A remote sensing identification method and system applied to ecosystem investigation

Abstract

The application relates to the technical field of remote sensing identification, in particular to a remote sensing identification method and system applied to ecological system investigation. The method comprises the following steps: acquiring multi-temporal remote sensing image data of a target region; extracting a ground cover type of the multi-temporal remote sensing image data, and identifying a region of explicit human activities according to the ground cover type of the multi-temporal remote sensing image data to generate data of the region of explicit human activities; acquiring night light data of the target region; performing space-time registration on the night light data of the target region and the multi-temporal remote sensing image data to generate fused night light remote sensing data; extracting an edge region of the multi-temporal remote sensing image data through the ground cover type to obtain edge region data; and calculating a vegetation index and a night light index of the edge region data based on the fused night light remote sensing data. Through multi-temporal, multi-source data fusion and multi-index time sequence analysis, the application improves the accuracy and comprehensiveness of remote sensing identification of ecological system investigation.

Description

Technical Field

[0001] This invention relates to the field of remote sensing identification technology, and in particular to a remote sensing identification method and system for ecosystem surveys. Background Technology

[0002] The rapid development of satellite remote sensing technology, particularly multispectral and hyperspectral imaging, has significantly enhanced the ability to identify ecosystems. Multispectral remote sensing captures spectral information across different bands, enabling the classification and monitoring of ecological elements such as vegetation types, water bodies, and soil. Hyperspectral remote sensing further enhances the ability to identify species and ecological structures in detail, promoting refined analysis of ecosystem diversity and health. With the development of computer technology and artificial intelligence, remote sensing image analysis methods based on machine learning and deep learning have emerged. These methods improve the efficiency and accuracy of ecosystem identification by automatically extracting image features, effectively processing large-scale, high-dimensional remote sensing data. However, current traditional methods primarily focus on overt human activities, neglecting the complex ecological impacts of peripheral areas. Furthermore, existing technologies often concentrate on identifying overt threats, ignoring potential latent threats within ecosystems, resulting in lower accuracy and comprehensiveness in remote sensing identification of ecosystem surveys. Summary of the Invention

[0003] Therefore, it is necessary to provide a remote sensing identification method and system for ecosystem surveys to solve at least one of the aforementioned technical problems.

[0004] To achieve the above objectives, a remote sensing identification method for ecosystem surveys is provided, the method comprising the following steps:

[0005] Step S1: Acquire multi-temporal remote sensing image data of the target area; extract the land cover type from the multi-temporal remote sensing image data, and identify the visible human activity area based on the land cover type to generate visible human activity area data.

[0006] Step S2: Acquire nighttime light data of the target area; perform spatiotemporal registration of the nighttime light data of the target area and multi-temporal remote sensing image data to generate fused nighttime light remote sensing data; extract the boundary areas of the multi-temporal remote sensing image data by land cover type to obtain edge area data; calculate the vegetation index and nighttime light index of the edge area data based on the fused nighttime light remote sensing data.

[0007] Step S3: Perform biological behavior correlation analysis on edge area data using vegetation index and nighttime light index to generate edge effect identification data; analyze the time series changes of low-frequency disturbances in data of areas with visible human activities to generate low-frequency disturbance identification data; perform ecosystem health assessment prediction based on low-frequency disturbance identification data and edge effect identification data to generate ecosystem health assessment prediction data for the target area.

[0008] Step S4: Based on the ecosystem health assessment and prediction data of the target area, output the regional latent threats to the ecosystem of the target area, thereby obtaining the latent threat level and distribution map.

[0009] This invention effectively enhances the collaborative processing capabilities of multi-source heterogeneous data in terms of spatial resolution and temporal consistency by spatiotemporal registration and fusion of multi-temporal remote sensing images and nighttime light data. It extracts visible human activity areas based on land cover type and combines vegetation and nighttime light indices in boundary areas for edge effect analysis, enhancing the ability to identify the spatial extent of human activities and areas with ambiguous boundaries. It performs biological behavior correlation analysis using vegetation and nighttime light change indices in edge areas and combines this with low-frequency disturbance time-series data from visible areas to achieve deep coupling analysis between dynamic ecological responses and human activity disturbances. Using edge effect identification data and low-frequency disturbance identification data as core input features, an ecosystem health prediction model is established, exhibiting strong data-driven characteristics and model scalability. Utilizing ecosystem health prediction data, a latent threat level classification mechanism is constructed and a spatial distribution map is output, effectively improving the accuracy and visualization capabilities of ecological threat identification. Therefore, this invention improves the accuracy and comprehensiveness of remote sensing identification in ecosystem surveys through multi-temporal, multi-source data fusion and multi-index time-series analysis.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: Perform multi-phase sampling control of remote sensing images on the target area to obtain multi-temporal remote sensing image data;

[0012] Step S12: Perform radiometric and geometric corrections on the multi-temporal remote sensing image data to generate corrected multi-temporal remote sensing image data;

[0013] Step S13: Extract the surface spectral features of the corrected multi-temporal remote sensing image data, and perform temporal variation analysis and classification processing on the surface spectral features to generate surface cover type data;

[0014] Step S14: Interpret human activity indicators from land cover type data to generate preliminary human activity mask data; perform spatial consistency screening and boundary correction on the preliminary human activity mask data to generate explicit human activity area data.

[0015] This invention utilizes a multi-phase sampling control mechanism to systematically capture remote sensing information of a target area at different time points, providing a high-quality spatiotemporal data foundation for subsequent change detection and dynamic analysis. It effectively eliminates radiation errors caused by factors such as sensor activity, solar angle, and atmospheric interference, as well as geometric distortions caused by changes in perspective and terrain, ensuring high analytical usability and spatial registration accuracy of the remote sensing data. By extracting spectral variation characteristics from multi-temporal remote sensing images and combining them with temporal dynamic models for classification analysis, the accuracy of identifying complex land cover types (such as the conversion between farmland, forest, and construction land) can be improved. Using land cover classification results to extract indicators of human activity, supplemented by spatial consistency screening and boundary correction, can significantly improve the spatial integrity and boundary accuracy of human activity masks, reducing the false positive rate.

[0016] Preferably, step S13 includes the following steps:

[0017] Step S131: Extract multi-band reflectance values ​​of surface pixels at different time phases from the corrected multi-temporal remote sensing image data to obtain a surface spectral feature sequence with time labels;

[0018] Step S132: Calculate the temporal variation trend of key band reflectance in multi-temporal remote sensing image data based on the surface spectral feature sequence, and determine whether its variation pattern meets any of the following conditions and mark it accordingly to generate temporal variation judgment label data: NDVI index has seasonal periodic fluctuations and the annual amplitude is greater than 0.3; the position of the crossover point of blue light band and red edge band reflectance changes by more than ±15nm in three consecutive temporal phases; the number of inflection points of the spectral curve varies by more than 2 in different growth stages;

[0019] Step S133: Use time-series change judgment marker data to identify the land cover type based on the spectral characteristics of the land surface. If any of the following conditions are met, it can be determined as the specified land cover type and land cover type data can be generated: If the maximum NDVI value exceeds 0.6 and the annual fluctuation range of NDVI is greater than 0.35, it is classified as "crops"; if the shortwave infrared reflectance changes by less than 5% throughout the year and the NDVI is consistently below 0.2, it is classified as "built surface"; if the normalized difference index of red light and near-infrared light shows a double peak throughout the year and the peak values ​​correspond to spring and autumn, respectively, it is classified as "deciduous woodland".

[0020] This invention extracts multi-band reflectance from different temporal phases to generate time-stamped spectral feature sequences, providing a high-dimensional and continuous remote sensing information foundation for subsequent temporal change analysis and significantly improving the ability to identify dynamic features such as seasonality and growth period of land cover. It innovatively combines multi-scale features such as NDVI seasonal fluctuation amplitude, blue-red edge intersection shift, and spectral curve inflection point changes to determine temporal change patterns, breaking through traditional classification methods based on single vegetation indices or static spectral values, and enhancing the discriminative power for complex ecological cover type changes. By performing rule matching on the labeling data of land surface temporal features and combining multiple biophysical parameters (such as NDVI, red / near-infrared index, and shortwave infrared stability) to set fine classification standards, it can effectively improve the identification accuracy of typical land cover types such as "crops," "built-up areas," and "deciduous forests," and enhance the classification system's generalization ability to cross-seasonal and cross-regional data. The overall process can automatically extract key change features and perform discriminative classification during large-scale remote sensing data processing, reducing the cost of manual intervention and adapting to the remote sensing classification needs of different regional ecological backgrounds.

[0021] Preferably, step S2 includes the following steps:

[0022] Step S21: Obtain nighttime light data for the target area;

[0023] Step S22: Perform radiation normalization and noise suppression on the raw nighttime light radiation data to generate net nighttime light radiation data; align the net nighttime light radiation data with the multi-temporal remote sensing image correction data to generate a spatiotemporal registration parameter set.

[0024] Step S23: Based on the spatiotemporal registration parameter set, perform raster-level registration and resampling fusion of nighttime light data and multi-temporal remote sensing image data of the target area to generate fused nighttime light remote sensing data; use land cover type data to perform edge detection and morphological region growth analysis on the boundary change areas of the remote sensing image to generate boundary area layers;

[0025] Step S24: Spatial extraction and segmentation of the boundary area layer to generate edge area data; vegetation index calculation of the edge area data based on the fused nighttime light remote sensing data to generate edge area vegetation index data;

[0026] Step S25: Extract the night light index from the edge area data based on the fused night light remote sensing data to generate edge area night light index data.

