A mine subsidence area landscape ecological risk early warning method based on big data analysis

Abstract

The application discloses a kind of based on big data analysis's mine area subsidence area landscape ecological risk early warning method, comprising: by synthetic aperture radar interferometry obtains and processes subsidence time sequence, extracts cumulative amount and rate parameter and divides subsidence grade area;Based on the area, obtain landscape type and species distribution data, construct hierarchical response model to obtain fragile feature description;Accordingly, extract vegetation coverage and land use conversion law, assess habitat quality trend and generate ecological risk grade;When risk exceeds threshold, trigger early warning and dynamically adjust model parameters, through iterative optimization sensitive grading scheme;Finally, based on time series analysis, determine long-term degradation trend and risk path, form landscape ecological risk assessment file.The application realizes the whole process dynamic early warning from ground deformation to landscape ecological risk, improves the accuracy and timeliness of mine area landscape ecological risk identification, and provides technical basis for ecological protection and sustainable management.

Description

Technical Field

[0001] This invention belongs to the field of ecological environment monitoring technology in mining areas, and in particular relates to a method for early warning of landscape ecological risks in mining subsidence areas based on big data analysis. Background Technology

[0002] Surface subsidence caused by mining is a significant issue affecting ecological and environmental security. As mining activities continue, the landscape pattern of subsidence areas undergoes profound changes, leading to decreased ecosystem stability, damage to biological habitats, and even regional environmental disasters. Research in this area is directly related to the coordination of sustainable development and ecological protection in mining areas. Currently, many early warning methods rely primarily on ground measurements or single-period remote sensing data for assessment, making it difficult to capture the continuous dynamic changes in the subsidence process. Especially when processing large-scale, multi-temporal data, insufficient spatiotemporal resolution often fails to accurately reflect the gradual damage of subsidence to landscape elements.

[0003] The correlation between the accuracy of subsidence inversion and landscape ecological risk is a key factor limiting the effectiveness of early warning systems. Because surface deformation monitoring requires millimeter-level accuracy to effectively identify the early impacts of minor subsidence on vegetation cover and land use, and landscape fragmentation often intensifies gradually from patch splitting and corridor blockage, it is difficult to accurately establish the response relationship between subsidence spatial distribution data and ecological indicators. Furthermore, when subsidence accumulates to a certain extent, the differences in ecological sensitivity across different regions become significantly amplified. Mild subsidence may only cause localized soil loosening, while severe subsidence leads to widespread tree tilting and death or permanent damage to the topsoil of farmland. These differences in landscape vulnerability under different subsidence severity levels are difficult to determine using a uniform threshold.

[0004] For example, in a certain mining area, when the initial subsidence rate is slow, only minor cracks appear on the grassland surface, and changes in vegetation cover are not easily detected. However, as subsidence deepens, the cracks expand, accelerating soil erosion, and the grassland quickly degrades into bare land. The difficulty of ecological restoration increases exponentially, yet the optimal early warning opportunity is missed due to the lack of a precise mapping between subsidence rate and changes in landscape elements. Therefore, how to establish a dynamic mapping relationship between high-precision subsidence time-series information and landscape ecological risk, and how to achieve sensitive and differentiated early warning under subsidence degree classification, has become a key issue in improving the early warning capability for landscape ecological risks in mining areas. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a method for early warning of landscape ecological risks in mining subsidence areas based on big data analysis, thereby resolving the issues present in the existing technologies.

[0006] To achieve the above objectives, this invention provides a method for early warning of landscape ecological risks in mining subsidence areas based on big data analysis, comprising: The time-series variation of subsidence was obtained by synthetic aperture radar interferometry and stored in a big data platform after noise filtering. The cumulative subsidence amount and rate parameters were extracted from the subsidence time series after noise filtering, and a clustering algorithm was used to divide the subsidence areas of different severity levels. Landscape type data and species distribution data are obtained based on the subsidence level areas. A hierarchical response model is constructed based on the obtained data. Ecological vulnerability characteristics are described based on the hierarchical response model. Association rule mining is performed on the ecological vulnerability characteristics to extract vegetation cover and land use conversion patterns. Based on the extracted patterns, the habitat quality change trend is determined; and the ecological risk level is obtained based on the habitat quality change trend. If the ecological risk level exceeds a preset threshold, the early warning threshold system is triggered, the corresponding level of ecological early warning signal is activated, and the updated data of the dynamic assessment system is obtained; the updated data and the subsidence distribution map are integrated, and the parameters of the graded response model are adjusted. Based on the graded response model with adjusted parameters, an optimized sensitivity grading scheme is obtained from the updated data of the dynamic assessment system; based on the optimized sensitivity grading scheme, it is determined whether iterative processing of the analysis of cumulative amount and rate parameters is required to maintain the continuity of landscape sensitivity monitoring. Time series analysis is used to obtain long-term ecological degradation trends from iterative processing results. Based on these trends, the dynamic change path of risk levels is determined and stored in a big data platform to form a complete landscape ecological risk assessment archive, thereby enabling early warning of landscape ecological risks in mining subsidence areas.

[0007] Optionally, the process of obtaining the subsidence time-series variation sequence through synthetic aperture radar interferometry and then filtering it for noise includes: Raw surface deformation data of the mining area was acquired from multi-temporal radar images using synthetic aperture radar interferometry (SAR). The raw surface deformation data was inverted using a phase difference calculation method to obtain preliminary surface deformation results. Noise filtering was applied to the preliminary surface deformation results using a big data platform, and an adaptive filtering algorithm was used to remove interference signals, resulting in a clear deformation data sequence. Subsidence temporal variation characteristics were extracted from the clear deformation data sequence, and time series analysis methods were applied to segment the variation characteristics to determine the subsidence trend of each time period. If the subsidence trend of a certain time period exceeded a preset threshold, the deformation data of that time period was weighted to obtain a precise subsidence temporal variation sequence, which was then stored in the big data platform. Optionally, cumulative subsidence amount and rate parameters are extracted based on the subsidence time-series change sequence after noise filtering, and a clustering algorithm is used to divide subsidence areas of different severity levels, including: Based on the subsidence time-series change sequence after noise filtering, the cumulative subsidence amount and subsidence rate parameters are calculated; the K-means clustering method is used to group the cumulative subsidence amount and subsidence rate parameters to obtain the distribution range of the parameters; if the distribution range of the parameters exceeds the preset threshold range, the corresponding abnormal data is labeled; based on the categories obtained from the grouping process and the abnormal data labeling results, combined with the cumulative subsidence amount parameter, subsidence levels from mild to severe are divided; the subsidence levels are mapped to specific geographical area data to obtain the correspondence between subsidence levels and regional locations, and the subsidence level regional division results are determined; based on the regional division results, a classification label for the subsidence level is generated to determine the severity of subsidence in each region.

