface ecological diversity described in the present invention solves the technical problems of inconsistent data processing, insufficient dynamic change analysis and low classification accuracy in surface ecological diversity monitoring. The present invention integrates Landsat and MODIS remote sensing images, meteorological data and land use data, and ensures data consistency and high quality through preprocessing processes such as radiation correction and atmospheric correction. It introduces a multidimensional calculation model of normalized difference vegetation index (NDVI), net primary productivity (NPP), vegetation coverage and aboveground biomass, which can comprehensively reflect the dynamic characteristics of the ecosystem, quantify the change trend of the ecosystem through trend analysis (Sen+MK model) and habitat fragmentation calculation, evaluate the spatial autocorrelation of ecological changes through the Moran's I index model, provide an accurate reference basis for ecological protection, adopt convolutional neural network (CNN) to classify the ecosystem, improve the accuracy of complex ecosystem classification by optimizing the convolution layer and the classification model, and use confusion matrix to evaluate the classification effect to ensure that the overall accuracy (OA) reaches more than 85%.

Claims

1. A method for satellite remote sensing monitoring and verification of surface ecological diversity, characterized in that: The steps include: Step 1: Establish a data collection module. The data collection module collects remote sensing images and auxiliary data from the Internet, pre-processes the remote sensing images and auxiliary data respectively, and obtains the original data set; The preprocessing of remote sensing images includes radiation correction, atmospheric correction, geometric correction and cropping of the study area; the preprocessing of auxiliary data includes unification into raster data and meteorological data standardization; Step 2: The data modeling module performs mathematical modeling on ecological data and calculates ecological parameters, including normalized vegetation index, net primary productivity, vegetation coverage and aboveground biomass; Step 3: The interpretation and classification module builds an interpretation symbol library to classify the ecosystems displayed by the ecological data; Step 4: The dynamic change analysis module dynamically analyzes ecological data, analyzes the spatial autocorrelation of ecosystem changes, and quantifies the ecosystem change process. The dynamic analysis includes trend analysis calculations and habitat fragmentation calculations; Step 5: The result output module outputs the final results, including ecosystem classification maps, dynamic change reports, and protection strategy reports.

2. The method for satellite remote sensing monitoring and verification of surface ecological diversity according to claim 1, characterized in that: When executing step 1, the remote sensing images are Landsat and MODIS image data obtained from the public platform, covering multi-temporal data within the preset year; Auxiliary data are land use data and meteorological data obtained from third-party platforms; The preprocessing of remote sensing images includes: Radiation correction processing, specifically using Landsat LED APS or MODIS correction tools to perform radiation correction and standardize the spectral reflectance of different image sources; Atmospheric correction processing, specifically using FLAASH or 6S model to remove atmospheric effects so that the image truly reflects the surface features; Geometric correction processing, specifically, making all remote sensing images consistent with the ground through geometric division; Crop the research area, specifically crop the remote sensing image according to the preset research scope; The preprocessing of land use data and meteorological data includes: The format of land use data is unified into raster data, and the spatial resolution is aligned with LandsatLEDAPS; The meteorological data are standardized and interpolated to the annual time scale on a monthly basis.