[0027] This invention constructs fused nighttime light remote sensing data under a unified spatiotemporal reference system by radiometric normalization, spatiotemporal registration, and fusion processing of nighttime light data and multi-temporal remote sensing images. This effectively integrates human activity intensity and natural surface information, enhancing the comprehensive perception capability of dynamic changes in edge areas. Based on precise spatiotemporal reference points, a raster-level registration model is constructed, and image data of different resolutions are resampled and fused, effectively alleviating spatial misalignment and scale mismatch problems during the fusion of heterogeneous remote sensing data, improving the consistency and reliability of data integration. By combining land cover type for edge detection and morphological region growing processing, areas of change in land cover boundaries can be located more accurately, particularly suitable for edge transition zone analysis scenarios such as urban expansion and land use conversion. Spatial extraction and segmentation methods are used to further refine the boundary areas into edge regions, providing accurate spatial boundary support for subsequent indicator extraction and ecological analysis, improving the analysis model's ability to identify ecological transition zones and human intervention boundaries. By simultaneously extracting vegetation index and nighttime light index from the edge area based on uniformly registered data, the interaction between ecological status and human disturbance can be reflected, providing highly sensitive and multi-dimensional indicator support for subsequent biological behavior analysis and ecological risk assessment.

[0028] Preferably, step S23, which utilizes land cover type data to perform edge detection and morphological region growing analysis on the boundary change areas of the remote sensing image, includes:

[0029] Edge extraction was performed on the boundary change area of ​​the remote sensing image using land cover type data. The high threshold was set to 100 pixels of gray value and the low threshold was set to 50 pixels of gray value. A 3×3 Gaussian kernel was used for smoothing to obtain the outline data with clear boundaries, which was then converted into vector polygon format.

[0030] Using contour line data as seed regions, a region growing algorithm is used to expand to regions with similar land features. The spectral similarity threshold is set to ±10% reflectance, and the area threshold is greater than or equal to 1000m². 2 The growth radius is within 90 meters, and the growth results are re-marked and rasterized into boundary areas;

[0031] The boundary area and the remote sensing image boundary change area are overlaid to generate a vector layer; each boundary area unit is assigned the following attribute fields: change type, change area, boundary length, and change time period, thereby outputting the boundary area layer.

[0032] This invention introduces land cover type as prior information during edge extraction, combined with a set dual threshold (high threshold 100, low threshold 50) and 3×3 Gaussian kernel smoothing, effectively suppressing noise and enhancing boundary contours. This results in clear, structurally complete contour line data, significantly improving the accuracy of remote sensing image boundary change detection. Using the contour line data as seed regions, spectral similarity (±10% reflectance) and area (≥1000m²) are used as the basis for further analysis. 2 Regional growth analysis was conducted using multi-dimensional constraints such as spatial growth radius (≤90 meters) to effectively prevent overexpansion or missed detection, thus improving the spatial modeling capability for complex boundary areas. During the region identification process, the extracted results were converted from contour lines to vector polygons and then rasterized, which not only improved the flexibility and accuracy of data representation but also facilitated subsequent spatial calculations and ecological indicator extraction, enhancing the engineering adaptability of the remote sensing processing. By overlaying boundary area images and assigning attribute fields such as change type, change area, boundary length, and change time period, a structured and quantifiable boundary area layer was constructed, providing strong data support and a visualization foundation for subsequent ecological monitoring, edge effect analysis, and regional dynamic modeling. Utilizing the boundary overlay and attribute annotation mechanism of multi-temporal images, precise tracking and classification of boundary area changes over time were achieved, improving the controllability and accuracy of ecological edge area change monitoring.

[0033] Preferably, step S3, which involves performing biological behavior correlation analysis on the edge area data using vegetation index and nighttime light index, includes:

[0034] Spatial grid division is performed on edge region data, and vegetation index changes are extracted by combining time-series remote sensing images to generate edge vegetation dynamic index data;

[0035] By overlaying nighttime light remote sensing images onto edge region data, nighttime light intensity and stability indices are extracted to generate edge nighttime light index data.

[0036] A bivariate spatiotemporal co-analysis of vegetation index and night light index was performed to generate vegetation-night light coupling characteristic data;

[0037] Spatiotemporal pattern clustering was performed on vegetation-night light coupling feature data to identify the coupling effect of human activity intervention and ecological disturbance, and to generate biological behavior response layer data.

[0038] Edge sensitivity analysis and biodiversity impact assessment are performed on biological behavior response layer data to generate edge effect identification data.

[0039] This invention, through spatial grid division of edge regions and dynamic extraction of multi-temporal vegetation indices, can meticulously capture the temporal variation trend of vegetation cover, providing high-precision quantitative data support for the spatiotemporal dynamic changes of ecosystems. By extracting nighttime light intensity and its stability from nighttime light data and combining it with vegetation indices to construct a composite index, it effectively reflects the intensity of human activities and their spatial distribution characteristics, enhancing the ability to identify both overt and covert disturbances. Through bivariate synergistic analysis of vegetation and nighttime light indices, it reveals the mutual influence and coupling mechanism between vegetation dynamics and human activities, helping to identify the human behavioral drivers behind ecological disturbances and deepening the understanding of ecosystem behavior. Using spatiotemporal pattern clustering technology to classify and identify coupling features, it effectively separates different intervention modes and ecological responses, enhancing the classification accuracy and stability of biological behavioral responses, facilitating the formulation of targeted ecological management and protection strategies. By assessing the edge sensitivity and biodiversity impact of biological behavioral response layers, it can identify vulnerable areas and potential risks in ecosystems, assisting in the scientific determination of ecosystem health and covert threats, and achieving early warning and precise intervention of ecological risks.

[0040] Preferably, the analysis of low-frequency perturbation time-series changes in the explicit human activity area data in step S3 includes:

[0041] Long-term resampling and smoothing of data from areas of visible human activity are performed to extract stable trends and generate baseline data of low-frequency activity changes.

[0042] Remote sensing index sequences are extracted from multi-temporal remote sensing image data and decomposed by wavelet to extract low-frequency perturbation principal components and generate perturbation frequency band feature data.

[0043] By performing time window sliding analysis on the characteristic data of the disturbance frequency band using baseline data of low-frequency activity changes, we can identify low-frequency trend deviations and breakpoints, and generate low-frequency change anomaly detection data.

[0044] Spatial consistency constraints are applied to low-frequency anomaly detection data and explicit human activity area data to filter out short-period high-frequency false interference and generate a steady-state disturbance response layer.

[0045] The steady-state disturbance response layer is classified for disturbance mode and its intensity is calibrated to generate low-frequency disturbance identification data.

[0046] This invention effectively filters out short-period fluctuations and noise through long-term resampling and smoothing, extracting stable and continuous low-frequency activity change baselines, providing reliable basic data for subsequent disturbance analysis. Wavelet decomposition can finely separate low-frequency disturbance components in remote sensing indicators, revealing potential anomalous changes in the ecosystem over time and improving the sensitivity and resolution of disturbance detection. Through sliding window analysis of disturbance frequency band feature data and baseline data, trend deviations and breakpoints can be identified in real time, enabling fine dynamic monitoring of low-frequency anomalous changes and improving the real-time performance and accuracy of time-series anomaly detection. Utilizing the spatial constraints of visible human activity areas, short-period high-frequency noise interference is effectively eliminated, ensuring the spatial coherence and stability of the disturbance response and improving the reliability of low-frequency disturbance identification. Pattern classification and intensity calibration of the steady-state disturbance response layer enable quantitative description of low-frequency disturbances, supporting scientific assessment and management decisions regarding ecosystem health.

[0047] Preferably, step S3, which involves predicting ecosystem health based on low-frequency disturbance identification data and edge effect identification data, includes:

[0048] Spatial semantic fusion is performed on low-frequency disturbance identification data and edge effect identification data to construct a comprehensive ecological disturbance index layer; ecological sensitivity weighted overlay analysis is performed on the comprehensive ecological disturbance index layer to generate multi-factor ecological vulnerability assessment data;

[0049] The multi-factor ecological vulnerability assessment data is divided into datasets to generate a model training set and a model test set;

[0050] The model training set is trained using a long short-term memory neural network algorithm to generate a pre-model for ecosystem health assessment; the pre-model is then optimized and iterated based on the model test set to generate the ecosystem health assessment model.

[0051] Multifactor ecological vulnerability assessment data are input into the ecosystem health assessment model for health assessment prediction, generating ecosystem health assessment prediction data for the target area.

[0052] This invention constructs a comprehensive ecological disturbance index layer by fusing low-frequency disturbance identification data and edge effect identification data through spatial semantic fusion, thereby enhancing the comprehensive perception of multi-dimensional disturbance factors in ecosystems. The comprehensive index layer is weighted and superimposed using ecological sensitivity weights to scientifically reflect the impact of different ecological factors on regional vulnerability, enhancing the accuracy and detail of ecological vulnerability assessment. By dividing the model into training and testing sets and combining the Long Short-Term Memory (LSTM) neural network algorithm for model training and optimization, dynamic learning and accurate prediction of ecosystem health status are achieved. Continuous optimization of the pre-model using the model testing set enhances the model's adaptability to unknown data, ensuring the stability and reliability of ecosystem health assessment results. By inputting multi-factor ecological vulnerability data into the trained ecosystem health assessment model, accurate prediction of the ecological health status of target areas can be made, assisting in ecological environmental protection and management decisions.

[0053] Preferably, step S4 includes the following steps:

[0054] Step S41: Perform factor stratification extraction on the ecosystem health assessment and prediction data to obtain data on potential latent disturbance factors within the region;

[0055] Step S42: Conduct a regional comprehensive analysis of the ecosystem of the target area based on the data of potential latent disturbance factors in the area to extract data of sensitive areas with latent threats. The regional comprehensive analysis includes land use, vegetation continuity and water connectivity analysis.

[0056] Step S43: Compare the fluctuations of sensitive area data over time to identify the trend of latent disturbances, generate a threat disturbance trend layer, and overlay the threat disturbance trend layer with the ecosystem health assessment and prediction data to divide the latent threat level ranges of different intensities.

[0057] Step S44: Geospatial mapping and distribution range definition of the hidden threat level range, generating a hidden threat level and distribution map of the target area.