[0008] Optionally, the process of obtaining a description of ecological vulnerability characteristics includes: The spatial overlay matching of the subsidence level areas with the landscape type layers obtained from remote sensing interpretation determines the spatial distribution pattern of the landscape types. Based on the spatial distribution pattern of the landscape types, species distribution data is extracted, and cluster analysis is used to group the species distribution data to determine the aggregation characteristics of the species distribution. A hierarchical response model is constructed based on the aggregation characteristics, and a pre-established response rule base is used to hierarchically map different aggregation characteristics to obtain hierarchical response results. Sensitivity index data is extracted from the hierarchical response results, and the sensitivity index data is standardized to obtain quantitative values ​​of the sensitivity indicators. Vulnerability characteristics are identified based on the quantitative values ​​of the sensitivity indicators. If a certain indicator value exceeds a preset threshold, the corresponding area is marked as a high-vulnerability area, determining the distribution range of the vulnerability characteristics. For the distribution range of the vulnerability characteristics, a text description containing the distribution range and response relationship is generated to obtain a description of the ecological vulnerability characteristics.

[0009] Optionally, the process of obtaining an ecological risk level includes: Association rule mining is performed on the ecological vulnerability characteristics to extract association rules between vegetation cover and land use conversion. For land use conversion patterns, frequent itemset identification is applied to determine high-frequency conversion patterns. Vegetation cover correspondences are extracted based on these high-frequency conversion patterns to identify habitat quality decline paths. If vegetation cover continuously decreases along a habitat quality decline path, the habitat quality change trend is marked as degraded. The degraded change trend is matched with a pre-established risk mapping relationship to determine the ecological risk level. Regional data is aggregated based on the ecological risk level to obtain the ecological risk level assessment result.

[0010] Optionally, the parameters of the graded response model are adjusted by integrating updated data and subsidence distribution maps, including: if the ecological risk level assessment result exceeds a preset threshold, triggering an early warning threshold system to generate an ecological early warning signal of the corresponding level; activating a dynamic assessment system through the ecological early warning signal to obtain real-time monitoring and update data; overlaying the real-time monitoring and update data with the subsidence distribution map to extract the current subsidence area range; using a spatial overlay analysis method to perform spatial intersection calculations between the current subsidence area range and the ecological unit layer to determine the ecological units affected by subsidence; and adjusting the parameter weights of the graded response model based on the ecological units affected by subsidence to obtain updated parameter values.

[0011] Optionally, based on the graded response model with adjusted parameters, an optimized sensitivity grading scheme is obtained from the updated data of the dynamic evaluation system; based on the optimized sensitivity grading scheme, it is determined whether iterative processing of the analysis of the cumulative amount and rate parameters is required, including: Based on the graded response model with adjusted parameters, basic information on sensitivity grading is extracted from the updated data of the dynamic evaluation system to determine a preliminary sensitivity grading optimization scheme. According to the preliminary sensitivity grading optimization scheme, the corresponding landscape monitoring data is obtained to determine the dynamic trend of landscape sensitivity. If the dynamic trend of landscape sensitivity exceeds a preset threshold, the iterative processing flow is triggered based on the correlation between the changes in the cumulative subsidence amount parameter and the subsidence rate parameter.

[0012] Optionally, time series analysis is used to obtain long-term ecological degradation trends from the iterative processing results, and the dynamic change path of risk levels is determined based on these long-term ecological degradation trends, including: Time series analysis is performed on the iterative processing results. Stationary time series data are obtained through stationarity testing and differencing. An autoregressive integral moving average model is used to fit the stationary time series data to obtain a long-term ecological degradation trend curve. Slope and fluctuation amplitude parameters are extracted from the long-term ecological degradation trend curve to determine the direction and intensity of the ecological degradation trend and the persistence level of ecological degradation. The persistence level of ecological degradation is mapped to a preset risk level threshold range. If the slope is greater than zero and the fluctuation amplitude exceeds the threshold, the risk level is determined to be in an upward trend. A dynamic change path is constructed based on the upward trend of the risk level, resulting in a trajectory point set of risk level evolution over time. This trajectory point set is overlaid and correlated with landscape ecological spatial data to obtain the dynamic change path of the risk level for each landscape unit. The dynamic change path of the risk level for each landscape unit is input into a big data platform for distributed storage, forming a complete landscape ecological risk assessment archive.

[0013] Compared with the prior art, the present invention has the following advantages and technical effects: This invention addresses the dynamic changes in ecological vulnerability and risk caused by surface subsidence in mining areas by innovatively integrating a complete logical chain of surface deformation monitoring, ecological response modeling, and risk early warning. It acquires subsidence time-series data through multi-temporal radar imagery, extracts cumulative amount and rate parameters after noise filtering, and uses clustering algorithms to classify subsidence level regions. A graded response model is constructed using landscape type and species distribution data to analyze vulnerability characteristics, uncover vegetation cover and land use conversion patterns, and generate ecological risk level assessment results. When the risk exceeds a threshold, an early warning signal is triggered, data is dynamically updated to adjust model parameters, optimize the sensitivity classification scheme, and predict long-term ecological degradation trends through time series analysis, ultimately forming a complete risk assessment archive. This invention achieves full-process monitoring and early warning from surface deformation to ecological risk, significantly improving the accuracy and timeliness of ecological protection in mining areas and providing a scientific basis for sustainable management. Attached Figure Description

[0014] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the landscape vulnerability feature analysis of an embodiment of the present invention. Figure 3 This is a flowchart of the long-term ecological risk trend analysis based on time series according to an embodiment of the present invention. Detailed Implementation

[0015] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0016] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0017] Example 1 like Figure 1 As shown, this embodiment provides a method for early warning of landscape ecological risks in mining subsidence areas based on big data analysis, including: S101. Obtain surface deformation data of the mining area through synthetic aperture radar interferometry, and obtain the subsidence time series change sequence based on the surface deformation data of the mining area; the subsidence time series change sequence is stored in the big data platform after noise filtering.