3. The method for satellite remote sensing monitoring and verification of surface ecological diversity according to claim 1, characterized in that: When executing step 2, the mathematical models for calculating the normalized vegetation index, net primary productivity, vegetation coverage, and aboveground biomass are as follows: Normalized Difference Vegetation Index NDVI calculation model: ； in, is the reflectivity in the near-infrared band, is the reflectivity of red light band; The Normalized Difference Vegetation Index (NDVI) calculation model finally outputs a multi-phase NDVI time series data grid map within the preset year; Net primary productivity NPP calculation model, namely CASA model: ； in, represents the net primary productivity of pixel x in month t; represents the photosynthetically active radiation absorbed by pixel x in month t; It represents the actual light energy utilization rate of pixel x in month t; Represents the vegetation growing season distribution curve factor, with a value in the range of [0,1]; Absorbed photosynthetically active radiation The calculation formula is as follows: ； in, It represents the total solar radiation at pixel x in month t; 0.5 is a constant, which represents the proportion of effective solar radiation that can be used by vegetation to the total solar radiation; Indicates the absorption ratio of the vegetation layer to the incident photosynthetically active radiation PAR: ； in, represents the NDVI value at pixel x in month t, and They represent the maximum and minimum NDVI values ​​corresponding to the i-th vegetation type; and The values ​​are 0.01 and 0.95 respectively; Dynamic adjustment based on sample statistics and : ; ； in, is the average value of NDVI at pixel x, obtained by statistically analyzing sample NDVI data; is the standard deviation of NDVI at pixel x, which is used to measure the data dispersion of NDVI; and It is an empirical parameter, and its value is 1.5; Actual light energy utilization The calculation formula is as follows: ； in, and They respectively represent the effects of plants on light energy conversion efficiency under different temperature conditions; It indicates the effect of plants on light energy conversion rate under different water conditions; Indicates the maximum light energy utilization under ideal conditions; ； in, It represents the average temperature of the month when the NDVI reaches the highest in the area where the pixel x is located, that is, the optimum temperature, which represents the optimum temperature of point x in month t; ； in, Represents the actual temperature at point x in month t; ； in, represents the actual evapotranspiration at point x in month t, It represents the regional potential evapotranspiration at point x in month t; The specific model for calculating vegetation coverage is as follows: ； in, Represents vegetation coverage, Indicates the maximum NDVI value of pure vegetation pixels, Indicates the minimum NDVI value of pure non-vegetated pixels; The aboveground biomass calculation model is as follows: ； Among them, B represents biomass, NDVI represents vegetation index, represents latitude, α, β and γ represent regression coefficients; Vegetation growing season distribution curve factor The calculation formula is as follows: ； in, represents the NDVI value at pixel x in month t, It is used to reflect the dynamic distribution of vegetation growing season, which depends on the change of NDVI over time.

4. The method for satellite remote sensing monitoring and verification of surface ecological diversity according to claim 1, characterized in that: When executing step 3, the specific steps include: Step 3-1: Collect characteristic samples of different ecosystem types and build an interpretation sign library, including collecting direct signs and indirect signs. Direct signs include shape, size, shadow, hue, color, texture, pattern, location, and layout, while indirect signs include water system, landform, and vegetation. Construct ecosystem change raster data and calculate the change value of each pixel; Step 3-2: Construct a spatial weight matrix based on Euclidean distance or adjacency relationship , the Moran's I index model was used to evaluate the spatial autocorrelation of ecosystem changes, and the specific formula is as follows: ； in, It represents the Moran's I index, which reflects spatial autocorrelation. Its value range is [-1,1]. Its positive value indicates positive correlation, and its negative value indicates negative correlation. Indicates the number of sample points; , represent the ecological change values ​​of the ith and jth locations, respectively; represents the average value of all samples; represents the weight matrix, which represents the spatial weight between position i and position j; Represents the normalization factor of the weight matrix: ； Step 3-3: Use deep learning methods to classify ecosystems. Specifically, use the convolutional neural network (CNN) model to classify ecosystems. The input data of the convolutional neural network (CNN) model is remote sensing images and raster data. The specific model formula is as follows: ； in, represents the kth feature map in the lth layer; Represents the convolution kernel weight; represents the bias term; Represents the activation function ReLU; The pooling layer of the convolutional neural network (CNN) model reduces the dimension of the feature map, retains the key features, performs classification probability prediction in the fully connected layer, and outputs the ecological type classification results; Step 3-4: Use the confusion matrix to evaluate the classification accuracy and calculate the overall accuracy OA and Kappa coefficient.

5. The method for satellite remote sensing monitoring and verification of surface ecological diversity according to claim 1, characterized in that: When executing step 4, the specific models for trend analysis and habitat fragmentation calculation are as follows: The model for trend analysis calculation is the Sen+MK model, and the specific formula is as follows: Sen slope formula: ； in, It represents the slope of time series change; Use the MK significance test to calculate the significance p-value and determine whether the trend is significant; The model formula for calculating habitat fragmentation is as follows: ； in, is the fragmentation degree of the i-th type of ecosystem, is the total area, is the number of plaques.

Images