[0058] This invention utilizes factor-layered extraction technology to effectively capture potential latent disturbance factors within a region, enhancing the depth and detail of identifying complex threats to ecosystems. By combining land use, vegetation continuity, and water connectivity analysis, it enables the scientific extraction of sensitive areas with latent threats to ecosystems, strengthening the spatial relevance and ecological rationality of latent threat identification. Through time-series fluctuation analysis of sensitive area data, it dynamically identifies the trends of latent disturbances, enhancing the understanding and mastery of the spatiotemporal evolution of latent threats to ecosystems. By overlaying threat disturbance trend layers with ecosystem health assessment and prediction data, and combining multi-source information for comprehensive analysis, it achieves quantitative classification of latent threat levels, improving the accuracy and practicality of threat assessment. Through geospatial mapping and boundary delineation, it generates intuitive latent threat levels and distribution maps, facilitating risk monitoring, early warning, and ecological protection decision support for ecological management departments.

[0059] This specification provides a remote sensing identification system for ecosystem surveys, used to perform the aforementioned remote sensing identification method for ecosystem surveys. The remote sensing identification system for ecosystem surveys includes:

[0060] The multi-temporal remote sensing identification module is used to acquire multi-temporal remote sensing image data of the target area; extract the land cover type of the multi-temporal remote sensing image data; and identify the visible human activity area of ​​the multi-temporal remote sensing image data according to the land cover type, thereby generating visible human activity area data.

[0061] The nighttime light remote sensing identification module is used to acquire nighttime light data of the target area; to perform spatiotemporal registration of the nighttime light data of the target area and multi-temporal remote sensing image data to generate fused nighttime light remote sensing data; to extract the boundary areas of the multi-temporal remote sensing image data by land cover type to obtain edge area data; and to calculate the vegetation index and nighttime light index of the edge area data based on the fused nighttime light remote sensing data.

[0062] The correlation analysis module is used to perform biological behavior correlation analysis on edge area data through vegetation index and night light index to generate edge effect identification data; analyze the time series changes of low-frequency disturbances in data of areas with obvious human activities to generate low-frequency disturbance identification data; and perform ecosystem health assessment prediction based on low-frequency disturbance identification data and edge effect identification data to generate ecosystem health assessment prediction data for the target area.

[0063] The ecosystem assessment module is used to output regional latent threats to the ecosystem of the target area based on the ecosystem health assessment and prediction data of the target area, thereby obtaining the latent threat level and distribution map.

[0064] The beneficial effects of this invention are that the multi-temporal remote sensing identification module can realize the dynamic acquisition and comprehensive analysis of multi-temporal remote sensing images of the target area, improving the temporal continuity and spatial accuracy of land cover type identification; it combines land cover type to identify areas of visible human activity, enhancing the spatial positioning and monitoring capabilities of human activity impacts, and providing a reliable data foundation for subsequent ecological analysis. The nighttime light remote sensing identification module achieves deep fusion of multi-source remote sensing data through precise spatiotemporal registration of nighttime light data and multi-temporal remote sensing images, improving the extraction accuracy of vegetation and nighttime light features in edge areas; it effectively extracts image boundary change areas using land cover type, accurately acquiring ecological edge area data and enhancing the scientific rigor of edge effect identification. The correlation analysis module comprehensively reveals the interaction between human activities and vegetation dynamics in the ecosystem through biological behavior correlation analysis of vegetation indices and nighttime light indices, enhancing the spatiotemporal understanding of ecological disturbances. Combined with low-frequency disturbance time-series variation analysis of overt human activity area data, it achieves accurate identification of long-term ecosystem disturbance trends. Based on multidimensional disturbance data, it performs ecosystem health assessment and prediction, improving the scientific rigor and predictability of ecological status assessments and providing decision support for ecological protection and restoration. The ecosystem assessment module, relying on ecosystem health assessment and prediction data, outputs regional latent threats, enabling in-depth identification and hierarchical management of potential ecological risks. Through the generation of latent threat levels and distribution maps, it provides intuitive and scientific spatial risk distribution information, facilitating ecological environment management and precise intervention. Therefore, this invention improves the accuracy and comprehensiveness of remote sensing identification in ecosystem surveys through multi-temporal, multi-source data fusion and multi-indicator time-series analysis. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the steps involved in a remote sensing identification method applied to ecosystem surveys.

[0066] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S2.

[0067] Figure 3 for Figure 1 A detailed flowchart illustrating the implementation steps of step S4.

[0068] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0069] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0070] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0071] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0072] To achieve the above objectives, please refer to Figures 1 to 3 A remote sensing identification method for ecosystem surveys, the method comprising the following steps:

[0073] Step S1: Acquire multi-temporal remote sensing image data of the target area; extract the land cover type from the multi-temporal remote sensing image data, and identify the visible human activity area based on the land cover type to generate visible human activity area data.

[0074] Step S2: Acquire nighttime light data of the target area; perform spatiotemporal registration of the nighttime light data of the target area and multi-temporal remote sensing image data to generate fused nighttime light remote sensing data; extract the boundary areas of the multi-temporal remote sensing image data by land cover type to obtain edge area data; calculate the vegetation index and nighttime light index of the edge area data based on the fused nighttime light remote sensing data.

[0075] Step S3: Perform biological behavior correlation analysis on edge area data using vegetation index and nighttime light index to generate edge effect identification data; analyze the time series changes of low-frequency disturbances in data of areas with visible human activities to generate low-frequency disturbance identification data; perform ecosystem health assessment prediction based on low-frequency disturbance identification data and edge effect identification data to generate ecosystem health assessment prediction data for the target area.

[0076] Step S4: Based on the ecosystem health assessment and prediction data of the target area, output the regional latent threats to the ecosystem of the target area, thereby obtaining the latent threat level and distribution map.

[0077] This invention effectively enhances the collaborative processing capabilities of multi-source heterogeneous data in terms of spatial resolution and temporal consistency by spatiotemporal registration and fusion of multi-temporal remote sensing images and nighttime light data. It extracts visible human activity areas based on land cover type and combines vegetation and nighttime light indices in boundary areas for edge effect analysis, enhancing the ability to identify the spatial extent of human activities and areas with ambiguous boundaries. It performs biological behavior correlation analysis using vegetation and nighttime light change indices in edge areas and combines this with low-frequency disturbance time-series data from visible areas to achieve deep coupling analysis between dynamic ecological responses and human activity disturbances. Using edge effect identification data and low-frequency disturbance identification data as core input features, an ecosystem health prediction model is established, exhibiting strong data-driven characteristics and model scalability. Utilizing ecosystem health prediction data, a latent threat level classification mechanism is constructed and a spatial distribution map is output, effectively improving the accuracy and visualization capabilities of ecological threat identification. Therefore, this invention improves the accuracy and comprehensiveness of remote sensing identification in ecosystem surveys through multi-temporal, multi-source data fusion and multi-index time-series analysis.

[0078] In this embodiment of the invention, reference Figure 1 The diagram shown illustrates the steps of a remote sensing identification method for ecosystem surveys according to the present invention. In this example, the remote sensing identification method for ecosystem surveys includes the following steps:

[0079] Step S1: Acquire multi-temporal remote sensing image data of the target area; extract the land cover type from the multi-temporal remote sensing image data, and identify the visible human activity area based on the land cover type to generate visible human activity area data.

[0080] Step S2: Acquire nighttime light data of the target area; perform spatiotemporal registration of the nighttime light data of the target area and multi-temporal remote sensing image data to generate fused nighttime light remote sensing data; extract the boundary areas of the multi-temporal remote sensing image data by land cover type to obtain edge area data; calculate the vegetation index and nighttime light index of the edge area data based on the fused nighttime light remote sensing data.

[0081] Step S3: Perform biological behavior correlation analysis on edge area data using vegetation index and nighttime light index to generate edge effect identification data; analyze the time series changes of low-frequency disturbances in data of areas with visible human activities to generate low-frequency disturbance identification data; perform ecosystem health assessment prediction based on low-frequency disturbance identification data and edge effect identification data to generate ecosystem health assessment prediction data for the target area.

[0082] Step S4: Based on the ecosystem health assessment and prediction data of the target area, output the regional latent threats to the ecosystem of the target area, thereby obtaining the latent threat level and distribution map.

[0083] In this embodiment of the invention, NPP-VIIRS nighttime light remote sensing data with the same time span as in step S1 is acquired, with a monthly temporal accuracy and a spatial resolution of 500 meters. The VIIRS data is resampled using the nearest neighbor interpolation method to achieve a resolution of 10 meters, ensuring spatial consistency with Sentinel-2 data. Spatial registration is performed using the gdalwarp tool from the GDAL library in Python. Control points are manually selected from six stable, invariant ground feature points for correction, with a correction error not exceeding 0.3 pixels. Temporal registration is achieved by taking the average VIIRSDN of three months within each quarter, ensuring consistency with the quarterly remote sensing image time points. Edge region extraction is achieved by constructing a classification boundary for land cover. The Canny edge detection algorithm is used, with threshold parameters set to a low threshold of 0.05 and a high threshold of 0.15, to extract the boundaries between built and non-built areas. Subsequently, a buffer analysis operation is applied, setting the buffer radius to ±150 meters to form a 300-meter-wide edge buffer zone. The vegetation index (NDVI) is calculated using the formula: NDVI = (B8 - B4) / (B8 + B4), where B8 is the near-infrared band and B4 is the red band. NDVI results are normalized to between 0 and 1 and extracted using edge region masks. Nightlight index is calculated using normalized VIIRSDN values, NLI = DN / 63, where the maximum DN value is 63. All processed results are output in GeoTIFF format with an accompanying WGS 84 spatial reference system. Edge region data is provided by the buffer output of step S2, with a spatial unit of 10 meters. A two-dimensional matrix of NDVI and NLI is constructed for all pixels within this region. A 5×5 sliding window (50m × 50m actual area) is constructed for local statistical processing, calculating the mean NDVI and mean NLI within each window. Pearson correlation analysis is used to obtain the correlation coefficient r between NDVI and NLI, and regions with |r| > 0.6 are selected as significantly behaviorally associated regions. All sliding window results are reconstructed into a new GeoTIFF format edge effect recognition map, with each pixel value being the r-value at that location. For data on visible human activity areas, the NDVI time series of their center points is extracted. Daubechies wavelet (db4) is used for wavelet decomposition from order 1 to 4 to extract low-frequency components with a period greater than one year. The standard deviation σ of this component is calculated, and areas meeting the condition are marked as low-frequency perturbation areas with σ > 0.12 as the perturbation threshold. The perturbation map is output in 10-meter resolution GeoTIFF format. Finally, the edge effect recognition map and the low-frequency perturbation map are superimposed pixel-wise, using the following ecological health index assessment model: H = 1 - (0.5·σNDVI + 0.3·NLI + 0.2·rabs), where σNDVI is the standard deviation of the low-frequency perturbation, NLI is the nighttime light index, and rabs is the absolute correlation coefficient between NDVI and NLI. The index normalization range is set to [0,1], and a 10-meter resolution health index layer is output.Based on the ecosystem health assessment map (health index layer) generated in S3, a latent threat level classification operation was performed on the target area. According to the health index value H, the following classification criteria were adopted: H ≥ 0.8 is Level I (no threat), 0.6 ≤ H < 0.8 is Level II (mild threat), 0.4 ≤ H < 0.6 is Level III (moderate threat), and H < 0.4 is Level IV (severe threat). During the classification process, the RasterCalculator tool was used to assign values ​​from 1 to 4 according to the above intervals. All results were output in GeoTIFF format with accompanying legends. To generate a regional spatial distribution map, the "Reclassify" tool in ArcGIS was used for layer coloring, setting color codes: green for Level I, yellow for Level II, orange for Level III, and red for Level IV. The "Fishnet" grid tool was used to construct 100m × 100m analysis cells, and the area proportion of each threat level in each grid was statistically analyzed to generate a raster heat map with a resolution of 100 meters. The final output includes a hidden threat level distribution map (GeoTIFF format), a grid statistical heatmap (Shapefile format), and a threat level spatial proportion table (CSV format).