[0018] Raw surface deformation data of the mining area was acquired from multi-temporal radar images using synthetic aperture radar interferometry. A pre-defined phase difference calculation method was used to process the image data, yielding preliminary results of surface deformation. For these preliminary results, a big data platform was used to filter noise, employing a pre-defined filtering algorithm to remove interference signals from the images and determine the clear sequence of deformation data. Subsidence temporal variation characteristics were extracted from the clear sequence of deformation data, and time series analysis methods were applied to segment the variation sequence and determine the subsidence trend within each time period. If the subsidence trend exceeded a pre-defined threshold within a certain time period, the deformation data for that time period was weighted to obtain a more accurate subsidence variation sequence. This data was then stored in the big data platform for later retrieval.

[0019] Synthetic Aperture Radar Interferometry (SAR) is a remote sensing method that uses the phase difference of multi-temporal SAR images to extract surface deformation information. It is widely used in mining area monitoring and can acquire millimeter-level precision data in all weather conditions. For example...

[0020] In one embodiment, Sentinel-1 satellite imagery is used to process the coal mining area. First, multiple images are registered, and image pairs with short spatiotemporal baselines are selected to generate interferograms. Phase differences are then calculated to obtain preliminary surface deformation results, such as a maximum subsidence of 11 cm in a mining area over 30 days. These preliminary results often include topographic residuals and atmospheric interference, requiring further processing to improve reliability.

[0021] In one possible implementation, a big data platform is used to filter noise from the preliminary results, and adaptive filtering algorithms such as Goldstein filtering are used to remove atmospheric phase and thermal noise interference, while retaining the highly coherent point sequence.

[0022] For example, atmospheric delay in mining area imagery can cause phase deviations of several centimeters. By using multi-temporal averaging or model correction, a clear deformation sequence can be obtained, improving the accuracy of the subsidence rate to 5-15 mm / year. This filtering helps suppress spatiotemporal incoherence, especially at the edges of vegetated mining areas, significantly improving data usability.

[0023] Specifically, the subsidence temporal change characteristics are extracted from the clear sequence, and time series analysis methods such as SBAS are applied to segment the sequence to determine the subsidence trend of each segment.

[0024] For example, monitoring data from 2003 to 2010 in the Changzhi mining area showed that the subsidence rate in residential areas was 5-15 mm / year, with a cumulative subsidence of 90 mm. Subsidence in older mining areas was slow, while in newer mining areas it was rapid and non-linear. This segmented approach can reveal differences in mining stages and identify potential collapse risks at an early stage.

[0025] It should be noted that if the subsidence trend exceeds a preset threshold in a certain period of time, such as a rate greater than 50 mm / month, the data in that segment will be weighted, giving higher coherence points greater weight to obtain a more accurate sequence.

[0026] For example, after weighting in the rapid settlement zone, the center settlement value is corrected from an underestimated 110 mm to a value closer to the actual 2415 mm. This weighting helps overcome large gradient incoherence, improves the accuracy of the mining area center, and supports disaster early warning.

[0027] For example, in a mining area in Shanxi Province, the entire process, from processing raw images to storing clear sequences on a big data platform, enables time-series monitoring, clearly displaying the cumulative subsidence basin at a maximum rate of 150 mm / year. This method has significant benefits, enabling timely detection of subsidence evolution, providing a basis for mine safety assessments, and reducing losses from geological disasters.

[0028] Preferably, PS and SBAS technologies are combined to obtain linear deformation on permanent scatterers such as buildings and capture nonlinear changes in distributed areas, which support each other to form a complete mining area subsidence map.

[0029] For example, in a mining area, the eastern part subsides by 170 cm and the western part by 70 cm. Weighted sequence storage facilitates subsequent multi-period comparative analysis. This comprehensive processing enhances monitoring robustness and is suitable for complex mining environments.

[0030] S102. Based on the subsidence time-series change sequence, extract the cumulative amount and rate parameters, and use a clustering algorithm to divide the subsidence level regions.

[0031] Based on the filtered time-series data, cumulative subsidence and subsidence rate parameters are calculated to obtain quantitative results for both parameters. K-means clustering is then used to group the cumulative subsidence and subsidence rate parameters, determining their distribution ranges and obtaining preliminary classifications. If the distribution ranges of the cumulative subsidence and subsidence rate parameters exceed preset thresholds, abnormal data are labeled to identify areas requiring further analysis. Based on the classifications and labeled abnormal areas, combined with the cumulative subsidence parameters, subsidence levels are categorized from mild to severe, yielding a level distribution result. This level distribution result is mapped to specific geographic area data to obtain the correspondence between subsidence levels and regional locations, determining the final regional division result. After obtaining the regional division result, classification labels for subsidence levels are generated to determine the severity of subsidence in each region, resulting in complete subsidence distribution information.

[0032] For example, in the field of surface subsidence monitoring in mining areas, the cumulative subsidence amount and subsidence rate parameters are first calculated based on the filtered time series data.

[0033] Specifically, cumulative subsidence refers to the total vertical displacement of the land surface from the start of monitoring to the present moment, while the subsidence rate reflects how fast the deformation changes per unit time.

[0034] In one embodiment, assuming that a mining area location shows a cumulative subsidence of 120 mm in a time series from January 2024 to June 2025, this value is directly taken as the cumulative subsidence amount; the average monthly subsidence during the same period is 8 mm, so the quantified subsidence rate is 8 mm / month. This quantification helps to intuitively grasp the subsidence evolution process and provides a reliable basis for subsequent analysis. For these parameters, K-means clustering is used for grouping. K-means clustering iteratively assigns data points to the nearest cluster center, achieving automatic identification of parameter distribution intervals.

[0035] For example, by using the cumulative subsidence amount and subsidence rate of thousands of monitoring points across the entire mining area as two-dimensional feature inputs and setting an initial K value of 4, four distribution intervals are obtained after multiple iterations, such as 0-50mm, 50-150mm, 150-300mm, and a category exceeding 300mm. This grouping method can effectively reveal the spatial heterogeneity of subsidence parameters and avoid the bias of manually setting thresholds.