[0084] Preferably, step S1 includes the following steps:

[0085] Step S11: Perform multi-phase sampling control of remote sensing images on the target area to obtain multi-temporal remote sensing image data;

[0086] Step S12: Perform radiometric and geometric corrections on the multi-temporal remote sensing image data to generate corrected multi-temporal remote sensing image data;

[0087] Step S13: Extract the surface spectral features of the corrected multi-temporal remote sensing image data, and perform temporal variation analysis and classification processing on the surface spectral features to generate surface cover type data;

[0088] Step S14: Interpret human activity indicators from land cover type data to generate preliminary human activity mask data; perform spatial consistency screening and boundary correction on the preliminary human activity mask data to generate explicit human activity area data.

[0089] In this embodiment of the invention, multi-phase sampling control of remote sensing images of the target area requires determining the sampling frequency and sampling period based on the temporal variation characteristics of the ecosystem. Taking the East Asian monsoon region as an example, remote sensing image data for the three typical seasons of spring, summer, and autumn each year should be acquired to cover the growth, flourishing, and decline periods of the ecosystem. The remote sensing image data source is Sentinel-2 MSI remote sensing data with a resolution of no less than 10 meters, covering the entire target area and ensuring a cloud cover filtering rate of no less than 95%. Multi-phase sampling adopts a monthly download method, using image date intervals (e.g., 10 days ± 3 days per month), spatial overlap greater than 95%, and cloud cover less than 5% as filtering conditions to download data from platforms such as ESA Copernicus Open Access Hub. The downloaded raw remote sensing images are organized according to naming rules and a multi-temporal data directory structure is established, sorted by timestamp to form a continuous temporal remote sensing data stack for subsequent processing. Radiometric correction was performed based on the sensor radiometric response (DN) values ​​of the original remote sensing imagery, using the SEN2COR tool for atmospheric correction and surface reflectance conversion. Parameters such as solar altitude angle, sensor observation angle, surface atmospheric model (e.g., mid-latitude summer atmospheric model), and aerosol type (continental type) were set, and radiometric conversion was performed uniformly. Geometric correction was performed based on high-precision geographic control point (GCP) data and DEM data. GCP data was matched against the coordinates of control points marked on national 1:50000 topographic maps, and a least-squares fitting algorithm was used for correction, combined with SRTM 30-meter resolution DEM for topographic distortion compensation. All corrected images were uniformly output in UTM projection format under the WGS84 coordinate system (specific zone numbers determined by the image center longitude), with spatial resolution uniformly resampled to 10 meters, ensuring that the overlap error of all image coverage areas within the boundary lines was less than 1 pixel. Surface spectral feature extraction employed a combination of principal component analysis (PCA) and vegetation index calculation. Using remote sensing imagery from each time phase as input, NDVI (Normalized Difference Vegetation Index), NDBI (Normalized Difference Building Index), and NDWI (Normalized Difference Water Index) are calculated. The specific formulas are: NDVI = (NIR - RED) / (NIR + RED), NDBI = (SWIR - NIR) / (SWIR + NIR), and NDWI = (GREEN - NIR) / (GREEN + NIR). Here, NIR represents the near-infrared band, RED represents the red band, SWIR represents the shortwave infrared band, and GREEN represents the green band. The band positions are based on the Sentinel-2MSI sensor specifications: Band 8 (NIR), Band 4 (RED), Band 11 (SWIR), and Band 3 (GREEN).Subsequently, NDVI sequence data was extracted according to pixel time series. The maximum value synthesis (MVC) strategy was used to obtain the NDVI peak layer for each season, and K-means clustering (k=5) was performed to classify land cover types into five categories: grassland, woodland, water bodies, bare land, and built-up areas. During classification, NDVI thresholds were set (e.g., NDVI>0.5 for woodland, NDVI<0.2 for bare land), and temporal consistency was assessed, removing pixels showing two or more major category changes in a year (determined as unstable areas). Human activity indicators were interpreted from the land cover type data, primarily through a mask screening based on the triple conditions of NDBI >0.2, NDVI <0.3, and NDWI <0, constructing preliminary human activity mask data. The thresholds were determined based on measured sample analysis; in the sample set, the NDBI threshold was set at the 95th percentile of pixels in the urban core area, while NDVI and NDWI were set based on seasonal minimum values. After initial mask data extraction, spatial consistency screening is performed, retaining connected regions with an area greater than 0.5 square kilometers and discarding regions smaller than this threshold as isolated noise. Boundary correction employs a boundary buffer fusion operation, specifically extending boundary pixels by 5 pixels to form a boundary buffer, applying a multi-value voting rule to determine their affiliation, and adjusting excessively fragmented boundary areas. The final output is data on explicit human activity areas, maintaining a spatial resolution of 10 meters, with a coordinate system consistent with the corrected image, and output in GeoTIFF format for direct loading and processing by subsequent analysis modules.

[0090] Preferably, step S13 includes the following steps:

[0091] Step S131: Extract multi-band reflectance values ​​of surface pixels at different time phases from the corrected multi-temporal remote sensing image data to obtain a surface spectral feature sequence with time labels;

[0092] Step S132: Calculate the temporal variation trend of key band reflectance in multi-temporal remote sensing image data based on the surface spectral feature sequence, and determine whether its variation pattern meets any of the following conditions and mark it accordingly to generate temporal variation judgment label data: NDVI index has seasonal periodic fluctuations and the annual amplitude is greater than 0.3; the position of the crossover point of blue light band and red edge band reflectance changes by more than ±15nm in three consecutive temporal phases; the number of inflection points of the spectral curve varies by more than 2 in different growth stages;

[0093] Step S133: Use time-series change judgment marker data to identify the land cover type based on the spectral characteristics of the land surface. If any of the following conditions are met, it can be determined as the specified land cover type and land cover type data can be generated: If the maximum NDVI value exceeds 0.6 and the annual fluctuation range of NDVI is greater than 0.35, it is classified as "crops"; if the shortwave infrared reflectance changes by less than 5% throughout the year and the NDVI is consistently below 0.2, it is classified as "built surface"; if the normalized difference index of red light and near-infrared light shows a double peak throughout the year and the peak values ​​correspond to spring and autumn, respectively, it is classified as "deciduous woodland".

[0094] In this embodiment of the invention, the reflectance values ​​of each pixel are extracted from the multi-temporal remote sensing image data after radiometric and geometric corrections have been completed in step S12. Ten main bands of Sentinel-2 MSI data are used, including blue light (Band 2, center wavelength 490nm), green light (Band 3, 560nm), red light (Band 4, 665nm), red edge (Band 5, 705nm; Band 6, 740nm; Band 7, 783nm), near-infrared (Band 8, 842nm), and shortwave infrared (Band 11, 1610nm; Band 12, 2190nm). Using geographic coordinates as an index, the reflectance values ​​of the aforementioned bands are extracted for each pixel across all temporal images, constructing a surface spectral feature sequence with the following structure: {Pixel ID:(t1,[R_b2,R_b3,...,R_b12]),(t2,[R_b2,R_b3,...,R_b12]),...,(tn,[R_b2,R_b3,...,R_b12])}, where ti represents the i-th timestamp and R_bj is the apparent reflectance of band j. All temporal data are archived in chronological order, with time tags recorded using the Julian Day encoding of the observation date, forming a spectral sequence with standard time attributes to ensure that each pixel has complete intra-annual spectral variation characteristics. Based on the spectral feature sequence extracted in step S131, the NDVI value sequence and the blue and red edge band reflectance sequences for each pixel are calculated sequentially, and temporal variation analysis is performed. The NDVI value is calculated using the formula: NDVI = (Band 8 - Band 4) / (Band 8 + Band 4). For each pixel, the maximum and minimum NDVI values ​​for the year are statistically analyzed, and the amplitude (maximum value - minimum value) is calculated to determine if it is greater than 0.3. Simultaneously, a Fast Fourier Transform (FFT) is used to detect seasonal periodic fluctuations. If the period of the principal component in the frequency domain is within the range of 300–370 days, it is marked as a significant seasonal fluctuation. For the red-edge band (Band 5–7) and the blue band (Band 2), linear interpolation is used to fit the spectral curves, and the wavelength position of their intersection is calculated. The intersection point is defined as the wavelength point where the reflectance curves of the two bands intersect. By reconstructing the curves every 1 nm within the range of 705 nm ± 30 nm, the wavelength corresponding to the reflectance intersection point is determined. Comparisons are made at three consecutive time points. If the change in the intersection point position exceeds ± 15 nm, it is marked as a significant change in the red-edge blue reflectance intersection. The number of inflection points in the spectral curve is statistically analyzed using the second derivative method. For each time phase, the second difference sequence of the spectral reflectance sequence is calculated, and the number of extreme points is used as the inflection point count index. If the difference between the maximum and minimum number of inflection points within a year exceeds 2, it is marked as a significant structural change.Finally, the above three judgment conditions are used to generate a three-dimensional logical vector (e.g., [1,0,1]), which is the "temporal change judgment marker data" of the pixel. The "temporal change judgment marker data" obtained in step S132 is jointly classified with the specific values ​​of NDVI and SWIR bands in step S131, and the land cover type is determined according to the following rules: For the NDVI sequence, if the maximum NDVI value in the year is greater than 0.6, and the difference between the maximum and minimum NDVI values ​​(intra-year amplitude) exceeds 0.35, and the seasonal fluctuation term of NDVI in the corresponding temporal change marker is 1 (i.e., there is a seasonal cycle), then the pixel is classified as "crops". A year-round time series analysis was performed on shortwave infrared reflectance (Band 11 and Band 12), calculating the difference between its maximum and minimum values. If the annual variation was less than 5% (i.e., the difference between the maximum and minimum values ​​was within 0.05), and the maximum NDVI was less than 0.2, with no significant fluctuations in NDVI throughout the year, and the "inflection point difference" item in its change marker was 0 (no obvious structural change), then the pixel was classified as "built-up land". Normalized difference sequences derived from NDVI were calculated for the red band (Band 4) and near-infrared band (Band 8), detecting whether they exhibited two peaks within the year. The peak locations were identified using extreme value analysis. If the two main peaks appeared between March and November, the peak difference was greater than 0.2, and both the "NDVI seasonal fluctuation" and "inflection point change" items in its change marker were 1, then it was classified as "deciduous woodland". Finally, each pixel was judged once according to the above conditions, and a land cover type label was output if any set of conditions was met. The classification results are output in GeoTIFF format, with different types encoded by value range in the single-band raster (e.g., 1 for crops, 2 for built-up areas, 3 for deciduous woodlands, etc.) for subsequent interpretation.

[0095] As an example of the present invention, reference is made to... Figure 2 As shown, step S2 in this example includes:

[0096] Step S21: Obtain nighttime light data for the target area;

[0097] Step S22: Perform radiation normalization and noise suppression on the raw nighttime light radiation data to generate net nighttime light radiation data; align the net nighttime light radiation data with the multi-temporal remote sensing image correction data to generate a spatiotemporal registration parameter set.

[0098] Step S23: Based on the spatiotemporal registration parameter set, perform raster-level registration and resampling fusion of nighttime light data and multi-temporal remote sensing image data of the target area to generate fused nighttime light remote sensing data; use land cover type data to perform edge detection and morphological region growth analysis on the boundary change areas of the remote sensing image to generate boundary area layers;

[0099] Step S24: Spatial extraction and segmentation of the boundary area layer to generate edge area data; vegetation index calculation of the edge area data based on the fused nighttime light remote sensing data to generate edge area vegetation index data;

[0100] Step S25: Extract the night light index from the edge area data based on the fused night light remote sensing data to generate edge area night light index data.

[0101] In this embodiment of the invention, by selecting VIIRSDNB (Visible Infrared Imaging Radiometer Suite Day / Night Band) nighttime light remote sensing image data, a nighttime light image sequence matching the time range of the multi-temporal remote sensing image is obtained. Monthly average nighttime light images can be downloaded from the NOAA Earth Observation Group platform; the spatial resolution is approximately 500m, the data format is GeoTIFF, and the value range unit is nW / cm². 2 / sr. The nighttime light imagery is cropped to a spatial range consistent with the remote sensing imagery using spatial clipping operations (e.g., gdalwarp), and the data is converted into a floating-point single-band layer for subsequent radiometric correction and fusion. The VIIRS nighttime light data undergoes min-max normalization to map the value range uniformly to between 0 and 1. The formula is: L_norm = (L_raw - L_min) / (L_max - L_min); where L_raw is the original nighttime light radiance, L_min is the minimum value within the region (statistic after removing background values), and L_max is the 95th percentile (excluding abnormally bright values). Gaussian noise is suppressed using median filtering (3x3 window) or wavelet denoising (e.g., Haar wavelet), preserving the structure of continuous illumination areas and removing discrete light spots. Basic geographic reference points (such as urban road intersections, water boundaries, and areas of high nighttime light intensity) consistent with the nighttime light data are extracted from multi-temporal remote sensing images. Co-location points are extracted using geographic location similarity algorithms (such as SIFT or ORB feature point matching), and the affine registration matrix (or TPS transformation) is calculated. The output is a "spatiotemporal registration parameter set" containing affine parameters, coordinate offsets, and time indices. For nightlight data and multi-temporal remote sensing image data, gdalwarp is used to perform a unified coordinate system transformation based on the spatiotemporal registration parameter set. A bilinear interpolation resampling method is used to adjust the nightlight data to a spatial resolution consistent with the remote sensing image (e.g., 10m or 30m). The nightlight layer is used as a new band and overlaid onto the multi-band remote sensing image to generate fused nightlight-fused remote sensing data (NFRSD). Using the generated land cover type data, boundary combination areas such as "crop-building surface" or "crop-deciduous woodland" are extracted. Edge detection is performed using the Sobel or Canny operator to extract abrupt changes in land cover type edges. Morphological region growing algorithms (such as grayscale region expansion + closing operation) are applied to generate a spatially expandable "boundary region layer". Vectorization operations (such as gdal_polygonize) are used on the boundary region layer to extract planar edge regions, generating shapefile-formatted vector edge data. Regions are divided and small-area error regions are filtered by combining attribute information (such as boundary category and spectral variation). Red light bands (Band 4) and near-infrared bands (Band 8) are extracted from the fused nighttime light remote sensing data. The standard NDVI formula is used: NDVI = (NIR - Red) / (NIR + Red); the average NDVI value of each pixel is calculated in each edge region to generate "edge region vegetation index data" (one vegetation index value per region). Nighttime light bands (newly added bands) are extracted from the fused nighttime light remote sensing data.The "Noctilucaluminum Index (NLI)" is defined as the average value of the luminous band within the region, or calculated using the normalized rate of variability in luminous intensity: NLI = mean(L_night) / max(L_night_in_region). For each edge region, its luminous index is statistically analyzed, and spatiotemporal comparative analysis is performed in conjunction with land cover type information. The final generated vector boundary file includes luminous index attributes, supporting further applications such as urban edge development intensity assessment and land use zoning identification.

[0102] Preferably, step S23, which utilizes land cover type data to perform edge detection and morphological region growing analysis on the boundary change areas of the remote sensing image, includes:

[0103] Edge extraction was performed on the boundary change area of ​​the remote sensing image using land cover type data. The high threshold was set to 100 pixels of gray value and the low threshold was set to 50 pixels of gray value. A 3×3 Gaussian kernel was used for smoothing to obtain the outline data with clear boundaries, which was then converted into vector polygon format.

[0104] Using contour line data as seed regions, a region growing algorithm is used to expand to regions with similar land features. The spectral similarity threshold is set to ±10% reflectance, and the area threshold is greater than or equal to 1000m². 2 The growth radius is within 90 meters, and the growth results are re-marked and rasterized into boundary areas;

[0105] The boundary area and the remote sensing image boundary change area are overlaid to generate a vector layer; each boundary area unit is assigned the following attribute fields: change type, change area, boundary length, and change time period, thereby outputting the boundary area layer.

[0106] In this embodiment of the invention, land cover type data in raster format is input (e.g., each pixel value represents a land use code: farmland = 1, building = 2, forest = 3, etc.). Current and historical land cover data are overlaid, and the changed areas are extracted as boundary change areas. A 3×3 Gaussian kernel is applied to the overlaid layer for blurring to eliminate small-scale noise and pseudo-change edges. Example Gaussian kernel: [[1,2,1],[2,4,2],[1,2,1]] / 16. A dual-threshold Canny edge detection algorithm is used to extract contour lines, with the following settings: high threshold: 100 grayscale value; low threshold: 50 grayscale value. The output is a pixel-level contour line image reflecting abrupt changes in land use boundaries. A contour tracking algorithm (such as findContours in OpenCV) is used to convert the edge lines into vector polygons. Simplification processing (Douglas–Peucker algorithm, threshold set to 2 pixels) is performed to reduce the number of vertices and improve subsequent analysis efficiency. The resulting "contour line vector data" is the initial boundary seed region. The outline vectors were converted into a raster seed layer and combined with multi-band data from fused remote sensing imagery. The growth conditions were set as follows: spectral similarity threshold: ±10% reflectance range (compare NDVI or blue / red / near-infrared bands to ensure ground cover consistency); area threshold: ≥1000m² 2 Small patch interference is removed; growth radius: ≤90 meters, meaning the expansion from the seed point does not exceed 90 meters. The growth results are reassigned as independent boundary area codes to maintain regional closure. Connectivity markers (4 or 8 neighborhoods) are used to ensure each growth area is an independent isometric unit. The output is rasterized to generate a "boundary area raster layer". A Boolean intersection operation is performed between the boundary area layer and the original remote sensing image boundary change area layer, retaining only the overlapping area. A final vector layer of the boundary area is generated using GIS overlay analysis methods (such as Intersect or gdal_rasterize+gdal_polygonize in ArcGIS). For each boundary area unit, the following attribute fields are calculated and assigned: Change type (change_type): assigned according to the combination of changed land types, such as "farmland → building"; Change area (area_m2): calculated from the vector isometric data, the actual area of ​​each polygon (unit: m²). 2 ); Boundary length (boundary_length): Calculates the total length of the polygon's boundary (unit: m); Change period (change_period): Combined with the image phase index, assigned a value such as "2021Q2–2023Q1". The output is a standard vector layer (such as Shapefile or GeoJSON format), supporting overlay analysis, chart statistics, and 3D display.

[0107] Preferably, step S3, which involves performing biological behavior correlation analysis on the edge area data using vegetation index and nighttime light index, includes:

[0108] Spatial grid division is performed on edge region data, and vegetation index changes are extracted by combining time-series remote sensing images to generate edge vegetation dynamic index data;

[0109] By overlaying nighttime light remote sensing images onto edge region data, nighttime light intensity and stability indices are extracted to generate edge nighttime light index data.