[0036] It should be noted that if the parameter distribution range of certain points exceeds the preset threshold range, such as the cumulative subsidence exceeding 300mm or the subsidence rate exceeding 15mm / month, these abnormal data will be marked.

[0037] In one possible implementation, the system automatically marks these locations as high-risk areas and records their coordinate range. This marking helps to prioritize the identification of potential disaster sources, reduce the workload of comprehensive investigation, and improve monitoring efficiency. Based on the classification and the marked abnormal areas, combined with the cumulative subsidence parameter, the subsidence levels are further divided into light to extremely severe levels.

[0038] For example, mild subsidence corresponds to a cumulative subsidence of less than 50 mm and a rate of less than 5 mm / month; moderate subsidence is 50-150 mm; severe subsidence is 150-300 mm; and extremely severe subsidence is defined as areas exceeding 300 mm or accompanied by anomaly markings. This multi-factor comprehensive classification can more accurately reflect the degree of subsidence hazard and avoid misjudgments caused by a single parameter. By mapping the grade distribution results to specific geographical area data, the correspondence between subsidence grade and location can be obtained.

[0039] For example, by overlaying extremely severe subsidence level points onto the boundary map of the mining area's goaf, it was found that these areas are mainly concentrated above the old mining areas, thus determining the final regional division results, such as dividing them into high-risk, medium-risk, and low-risk zones. This spatial correspondence provides a precise basis for mine production scheduling and disaster prevention. After obtaining the regional division results, classification labels for subsidence levels are generated, and the severity of subsidence in each area is determined, ultimately forming complete subsidence distribution information.

[0040] For example, areas with extremely severe subsidence are marked in red, severe in orange, and mild in yellow. This visual identification helps managers quickly identify key areas for prevention. The beneficial effect is that it supports dynamic adjustments to mining plans and timely implementation of surface reinforcement or relocation measures, thereby significantly reducing the safety risks and economic losses caused by ground subsidence.

[0041] S103. Obtain landscape type and species distribution data from the subsidence-level areas, construct a graded response model, and obtain a vulnerability characteristic description through sensitivity index analysis. The specific process of this step is as follows: Figure 2 As shown.

[0042] Based on the preliminary regional classification results, landscape type data is acquired. Using data matching methods, landscape types are correlated with each regional classification to determine their spatial distribution patterns. Species distribution data is extracted based on these patterns. If species distribution data shows a concentrated trend within a certain region, cluster analysis is used to group them and determine their aggregation characteristics. A hierarchical response model is constructed based on these aggregation characteristics. A pre-established response rule base is used to map different aggregation characteristics hierarchically, yielding hierarchical response results. Sensitivity index data is extracted from the hierarchical response results. Sensitivity indices are standardized using an index quantification tool to obtain their quantified values. Based on these quantified values, vulnerability characteristics are identified using analytical methods. If an index value exceeds a preset threshold, it is marked as a high-vulnerability area, determining the distribution range of the vulnerability characteristics. For each vulnerability characteristic's distribution range, a feature description is generated. A text generation tool is used to correlate the distribution range with the response relationship, obtaining the final vulnerability characteristic description results.

[0043] Based on the preliminary regional classification results, landscape type data is obtained. Using data matching methods, landscape types are correlated with each regional classification to determine the spatial distribution pattern of landscape types. For example...

[0044] In one possible implementation, the subsidence level area is spatially overlaid and matched with the landscape type layer obtained by remote sensing interpretation to obtain the proportion distribution of landscapes such as forests, grasslands, farmland and construction land in different subsidence levels, thereby clearly revealing the spatial distribution pattern of landscape types.

[0045] Specifically, based on the spatial distribution pattern of landscape types, species distribution data are extracted. If the species distribution data shows a concentrated trend in a certain area, cluster analysis is used to group them and determine the aggregation characteristics of species distribution.

[0046] For example, in areas with mild subsidence, the habitat data of a certain key protected bird species show obvious concentration. Multiple clustering core areas can be identified through density clustering algorithms, and it can be determined that these clustering characteristics are mainly dominated by the forest landscape, thus providing a basis for subsequent response analysis.

[0047] In one possible implementation, a hierarchical response model is constructed based on the clustering characteristics of species distribution. A pre-established response rule base is used to hierarchically map different clustering characteristics to obtain hierarchical response results.

[0048] Specifically, the response rule base can include the correspondence between aggregation density and subsidence impact. For example, when the aggregation density is higher than 15 habitat points per square kilometer, it is mapped to a high response level. Finally, the classification results such as mild response, moderate response and high response are obtained, which effectively quantifies the sensitivity of the ecosystem to subsidence.

[0049] For example, sensitivity index data can be extracted from the graded response results, and the sensitivity indexes can be standardized using index quantification tools to obtain quantified values ​​of the sensitivity indexes.

[0050] In one possible implementation, species richness, aggregation intensity, and landscape connectivity are selected as sensitivity indicators, and their values ​​are converted to the range of 0 to 1 using the min-max normalization method. For example, the species richness of a certain area is 0.85 after normalization, indicating high sensitivity.

[0051] Specifically, based on the quantitative values ​​of sensitivity indicators, analytical methods are applied to identify vulnerability characteristics. If a certain indicator value exceeds a preset threshold, it is marked as a high-vulnerability area, thus determining the distribution range of vulnerability characteristics.

[0052] For example, a threshold of 0.7 is set. When the combined value of multiple indicators exceeds this threshold, it is marked as highly vulnerable. The distribution range is mainly concentrated in areas with moderate to severe subsidence and a high proportion of forest landscape, thereby accurately identifying ecologically vulnerable hotspots.

[0053] In one possible implementation, feature descriptions are generated based on the distribution range of vulnerability features. A text generation tool is then used to associate the distribution range with the response relationship to obtain the final vulnerability feature description result.

[0054] For example, the generated description is "This highly vulnerable area is located in the northern part of the severely subsided area, with forest landscape accounting for 65%, species aggregation showing a high responsiveness, and is susceptible to further subsidence leading to habitat fragmentation." This kind of related description helps to intuitively convey the causes and consequences of vulnerability, supports subsequent ecological protection decisions, and significantly improves the pertinence and practicality of ecological risk assessment in subsidence areas.