[0110] A bivariate spatiotemporal co-analysis of vegetation index and night light index was performed to generate vegetation-night light coupling characteristic data;

[0111] Spatiotemporal pattern clustering was performed on vegetation-night light coupling feature data to identify the coupling effect of human activity intervention and ecological disturbance, and to generate biological behavior response layer data.

[0112] Edge sensitivity analysis and biodiversity impact assessment are performed on biological behavior response layer data to generate edge effect identification data.

[0113] In this embodiment of the invention, edge region data is divided into regular grid units according to a certain spatial resolution (e.g., 500m × 500m or 1km × 1km) to form an edge region spatial grid system. Multi-temporal remote sensing images (e.g., MODIS, Landsat, Sentinel-2, etc.) are collected to cover the edge region, ensuring the continuity and integrity of the time series. Vegetation indices (e.g., NDVI, EVI, etc.) are calculated for the remote sensing images at each time point to obtain the vegetation index value for each spatial grid unit. Combined with the time series, the temporal variation characteristics of the vegetation index of each grid unit (e.g., seasonal variation trends, anomalous fluctuations) are extracted to generate an edge vegetation dynamic index dataset. The spatial grid of the edge region is spatially overlaid with nighttime light remote sensing data (e.g., DMSP-OLS or VIIRS data) for the corresponding time period. The nighttime light radiation intensity of each grid unit is statistically analyzed to obtain a nighttime light index representing the intensity of human activity. Stability indicators of the nighttime light index (e.g., time series standard deviation, fluctuation coefficient, etc.) are calculated to distinguish between stable and fluctuating human activities. An edge nighttime light index dataset containing nighttime light intensity and stability indicators is formed. The vegetation index and the noctilucent index were normalized to eliminate dimensional differences, ensuring temporal and spatial synchronization. Bivariate spatiotemporal correlation analysis (e.g., spatiotemporal cross-correlation function) was used to assess the coordinated changes of vegetation and noctilucent at different temporal and spatial scales. A coupling coordination model was employed to calculate the coupling strength and coordination degree between the vegetation index and the noctilucent index, quantifying their relationship. A spatiotemporal dataset describing the coupling characteristics of vegetation and noctilucent was created, reflecting the interaction between ecology and human activities. Based on the vegetation-noctilucent coupling characteristic data, typical spatiotemporal pattern features (e.g., periodicity, trends, abrupt change points) were extracted. Spatiotemporal clustering algorithms (e.g., spatiotemporal K-means, DBSCAN, spatiotemporal spectral clustering) were used for pattern recognition. Areas of strong human intervention (high noctilucent, low vegetation coupling pattern) and areas of ecological disturbance (abnormal vegetation decline accompanied by noctilucent changes) were identified. A partitioned biological behavior response layer was generated to identify the spatial distribution of different coupling effects. Spatial statistical analysis was performed on the biological behavior response layer to identify grid cells in the edge regions sensitive to human activities or ecological changes. Spatial weighting matrices are applied to analyze the spatial diffusion and impact range of edge effects. Combined with existing biodiversity monitoring data or model predictions, the impact of coupling effect zones on species diversity and ecosystem function is assessed. A multi-indicator comprehensive evaluation method (such as species richness, species evenness, and ecological function index) is used to classify the impact levels. A comprehensive dataset reflecting the impact and sensitivity of biodiversity in edge areas is output for subsequent ecological protection or intervention decisions.

[0114] Of particular importance, spatiotemporal pattern clustering of vegetation-night light coupling characteristic data to identify the coupling effects of human intervention and ecological disturbance also includes:

[0115] Spatiotemporal features of vegetation-night light coupling data were extracted, and spatiotemporal pattern recognition was performed using density peak clustering algorithm to generate cluster label data;

[0116] Spatiotemporal distribution mapping is performed on cluster label data to construct a spatiotemporal pattern clustering layer;

[0117] The coupling effect index between vegetation characteristics and nighttime light intensity was calculated using statistical correlation analysis, and a coupling effect coefficient matrix was generated. The coupling effect was comprehensively evaluated by combining the spatiotemporal pattern clustering layer and the coupling effect coefficient matrix, and biological behavioral response intensity distribution data was generated.

[0118] The distribution data of biological behavioral response intensity is mapped to a spatial coordinate system to generate biological behavioral response layer data.

[0119] In this embodiment of the invention, edge region data is spatially gridded, and vegetation indices (such as NDVI) are extracted over time from long-term remote sensing images to generate edge vegetation dynamic index data. Simultaneously, nighttime light remote sensing images are overlaid with edge region data to extract indicators such as nightlight intensity and nightlight stability, generating edge nightlight index data. Based on this, a bivariate spatiotemporal co-analysis of the vegetation index and nightlight index is performed to construct a coupled multidimensional feature set and generate vegetation-nightlight coupling feature data. Subsequently, the main spatiotemporal features of the vegetation-nightlight coupling feature data are extracted, including dynamic features such as mean, trend, volatility, periodicity, and abrupt change points in the time series, as well as indicators such as spatial distribution concentration, expansion, spatial gradient, and local spatial autocorrelation. Based on this, the Density Peaks Clustering (DPC) algorithm is used to cluster and identify high-dimensional spatiotemporal coupling features. By calculating the local density and relative distance between samples, significantly different coupling patterns are identified, generating cluster label data. Cluster labels are mapped to geospatial data to construct a spatiotemporal pattern clustering layer, reflecting the distribution patterns and evolution trends of different coupling patterns within the region. Subsequently, statistical correlation analysis methods (such as Pearson correlation coefficient, mutual information, and canonical correlation analysis) are used to calculate the synergistic relationship between vegetation change and nocturnal light change in each cluster category, generating a coupling effect coefficient matrix. Further, this coefficient matrix is ​​combined with the clustering layer to conduct a comprehensive assessment of the coupling effect, forming data on the distribution of biological behavioral response intensity. This data reflects the synergistic mechanism and spatiotemporal differences in strength between changes in human activity intensity and ecological disturbance response. Finally, the biological behavioral response intensity distribution data is mapped to a unified spatial coordinate system to generate high-resolution biological behavioral response layer data, providing spatial support for subsequent edge sensitivity analysis, biodiversity risk assessment, and ecosystem health monitoring.

[0120] Of particular importance, the comprehensive evaluation of coupling effects by combining the spatiotemporal pattern clustering layer and the coupling effect coefficient matrix also includes:

[0121] For each clustering unit in the spatiotemporal pattern clustering layer, the corresponding coupling effect coefficient matrix subset is extracted to generate local coupling index data; based on the local coupling index data, a weighted fusion algorithm is used to integrate the coupling strength of multiple factors to generate a comprehensive coupling strength score.

[0122] Spatial interpolation is performed on the comprehensive coupling strength score to fill the uncovered areas of the spatiotemporal pattern clustering layer and generate a continuous coupling strength distribution map.

[0123] Threshold segmentation technology was used to identify regions with significant coupling strength and to determine hotspots of high-response biological behavior.

[0124] Spatiotemporal variation distribution analysis of hotspots of high-response biological behavior is performed to generate data on the distribution of biological behavior response intensity.

[0125] In this embodiment of the invention, edge region data is spatially gridded, and vegetation indices (such as NDVI) are extracted over time from long-term remote sensing images to generate edge vegetation dynamic index data. Simultaneously, nighttime light remote sensing images are overlaid with edge region data to extract indicators such as night light intensity and night light stability, generating edge night light index data. Based on this, a bivariate spatiotemporal co-analysis of the vegetation index and night light index is performed to construct a coupled multidimensional feature set, generating vegetation-night light coupled feature data. Subsequently, the main spatiotemporal features of the coupled feature data are extracted, including dynamic features such as mean, trend, volatility, periodicity, and abrupt change points in the time series, as well as indicators such as spatial distribution concentration, spatial gradient, and local spatial autocorrelation. Based on this, the Density Peaks Clustering (DPC) algorithm is used to cluster and identify high-dimensional spatiotemporal coupled features. Through analysis of local density and relative distance between samples, significantly different coupling patterns are identified, generating cluster label data, which is then mapped to geospatial data to construct a spatiotemporal pattern clustering layer. Next, statistical correlation analysis methods (such as Pearson correlation coefficient, canonical correlation coefficient, and mutual information) were used to calculate the synergistic relationship between vegetation change and noctilucentness, forming a coupling effect coefficient matrix. Based on this, a comprehensive assessment of the coupling effect was conducted by combining the spatiotemporal pattern clustering layer and the coupling effect coefficient matrix. Specifically, this included: extracting a subset of the corresponding coupling effect coefficient matrix for each cluster unit in the spatiotemporal pattern clustering layer to generate local coupling index data; based on these indices, a weighted fusion algorithm was used to fuse multiple coupling factors (such as the correlation between noctilucentness intensity stability and vegetation change rate) to obtain comprehensive coupling strength score data. To improve spatial continuity, spatial interpolation was performed on the comprehensive coupling strength score data to fill the uncovered areas of the spatiotemporal pattern clustering layer, generating a continuous coupling strength distribution map. Subsequently, threshold segmentation technology was used to identify significant regions of comprehensive coupling strength, extracting high-response regions and identifying them as high-response biological behavior hotspots. Furthermore, spatiotemporal distribution and evolution trend analysis was performed on these high-response biological behavior hotspots to identify their expansion, migration, or decline patterns, ultimately generating biological behavior response intensity distribution data.

[0126] Preferably, the analysis of low-frequency perturbation time-series changes in the explicit human activity area data in step S3 includes:

[0127] Long-term resampling and smoothing of data from areas of visible human activity are performed to extract stable trends and generate baseline data of low-frequency activity changes.

[0128] Remote sensing index sequences are extracted from multi-temporal remote sensing image data and decomposed by wavelet to extract low-frequency perturbation principal components and generate perturbation frequency band feature data.

[0129] By performing time window sliding analysis on the characteristic data of the disturbance frequency band using baseline data of low-frequency activity changes, we can identify low-frequency trend deviations and breakpoints, and generate low-frequency change anomaly detection data.

[0130] Spatial consistency constraints are applied to low-frequency anomaly detection data and explicit human activity area data to filter out short-period high-frequency false interference and generate a steady-state disturbance response layer.

[0131] The steady-state disturbance response layer is classified for disturbance mode and its intensity is calibrated to generate low-frequency disturbance identification data.

[0132] In this embodiment of the invention, multi-temporal remote sensing images and related activity index data of areas with visible human activity are collected to ensure a long time span (e.g., several years) and uniform or nearly uniform time intervals. The original time series data is resampled, and time intervals are standardized (e.g., monthly, quarterly, or annually) to fill in missing values. Interpolation methods (e.g., linear interpolation, spline interpolation) are used to ensure the continuity of the time series. Moving average, LOESS smoothing, or locally weighted regression methods are employed to remove high-frequency fluctuations and extract long-term stable trends, forming low-frequency activity change baseline data. Key remote sensing indicators from multi-temporal images, such as NDVI, NDBI (Building Index), and nighttime light intensity, are calculated to form multi-time-point indicator sequences. A suitable wavelet basis (e.g., Daubechies wavelet db4) is selected, and multi-scale wavelet decomposition is performed on each indicator sequence to separate low-frequency components (approximation coefficients) from high-frequency components (detail coefficients). Principal components reflecting low-frequency disturbances are extracted to generate disturbance frequency band feature data. An appropriate time window size (e.g., 12 months or 24 months) is set and slid across the disturbance frequency band feature data sequence. Calculate statistical indicators (mean, variance, rate of change of trend, etc.) within each window. Use breakpoint detection algorithms (such as CUSUM and Pelt) to identify trend deviations and breakpoints, generating low-frequency change anomaly detection data containing low-frequency trend deviations and breakpoints. Use the low-frequency activity change baseline data generated in step 1 as a reference to perform sliding window comparative analysis on the disturbance frequency band feature data. Confirm whether the detected trend deviations and breakpoints are significantly different from the baseline trend to avoid misjudgments. Combine the baseline and disturbance feature analysis results to generate more accurate low-frequency change anomaly detection data. Utilize the overlap between the spatial boundaries of explicit human activity areas and the spatial location of the anomaly detection data to remove inconsistent anomalies. Employ spatial clustering (such as DBSCAN) to identify the spatial coherence of anomalous regions, and combine this with time-series high-frequency signal analysis to remove noise interference with periods less than a set threshold, generating a steady-state disturbance response layer constrained by spatial consistency. Based on the spatial morphology and temporal characteristics of the disturbance response, classify and identify disturbance patterns (such as slowly increasing, abrupt, and periodic). Machine learning classification algorithms (such as random forests and support vector machines) combined with expert knowledge can be used to classify the intensity of disturbances (weak, medium, and strong) based on the amplitude, duration, and spatial range of the disturbances. This results in low-frequency disturbance identification data that reflects the characteristics of low-frequency disturbances in areas of overt human activity, which can be used for subsequent ecological impact analysis and management decisions.

[0133] Preferably, step S3, which involves predicting ecosystem health based on low-frequency disturbance identification data and edge effect identification data, includes:

[0134] Spatial semantic fusion is performed on low-frequency disturbance identification data and edge effect identification data to construct a comprehensive ecological disturbance index layer; ecological sensitivity weighted overlay analysis is performed on the comprehensive ecological disturbance index layer to generate multi-factor ecological vulnerability assessment data;

[0135] The multi-factor ecological vulnerability assessment data is divided into datasets to generate a model training set and a model test set;

[0136] The model training set is trained using a long short-term memory neural network algorithm to generate a pre-model for ecosystem health assessment; the pre-model is then optimized and iterated based on the model test set to generate the ecosystem health assessment model.

[0137] Multifactor ecological vulnerability assessment data are input into the ecosystem health assessment model for health assessment prediction, generating ecosystem health assessment prediction data for the target area.

[0138] In this embodiment of the invention, low-frequency disturbance identification data (reflecting overt human activity disturbance) and edge effect identification data (reflecting the sensitivity and diversity impact of ecosystem edges) are read. Spatial coordinate unification and grid alignment are performed on the two datasets to ensure direct comparison of data between corresponding spatial units. Spatial semantic fusion methods (such as weighted multi-index fusion or fuzzy logic-based fusion methods) are used to integrate the spatial information from the two data sources. A comprehensive index calculation model is designed, for example: Ieco = w1 × Dlowfreq + w2 × Eedge; where Ieco is the comprehensive ecological disturbance index, Dlowfreq is the low-frequency disturbance identification index, Eedge is the edge effect identification index, and w1 and w2 are weight coefficients (determined through expert scoring or data-driven approaches). An ecological disturbance comprehensive index layer covering the target area is generated. Relevant ecological sensitivity factors, such as soil type, terrain slope, precipitation, and vegetation cover, are collected. Weights are assigned to each sensitivity factor, and the comprehensive ecological disturbance index is combined for superposition calculation to form a multi-factor comprehensive evaluation. For example, a weighted linear superposition (WLA) model is used. Where V eco F is the ecological vulnerability assessment value. i Let w be the value of the i-th ecological sensitivity factor. i For the corresponding weights, w eco As the weight of the comprehensive ecological disturbance index, I ecoThis involves generating a comprehensive ecological disturbance index. Multi-factor ecological vulnerability assessment data reflecting the ecological vulnerability of the target area is then generated. The ecological vulnerability assessment data is compiled along with corresponding ecosystem health assessment reference data (such as historical monitoring indicators and biodiversity indicators). The dataset is divided into a training set (e.g., 70%-80%) and a test set (e.g., 20%-30%) using random sampling, stratified spatial sampling, or time-series partitioning methods. Data standardization and normalization are performed to ensure the numerical stability of the model training. An LSTM network structure is constructed, inputting the multi-factor ecological vulnerability assessment data and outputting ecosystem health assessment prediction results. Appropriate network layers, units, and activation functions are set to prevent overfitting. Iterative training is performed using the training set, and the prediction error is evaluated using a loss function (e.g., mean squared error, MSE). Optimization algorithms (e.g., Adam) are used to update the network weights. The model performance is evaluated using the test set, and model adjustments and hyperparameter optimization are performed using techniques such as early stopping and cross-validation. The model parameters are iteratively updated to obtain a stable and highly generalizable ecosystem health assessment model. The multi-factor ecological vulnerability assessment data is then input into the trained LSTM model. The model outputs predicted data on the ecosystem health assessment of the target area, quantifying the health status of the ecosystem. The prediction results may include health level classifications (such as healthy, sub-healthy, sub-poor, and deteriorating) or continuous numerical indicators.

[0139] As an example of the present invention, reference is made to... Figure 3 As shown, step S4 in this example includes:

[0140] Step S41: Perform factor stratification extraction on the ecosystem health assessment and prediction data to obtain data on potential latent disturbance factors within the region;

[0141] Step S42: Conduct a regional comprehensive analysis of the ecosystem of the target area based on the data of potential latent disturbance factors in the area to extract data of sensitive areas with latent threats. The regional comprehensive analysis includes land use, vegetation continuity and water connectivity analysis.

[0142] Step S43: Compare the fluctuations of sensitive area data over time to identify the trend of latent disturbances, generate a threat disturbance trend layer, and overlay the threat disturbance trend layer with the ecosystem health assessment and prediction data to divide the latent threat level ranges of different intensities.

[0143] Step S44: Geospatial mapping and distribution range definition of the hidden threat level range, generating a hidden threat level and distribution map of the target area.

[0144] In this embodiment of the invention, ecosystem health assessment and prediction data is input, including quantitative indicators of the current health status of the ecosystem within the region. Multi-factor hierarchical analysis is employed to decompose the comprehensive prediction indicators into different ecological factor levels, such as potential latent disturbance factors like soil quality, groundwater status, and ecological connectivity. Principal component analysis (PCA), factor analysis (FA), or hierarchical clustering methods are applied to identify hidden key influencing factors. A dataset of potential latent disturbance factors within the region is generated, with each factor represented as a spatial grid or vector unit, reflecting its spatial distribution characteristics. The data of potential latent disturbance factors and basic spatial data such as regional land use, vegetation cover, and water body distribution are input. Land use classification data is used to identify types such as cultivated land, construction land, and forest land, and the land use change rate and ecological disturbance potential are calculated. Remote sensing vegetation indices (such as NDVI) data, combined with spatial connectivity indicators (such as Moran's index and connectivity index), are used to assess the degree of vegetation contiguousness and the risk of fragmentation. Based on remote sensing images of water bodies and river system vector data, water body spatial connectivity indicators (such as water connectivity index) are calculated to assess the supporting function of water resources for the ecosystem. A spatial layer of sensitive areas for latent threats is generated through multi-factor spatial overlay and weighted scoring. Data on sensitive areas for latent threats within the target region are extracted to identify potential risk areas in the ecosystem. Multi-temporal time series fluctuation analysis is performed on the data of sensitive areas for latent threats, using statistical fluctuation indicators (such as standard deviation and coefficient of variation) and time series analysis methods (such as sliding window trend detection and change point detection). The changing trends and abnormal fluctuation points of latent disturbance factors over time are identified, revealing the dynamic evolution of potential threats. Based on the trend analysis results, a spatially distributed threat disturbance trend layer is generated, showing the strength and direction of latent disturbance trends. The threat disturbance trend layer is overlaid with ecosystem health assessment and prediction data, and combined with spatial weights and a multi-factor scoring system, latent threat level intervals (such as low, medium, and high levels) are divided. A spatial layer of latent threat level intervals is generated to provide a basis for subsequent spatial mapping. Using a GIS platform, vectorization processing is performed based on the latent threat level interval layer to draw threat level boundaries. Spatial analysis tools, such as buffer zone analysis and hotspot analysis, are applied to help clarify the distribution range of threat areas. Combining ecosystem boundaries and natural geographical conditions, the spatial expansion range of latent threats is reasonably defined. Generate map products that include the classification of latent threats and their spatial distribution, indicating the level and distribution range of latent threats within the target area, and providing spatial decision support for ecological management and risk prevention and control.