[0055] S104. Using association rule mining, vegetation cover and land use conversion patterns are extracted from the vulnerability feature descriptions to determine the habitat quality change trend. Input risk mapping relationships to generate ecological risk level assessment results.

[0056] Association rule mining was employed to process vulnerable feature descriptions and obtain vegetation cover and land use conversion patterns. Frequent itemset identification was used to determine high-frequency conversion patterns for land use conversion. Vegetation cover correspondences were extracted from these high-frequency conversion patterns to identify habitat quality decline paths. If vegetation cover continuously decreased along a habitat quality decline path, the habitat quality change trend was marked as degraded. Ecological risk levels were determined by matching degraded change trends with pre-established risk mapping relationships. Ecological risk level assessment results were obtained by aggregating regional data based on ecological risk levels.

[0057] For example, when studying the vulnerability characteristics of regional ecosystems, association rule mining can be used to analyze the relationship between vegetation cover and land use conversion. Suppose that in a specific wetland protected area, historical data shows that over the past 10 years, the conversion rate of farmland to construction land has reached as high as 40%, while vegetation cover has decreased from 75% to 50%. Association rule mining reveals a strong association between farmland conversion to construction land and the decline in vegetation cover. The support of this association rule reaches 0.8, and the confidence level is 0.9, indicating that this conversion pattern is one of the main factors leading to vegetation reduction.

[0058] For example, when applying frequent itemset identification methods to analyze land use transformation patterns, the focus can be on high-frequency transformation patterns. In the aforementioned wetland reserve, data analysis shows that the conversion from farmland to construction land accounts for the highest proportion of all transformation types, reaching an annual conversion rate of approximately 15%, and this pattern has persisted for the past five years. Through frequent itemset identification, this transformation pattern can be marked as a high-frequency pattern, thus providing data support for subsequent analysis. This method helps to quickly identify the land use change types with the greatest ecological impact.

[0059] For example, when extracting vegetation cover correlations to determine habitat quality decline pathways, spatial distribution data can be used for analysis. Suppose that in the core area of ​​a wetland reserve, vegetation cover gradually decreases from 60% to 30%, while bird habitat observations show a reduction in the number of key species within those habitats by about one-third. By comparing the trend of declining vegetation cover with changes in species numbers, it can be inferred that the habitat quality decline pathway is primarily influenced by vegetation reduction. This analytical approach can clearly reveal the key driving factors of ecological degradation.

[0060] For example, if vegetation cover continues to decrease along a habitat quality decline path, the trend is marked as degraded. In the case above, vegetation cover declined by more than 10% for three consecutive years without any signs of recovery; therefore, it was marked as a degraded trend. This marking method helps to quickly identify the direction of deterioration in ecosystem health and provides a basis for subsequent risk assessment.

[0061] For example, ecological risk levels can be determined by matching degradation trends with pre-established risk mapping relationships. For instance, the risk mapping relationship might define a high-risk level as a vegetation cover decline exceeding 20% ​​over a period of more than two years. In the case of the wetland reserve, the core area met this condition and was therefore assessed as high-risk. This method simplifies complex ecological changes into actionable risk levels, facilitating management decisions.

[0062] For example, when aggregating regional data based on ecological risk levels, the risk levels of different sub-regions within a wetland reserve can be summarized to obtain an overall assessment result. Assuming the core area is high-risk and the surrounding buffer zone is medium-risk, the overall ecological risk level of the region might be classified as medium-to-high risk. This aggregation method comprehensively reflects the regional ecological condition, providing a comprehensive reference for formulating protection measures. The advantage of this analytical method lies in its ability to combine local issues with overall risk, enhancing the scientific rigor and practicality of the assessment.

[0063] S105. If the ecological risk level assessment result exceeds the preset threshold, the early warning threshold system will be triggered to activate the corresponding level of ecological early warning signal, obtain the dynamic assessment system update data fusion subsidence distribution map and adjust the response model parameters.

[0064] If the ecological risk level exceeds a preset threshold, an early warning threshold system is triggered to generate an ecological early warning signal. The ecological early warning signal activates a dynamic assessment system to obtain real-time monitoring and update data. The subsidence area is extracted by overlaying a subsidence distribution map onto the real-time monitoring and update data. Spatial overlay analysis is used to determine the ecological units affected by the subsidence within the subsidence area. Response model parameters are adjusted based on the affected ecological units to obtain updated parameter values. The ecological risk level is recalculated based on the updated parameter values ​​in the response model to obtain the adjusted ecological risk level. The adjusted ecological risk level is compared with the preset threshold to determine whether a high-level ecological early warning signal should continue to be triggered.

[0065] For example, in the field of ecological risk assessment, when the adjusted ecological risk level exceeds a preset threshold, a high-level ecological early warning signal will be continuously triggered, thereby initiating a more refined dynamic management process.

[0066] Specifically, if the ecological risk level exceeds the preset threshold, the early warning threshold system will be immediately triggered to generate an ecological early warning signal. This mechanism can promptly capture the trend of risk deterioration and prevent the potential ecological damage from expanding.

[0067] In one embodiment, the warning threshold can be set to 0.7. When the risk level reaches 0.8, the system automatically generates a red warning signal to notify relevant departments to intervene quickly.

[0068] Understandably, activating a dynamic assessment system through ecological early warning signals allows for the acquisition of real-time monitoring and updates, such as daily updated NDVI vegetation index data from remote sensing satellites and soil moisture information collected by ground sensors. Comparing this real-time data with historical data can accurately reflect the latest state of the ecosystem.

[0069] For example, in the ecological monitoring of coal mining areas, if an early warning signal is triggered, the dynamic assessment system will prioritize pulling the land subsidence monitoring data and vegetation cover change data of the most recent week to ensure that subsequent analysis is based on the latest situation.

[0070] Specifically, when extracting the subsidence area range by overlaying the subsidence distribution map with real-time monitoring and update data, a grid overlay method can be used to designate areas with a subsidence depth greater than 5 cm as high-impact areas.

[0071] For example, in a certain mining area, the subsidence distribution map showed that the central area subsided by 15 centimeters. By overlaying it with real-time vegetation data, the core subsidence area of ​​about 120 hectares was extracted. This area was defined as the core object of subsequent analysis, effectively focusing on high-risk areas.

[0072] In one possible implementation, spatial overlay analysis is used to determine the ecological units affected by the subsidence area. Spatial intersection calculations can be performed between the subsidence layer and ecological unit layers such as forests, grasslands, and wetlands.

[0073] For example, if the subsidence area overlaps with the original forest area by 80 hectares, the forest ecological unit is marked as a highly affected unit. This process helps to accurately identify the most vulnerable habitat segments and support targeted conservation measures.

[0074] For example, when adjusting the parameters of the response model to obtain updated values ​​by adjusting the parameters of the ecological units affected by subsidence, the model weights can be dynamically adjusted according to the degree of impact.

[0075] In one embodiment, the vegetation cover weight in the original model is 0.4. If the vegetation cover of the affected unit decreases by 20%, the weight is increased to 0.55 to reflect the amplifying effect of subsidence on ecological sensitivity. This parameter update makes the model more closely reflect actual risk scenarios.

[0076] Understandably, by recalculating the ecological risk level based on the updated parameter values ​​input into the response model, the adjusted ecological risk level can be obtained. This iterative calculation can significantly improve the accuracy of the assessment.

[0077] For example, the original risk level was 0.65, but after parameter adjustment, it was recalculated to 0.82, reflecting the real threat after the subsidence intensified.

[0078] Specifically, the system compares the adjusted ecological risk level with a preset threshold to determine whether a high-level ecological early warning signal continues to be triggered. If the level remains above 0.7, the high-level warning status is maintained until the risk decreases. This closed-loop judgment mechanism ensures the continuity and reliability of the early warning signal, effectively supports the dynamic tracking and timely intervention of ecological risks, and ultimately reduces the rate of habitat degradation and maintains regional ecological stability.

[0079] S106. After adjusting the response model parameters, obtain the optimized sensitivity classification scheme from the updated data, and determine whether the cumulative parameters and rate analysis need to be iteratively processed to maintain the continuity of landscape sensitivity monitoring.

[0080] By correlating the response model with parameter adjustments, basic information on sensitivity classification is obtained from updated data, and a preliminary optimization scheme is determined using data extraction. Based on the preliminary optimization scheme, and considering the relationship between sensitivity classification and landscape sensitivity, corresponding monitoring data is obtained to determine the dynamic trend of landscape sensitivity. If the dynamic trend of landscape sensitivity exceeds a preset threshold, the correlation between accumulated parameters and rate changes is used to determine whether to trigger an iterative processing flow. For the results of the iterative processing flow, the update frequency of continuous monitoring is obtained by combining the relationship between rate changes and monitoring continuity, and the priority of monitoring adjustments is determined. Based on the priority of monitoring adjustments, and considering the relationship between the optimization scheme and accumulated parameters, a hierarchical analysis tool is used to obtain the specific direction of parameter optimization. Based on the specific direction of parameter optimization, and considering the correlation between data extraction and scheme analysis, the final sensitivity classification adjustment strategy is obtained, and the execution plan for monitoring continuity is determined. Using the final sensitivity classification adjustment strategy, and considering the relationship between landscape sensitivity and monitoring continuity, the response model is used to update monitoring data, obtaining the optimization results for continuous monitoring.

[0081] For example, in the field of ecological risk monitoring of land subsidence caused by coal mining, basic information on ecological sensitivity classification can be extracted from real-time updated data by correlating response models with parameter adjustments. The core of this correlation lies in the fact that the response model can dynamically reflect parameter changes based on historical subsidence data and ecological indicators, thereby guiding the data extraction process.

[0082] In one embodiment, sensitive classification information such as vegetation coverage, soil moisture, and biodiversity index are first extracted from the monitoring update data obtained from the dynamic assessment system. This information comes directly from the real-time dataset after the subsidence distribution map is overlaid.

[0083] Understandably, when using data extraction to determine a preliminary optimization plan, it is necessary to prioritize the selection of ecological units with higher sensitivity.

[0084] For example, when the vegetation coverage in a certain area drops below 60% and the soil moisture decreases by 15%, it can be preliminarily determined to be at a high sensitivity level, thus forming the prototype of an optimization plan. This plan emphasizes targeted adjustments to the subsidence-affected area to avoid resource waste caused by changes in global parameters.

[0085] In one possible implementation, the relationship between sensitivity classification and landscape sensitivity is combined to obtain corresponding monitoring data in order to determine the dynamic trend of landscape sensitivity.

[0086] Specifically, by overlaying data on the extent of subsidence areas with landscape patterns, changes in hilly areas or water buffer zones with slopes greater than 15 degrees are monitored. If landscape-sensitive trends show a 20% increase in the vegetation fragmentation index within a short period, it indicates that the dynamic changes have exceeded a preset threshold of 0.15, requiring timely intervention. This assessment helps to detect ecological degradation signals early and maintain stable landscape connectivity.

[0087] For example, if the dynamic change trend exceeds a threshold, the correlation between accumulated parameters and rate changes determines whether to trigger an iterative processing flow. Accumulated parameters may include the cumulative value of subsidence depth and the rate of ecological loss.

[0088] In one embodiment, an iterative process is triggered when the cumulative subsidence depth exceeds 2 meters and the loss rate reaches 5% per month. This triggering mechanism ensures that the system proactively optimizes before the risk accumulates to a critical point, avoiding irreversible ecological damage caused by passive responses. Based on the results of the iterative process, the update frequency of continuous monitoring is obtained by combining the relationship between rate changes and monitoring continuity.

[0089] For example, when the rate of change is high, the update frequency can be adjusted from once a day to once every 6 hours, thereby determining the priority of monitoring adjustments. Higher priority areas will be allocated more sensor resources to improve data timeliness.

[0090] Understandably, by monitoring the priority of adjustments and combining the relationship between optimization schemes and cumulative parameters, hierarchical analysis tools can be used to obtain specific directions for parameter optimization.

[0091] In one embodiment, the stratified analysis first processes the high-sensitivity layer, then extends to the medium-sensitivity layer, ultimately determining specific directions such as increasing the weight of vegetation restoration parameters by 1.5 times. This stratified approach improves the accuracy of optimization and significantly reduces overall ecological risk.