[0145] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0146] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. An application method for remote sensing identification of an ecological system survey, characterized in that, The method comprises the following steps: Step S1: acquiring multi-temporal remote sensing image data of a target region; extracting a ground cover type of the multi-temporal remote sensing image data, and identifying a region of explicit human activity of the multi-temporal remote sensing image data according to the ground cover type, to generate data of the region of explicit human activity; Step S2: acquiring night light data of the target region; spatiotemporally registering the night light data of the target region and the multi-temporal remote sensing image data to generate fused night light remote sensing data; extracting an edge region data of the multi-temporal remote sensing image data according to the ground cover type; and calculating a vegetation index and a night light index of the edge region data based on the fused night light remote sensing data; Step S3: performing biological behavior correlation analysis on the edge region data by using the vegetation index and the night light index to generate edge effect identification data; and analyzing a low-frequency disturbance time series change of the data of the region of explicit human activity to generate low-frequency disturbance identification data; performing ecosystem health evaluation prediction according to the low-frequency disturbance identification data and the edge effect identification data to generate ecosystem health evaluation prediction data of the target region; wherein the biological behavior correlation analysis on the edge region data by using the vegetation index and the night light index comprises: performing spatial grid division on the edge region data, combining a time series remote sensing image to extract a vegetation index change, and generating edge vegetation dynamic index data; superimposing a night light remote sensing image on the edge region data to extract a night light intensity and stability index, and generating edge night light index data; performing bivariate spatiotemporal collaborative analysis on the vegetation index and the night light index to generate vegetation-night light coupling feature data; performing spatiotemporal pattern clustering on the vegetation-night light coupling feature data to identify a coupling effect of human activity intervention and ecological disturbance, and generating biological behavior response layer data; performing edge sensitivity analysis and biological diversity influence determination on the biological behavior response layer data to generate the edge effect identification data; wherein the ecosystem health evaluation prediction according to the low-frequency disturbance identification data and the edge effect identification data comprises: performing spatial semantic fusion on the low-frequency disturbance identification data and the edge effect identification data to construct an ecological disturbance comprehensive index layer; and performing ecological sensitivity weighted superimposition analysis on the ecological disturbance comprehensive index layer to generate multi-factor ecological vulnerability evaluation data; dividing the multi-factor ecological vulnerability evaluation data into a model training set and a model test set; training a model by using a long short-term memory neural network algorithm on the model training set to generate an ecosystem health evaluation pre-model; and iteratively optimizing the ecosystem health evaluation pre-model according to the model test set to generate an ecosystem health evaluation model; inputting the multi-factor ecological vulnerability evaluation data into the ecosystem health evaluation model to perform health evaluation prediction, and generating the ecosystem health evaluation prediction data of the target region; Step S4: outputting a region implicit threat of an ecosystem of the target region based on the ecosystem health evaluation prediction data of the target region, to obtain an implicit threat level and a distribution map.

2. The method for remote sensing identification applied to ecological system investigation according to claim 1, characterized in that, Step S1 comprises the following steps: Step S11: controlling multi-period sampling of remote sensing images of the target region to acquire the multi-temporal remote sensing image data; Step S12: Perform radiation correction and geometric correction on the multi-temporal remote sensing image data to generate corrected multi-temporal remote sensing image data; Step S13: Extract the ground surface spectral features of the corrected multi-temporal remote sensing image data, and perform time series change analysis and classification processing on the ground surface spectral features to generate ground surface cover type data; Step S14: Perform human activity indication interpretation on the ground surface cover type data to generate preliminary human activity mask data; perform spatial consistency screening and boundary correction on the preliminary human activity mask data to generate dominant human activity region data.

3. The method for remote sensing identification applied to ecological system investigation according to claim 2, characterized in that, Step S13 includes the following steps: Step S131: Extract multi-band reflectance values of ground surface pixels at different time phases from the corrected multi-temporal remote sensing image data to obtain a sequence of ground surface spectral features with time labels; Step S132: Calculate the time series change trend of key band reflectance in the multi-temporal remote sensing image data based on the sequence of ground surface spectral features, and determine whether the change pattern meets any of the following conditions and mark accordingly to generate time series change judgment mark data: the NDVI index has seasonal periodic fluctuations and the amplitude within the year is greater than 0.3; the intersection point position of blue band and red edge band reflectance changes by more than ±15nm in three consecutive time phases; the number of spectral curve inflection points differs by more than 2 in different growth periods; Step S133: Use the time series change judgment mark data to perform cover type recognition on the ground surface spectral features, and if any of the following conditions is met, it can be determined as the specified cover type to generate ground surface cover type data: if the maximum NDVI value exceeds 0.6 and the NDVI annual fluctuation range is greater than 0.35, it is classified as "crop"; if the short-wave infrared reflectance changes by less than 5% throughout the year, and the NDVI is long-term stable and lower than 0.2, it is classified as "building surface"; if the normalized difference index of red light and near infrared presents double peaks within the year, and the peak values correspond to spring and autumn respectively, it is classified as "deciduous forest land".

4. The method for remote sensing identification applied to ecological system investigation according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Obtain the target area night light data; Step S22: Perform radiation normalization and noise suppression on the original radiation data of the night light to generate night light net radiation data; align the night light net radiation data and the multi-temporal remote sensing image correction data at the time and space reference points to generate a set of time and space registration parameters; Step S23: Perform grid-level registration and resampling fusion on the target area night light data and the multi-temporal remote sensing image data according to the set of time and space registration parameters to generate fused night light remote sensing data; perform edge detection and morphological region growing analysis on the remote sensing image boundary change area using the ground surface cover type data to generate an interface area layer; Step S24: Perform spatial extraction and segmentation on the interface area layer to generate edge region data; calculate the vegetation index of the edge region data based on the fused night light remote sensing data to generate edge region vegetation index data; Step S25: Extract the night light index of the edge region data based on the fused night light remote sensing data to generate edge region night light index data.

5. The method for remote sensing identification applied to ecological system investigation according to claim 4, characterized in that, The edge detection and morphological region growing analysis of the remote sensing image boundary change area by using the ground cover type data in step S23 includes: Edge extraction of the remote sensing image boundary change area by using the ground cover type data, setting a high threshold value of 100 pixel gray values and a low threshold value of 50 pixel gray values, and using a 3*3 Gaussian kernel for smoothing processing, so as to obtain a clear contour line data and convert it into a vector polygon format; Using the contour line data as a seed region, extending to a similar ground object region by using a region growing algorithm, setting a spectral similarity threshold value of ±10% reflectivity, an area threshold value of greater than or equal to 1000 m2, and a growth radius of within 90 meters, and re-labeling and rasterizing the growth result as a boundary area; Image superimposition of the boundary area and the remote sensing image boundary change area to generate a vector layer; assigning the following attribute fields to each boundary area unit: change type, change area, boundary length, and change time period, so as to output a boundary area layer. 6.The method for remote sensing identification applied to ecological system investigation according to claim 1, characterized in that, The low-frequency disturbance time sequence change analysis of the explicit human activity region data in step S3 includes: Long-time sequence resampling and smoothing processing of the explicit human activity region data to extract a stable change trend and generate low-frequency activity change baseline data; Extraction of a remote sensing index sequence of the multi-temporal remote sensing image data, wavelet decomposition, extraction of a low-frequency disturbance principal component, and generation of disturbance frequency band feature data; Time window sliding analysis of the disturbance frequency band feature data by using the low-frequency activity change baseline data to identify low-frequency trend deviation and breakpoints, and generation of low-frequency change anomaly detection data; Spatial consistency constraint of the low-frequency change anomaly detection data and the explicit human activity region data to filter short-period high-frequency false interference and generate a steady-state disturbance response layer; Disturbance mode classification and intensity calibration of the steady-state disturbance response layer to generate low-frequency disturbance identification data.

7. The method for remote sensing identification applied to ecological system investigation according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: Factor stratification extraction of the ecosystem health evaluation prediction data to obtain potential implicit interference factor data in the region; Step S42: Regional comprehensive analysis of the target region's ecosystem according to the potential implicit interference factor data in the region to extract sensitive area data of implicit threats, wherein the regional comprehensive analysis includes land use, vegetation continuity, and water connectivity analysis; Step S43: Fluctuation change comparison of the sensitive area data according to a time sequence to identify an implicit interference trend, generate a threat interference trend layer, and perform superimposition analysis of the threat interference trend layer and the ecosystem health evaluation prediction data to divide implicit threat grade intervals of different intensities; Step S44: Geographic space plotting and distribution range definition of the implicit threat grade intervals to generate an implicit threat grade and distribution map of the target region.

8. An application for an ecological system survey remote sensing identification system, characterized by, The application of the ecosystem investigation remote sensing identification method as claimed in claim 1, the ecosystem investigation remote sensing identification system includes: The multi-temporal remote sensing identification module is configured to acquire multi-temporal remote sensing image data of a target region; extract a ground cover type of the multi-temporal remote sensing image data, and perform explicit human activity region identification on the multi-temporal remote sensing image data according to the ground cover type, to generate explicit human activity region data; The noctilucent remote sensing identification module is configured to acquire target region night light data; perform space-time registration on the target region night light data and the multi-temporal remote sensing image data, to generate fused noctilucent remote sensing data; extract edge region data by performing boundary region extraction on the multi-temporal remote sensing image data according to the ground cover type; and calculate a vegetation index and a noctilucent index of the edge region data based on the fused noctilucent remote sensing data; The correlation analysis module is configured to perform biological behavior correlation analysis on the edge region data by using the vegetation index and the noctilucent index, to generate edge effect identification data; analyze low-frequency disturbance time series changes of the explicit human activity region data, to generate low-frequency disturbance identification data; and perform ecosystem health evaluation prediction according to the low-frequency disturbance identification data and the edge effect identification data, to generate ecosystem health evaluation prediction data of the target region; The ecosystem evaluation module is configured to output a region implicit threat of the ecosystem of the target region based on the ecosystem health evaluation prediction data of the target region, to obtain an implicit threat level and a distribution map.

Images