[0092] For example, based on the specific direction of parameter optimization and the correlation between data extraction and scheme analysis, the final sensitivity grading adjustment strategy can be obtained.

[0093] For example, areas that were originally considered moderately sensitive can be reclassified as highly sensitive, and correspondingly, more stringent monitoring protocols can be implemented. This strategy directly improves the accuracy of early warning systems.

[0094] In one possible implementation, by adjusting the final sensitivity classification strategy, combining the relationship between landscape sensitivity and monitoring continuity, and using a response model to update monitoring data, the optimized results of continuous monitoring are obtained.

[0095] For example, the updated model can reduce risk prediction errors by 10%, achieving more stable ecological protection results. This optimization helps maintain the ecological balance of the mining area in the long term and promotes sustainable development.

[0096] S107. Use time series analysis to obtain long-term ecological degradation trends from iterative processing results, determine the dynamic change path of risk level, input the data into a big data platform for storage, and form a complete landscape ecological risk assessment archive.

[0097] The stationarity of the iterative processing results was tested and differencing was performed using time series analysis to obtain stationary time series data. An autoregressive integral moving average model was used to fit the sequence based on the stationary time series data, resulting in a long-term degradation trend curve. Slope and fluctuation amplitude parameters were extracted from the long-term degradation trend curve to determine the trend direction and intensity, and to ascertain the level of ecological degradation persistence. The ecological degradation persistence level was mapped to a preset risk level threshold range; if the slope was greater than zero and the fluctuation amplitude exceeded the threshold, the risk level was determined to be in an upward trend. A dynamic change path was constructed from the risk level upward trend sequence, obtaining a trajectory point set showing the evolution of the risk level over time. This trajectory point set was overlaid and correlated with landscape ecological spatial data to obtain the dynamic change path of the risk level for each landscape unit. The dynamic change path sequence of the risk level for each landscape unit was input into a distributed storage platform to obtain a complete landscape ecological risk assessment file. The process is as follows: Figure 3 As shown.

[0098] For example, in the field of landscape ecological monitoring, time series analysis can be used to perform stationarity tests and differencing on the results of iterative processing. First, the collected ecological degradation index sequences can be subjected to an ADF test to determine their stationarity. If the test shows the presence of a unit root, first-order differencing is performed until the sequence reaches a stationary state, thus obtaining stationary time series data. This process helps eliminate trend and seasonal disturbances, providing a reliable foundation for subsequent modeling.

[0099] Specifically, the autoregressive integral moving average model is used to fit the sequence based on stationary time series data.

[0100] One possible implementation is to use the ARIMA(1,1,1) model to fit the monthly vegetation cover data to obtain a long-term degradation trend curve. This curve can clearly reflect the slow changes in the ecosystem over many years, avoiding misleading judgments due to short-term fluctuations.

[0101] In one embodiment, slope and fluctuation amplitude parameters are extracted from the long-term degradation trend curve. For example, a slope of -0.02 indicates a 2% annual decrease in vegetation cover, and a fluctuation amplitude of 0.15 indicates a negative degradation trend with a moderate intensity. The persistence level of ecological degradation is mapped to a preset risk level threshold range. If the slope is greater than zero and the fluctuation amplitude exceeds the 0.1 threshold, the risk level is determined to be rising. This mapping mechanism can directly transform quantitative indicators into management decision-making criteria, improving the timeliness of risk identification.

[0102] For example, if the slope is greater than zero and the fluctuation amplitude exceeds a threshold, a dynamic change path can be constructed from the risk level ascending state sequence. By connecting the risk level values ​​at each time point, an upward-sloping trajectory curve can be formed, resulting in a set of trajectory points showing the evolution of the risk level over time. This set of points records the gradual transition from low risk to medium-high risk, facilitating the tracing of degradation driving factors.

[0103] Specifically, by overlaying and correlating trajectory point sets with landscape ecological spatial data—for example, projecting trajectory points onto landscape unit grids such as forests, grasslands, and wetlands—the dynamic change paths of risk levels for each landscape unit can be obtained. Forest units may show a continuously rising path, while grassland units may show a relatively gentle path. This differential analysis helps to prioritize high-risk units.

[0104] In one possible implementation, the path sequences dynamically changing according to the risk level of each landscape unit are input into a distributed storage big data platform. Hadoop or similar frameworks can be used to achieve parallel storage and indexing of massive path data, resulting in a complete landscape ecological risk assessment archive. This archive supports rapid querying of historical trajectories and enables cross-year comparative analysis, thereby providing data support for developing targeted restoration measures and improving the continuity of monitoring and decision-making efficiency.

[0105] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for early warning of landscape ecological risks in mining subsidence areas based on big data analysis, characterized in that, Includes the following steps: The time-series variation of subsidence was obtained by synthetic aperture radar interferometry and stored in a big data platform after noise filtering. The cumulative subsidence amount and rate parameters were extracted from the subsidence time series after noise filtering, and a clustering algorithm was used to divide the subsidence areas of different severity levels. Landscape type data and species distribution data are obtained based on the subsidence level areas. A hierarchical response model is constructed based on the obtained data. Ecological vulnerability characteristics are described based on the hierarchical response model. Association rule mining is performed on the ecological vulnerability characteristics to extract vegetation cover and land use conversion patterns. Based on the extracted patterns, the habitat quality change trend is determined; and the ecological risk level is obtained based on the habitat quality change trend. If the ecological risk level exceeds a preset threshold, the early warning threshold system is triggered, the corresponding level of ecological early warning signal is activated, and the updated data of the dynamic assessment system is obtained; the updated data and the subsidence distribution map are integrated, and the parameters of the graded response model are adjusted. Based on the graded response model with adjusted parameters, an optimized sensitivity grading scheme is obtained from the updated data of the dynamic assessment system; based on the optimized sensitivity grading scheme, it is determined whether iterative processing of the analysis of cumulative amount and rate parameters is required to maintain the continuity of landscape sensitivity monitoring. Time series analysis is used to obtain long-term ecological degradation trends from iterative processing results. Based on these trends, the dynamic change path of risk levels is determined and stored in a big data platform to form a complete landscape ecological risk assessment archive, thereby enabling early warning of landscape ecological risks in mining subsidence areas.

2. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, The process of obtaining the temporal variation sequence of subsidence through synthetic aperture radar interferometry and then filtering it for noise includes: Raw surface deformation data of the mining area is acquired from multi-temporal radar images using synthetic aperture radar interferometry (SAR). The raw surface deformation data is then inverted using a phase difference calculation method to obtain preliminary surface deformation results. Noise filtering is applied to these preliminary results using a big data platform, and an adaptive filtering algorithm is used to remove interference signals, resulting in a clear deformation data sequence. Subsidence temporal variation characteristics are extracted from this clear deformation data sequence, and time series analysis methods are applied to segment these characteristics to determine the subsidence trend for each time period. If the subsidence trend for a certain time period exceeds a preset threshold, the deformation data for that time period is weighted to obtain a precise subsidence temporal variation sequence, which is then stored on the big data platform.

3. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, The cumulative subsidence amount and rate parameters were extracted from the subsidence time-series change sequence after noise filtering, and a clustering algorithm was used to divide the subsidence areas into regions with different severity levels, including: Based on the subsidence time-series change sequence after noise filtering, the cumulative subsidence amount and subsidence rate parameters are calculated; the K-means clustering method is used to group the cumulative subsidence amount and subsidence rate parameters to obtain the distribution range of the parameters; if the distribution range of the parameters exceeds the preset threshold range, the corresponding abnormal data is labeled; based on the categories obtained from the grouping process and the abnormal data labeling results, combined with the cumulative subsidence amount parameter, subsidence levels from mild to severe are divided; the subsidence levels are mapped to specific geographical area data to obtain the correspondence between subsidence levels and regional locations, and the subsidence level regional division results are determined; based on the regional division results, a classification label for the subsidence level is generated to determine the severity of subsidence in each region.

4. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, The process of obtaining a description of ecological vulnerability characteristics includes: The spatial overlay matching of the subsidence level areas with the landscape type layers obtained from remote sensing interpretation determines the spatial distribution pattern of the landscape types. Based on the spatial distribution pattern of the landscape types, species distribution data is extracted, and cluster analysis is used to group the species distribution data to determine the aggregation characteristics of the species distribution. A hierarchical response model is constructed based on the aggregation characteristics, and a pre-established response rule base is used to hierarchically map different aggregation characteristics to obtain hierarchical response results. Sensitivity index data is extracted from the hierarchical response results, and the sensitivity index data is standardized to obtain quantitative values ​​of the sensitivity indicators. Vulnerability characteristics are identified based on the quantitative values ​​of the sensitivity indicators. If a certain indicator value exceeds a preset threshold, the corresponding area is marked as a high-vulnerability area, determining the distribution range of the vulnerability characteristics. For the distribution range of the vulnerability characteristics, a text description containing the distribution range and response relationship is generated to obtain a description of the ecological vulnerability characteristics.

5. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, The process of obtaining an ecological risk level includes: Association rule mining is performed on the ecological vulnerability characteristics to extract association rules between vegetation cover and land use conversion. For land use conversion patterns, frequent itemset identification is applied to determine high-frequency conversion patterns. Vegetation cover correspondences are extracted based on these high-frequency conversion patterns to identify habitat quality decline paths. If vegetation cover continuously decreases along a habitat quality decline path, the habitat quality change trend is marked as degraded. The degraded change trend is matched with a pre-established risk mapping relationship to determine the ecological risk level. Regional data is aggregated based on the ecological risk level to obtain the ecological risk level assessment result.

6. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, By integrating updated data and subsidence distribution maps, the parameters of the graded response model are adjusted, including: If the ecological risk level assessment result exceeds a preset threshold, the early warning threshold system is triggered to generate an ecological early warning signal of the corresponding level. The ecological early warning signal activates the dynamic assessment system to obtain real-time monitoring and update data. The real-time monitoring and update data is overlaid with the subsidence distribution map to extract the current subsidence area range. Using a spatial overlay analysis method, the current subsidence area range is spatially intersected with the ecological unit layer to determine the ecological units affected by subsidence. Based on the ecological units affected by subsidence, the parameter weights of the graded response model are adjusted to obtain parameter update values.

7. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, Based on the graded response model with adjusted parameters, an optimized sensitivity grading scheme is obtained from the updated data of the dynamic evaluation system. Based on the optimized sensitivity classification scheme, determine whether iterative processing of the analysis of cumulative amount and rate parameters is required, including: Based on the graded response model with adjusted parameters, basic information on sensitivity grading is extracted from the updated data of the dynamic evaluation system to determine a preliminary sensitivity grading optimization scheme. According to the preliminary sensitivity grading optimization scheme, the corresponding landscape monitoring data is obtained to determine the dynamic trend of landscape sensitivity. If the dynamic trend of landscape sensitivity exceeds a preset threshold, the iterative processing flow is triggered based on the correlation between the changes in the cumulative subsidence amount parameter and the subsidence rate parameter.

8. The method for early warning of landscape and ecological risks in mining subsidence areas based on big data analysis according to claim 1, characterized in that, Time series analysis is used to obtain long-term ecological degradation trends from iterative processing results. Based on these trends, the dynamic change path of risk levels is determined, including: Time series analysis is performed on the iterative processing results. Stationary time series data are obtained through stationarity testing and differencing. An autoregressive integral moving average model is used to fit the stationary time series data to obtain a long-term ecological degradation trend curve. Slope and fluctuation amplitude parameters are extracted from the long-term ecological degradation trend curve to determine the direction and intensity of the ecological degradation trend and the persistence level of ecological degradation. The persistence level of ecological degradation is mapped to a preset risk level threshold range. If the slope is greater than zero and the fluctuation amplitude exceeds the threshold, the risk level is determined to be in an upward trend. A dynamic change path is constructed based on the upward trend of the risk level, resulting in a trajectory point set of risk level evolution over time. This trajectory point set is overlaid and correlated with landscape ecological spatial data to obtain the dynamic change path of the risk level for each landscape unit. The dynamic change path of the risk level for each landscape unit is input into a big data platform for distributed storage, forming a complete landscape ecological risk assessment archive.

Images