Natural resource multi-element linkage ecosystem mapping method and system

Abstract

The present application relates to the technical field of ecosystem mapping, and particularly relates to a natural resource multi-element linkage ecosystem mapping method and system; the method comprises collecting multi-source natural resource investigation and monitoring data, and performing semantic cleaning and fusion to construct a natural resource data bottom plate; based on an artificial intelligence coupling model, the natural resource data bottom plate is mapped to an initial ecosystem classification map; the update flow of the multi-source natural resource investigation and monitoring data is continuously monitored, and data change events are automatically detected; based on multi-source change information, a change intensity index corresponding to the change event is calculated; and a target ecosystem classification map is generated. Through the technology of data fingerprint comparison, the multi-source data flow is continuously monitored, and change events are automatically captured. For the detected changes, the system performs intelligent classification and decision-making, and through the knowledge graph-based polygon aggregation, the automation and intelligentization of mapping are realized under the premise of ensuring data topology and area balance.

Description

Technical Field

[0001] This invention relates to the field of ecosystem mapping technology, and in particular to an ecosystem mapping method and system that links multiple elements of natural resources. Background Technology

[0002] With the increasing urgency of ecological civilization construction and refined management of natural resources, how to accurately and dynamically map the distribution of ecosystems has become a key technical challenge. Currently, the field of natural resource management has formed a rich resource of massive data from multiple sources, multiple elements, and continuous updates, such as data from annual land change surveys, comprehensive monitoring of forests, grasslands, and wetlands, land spatial planning, and farmland quality monitoring. Theoretically, this data provides the possibility for high-precision and up-to-date ecosystem mapping. Existing ecosystem mapping methods usually rely on a single or a small number of data sources, using static classification models to classify data at a certain point in time in a one-time manner, generating static ecosystem classification maps. For data updates, existing technologies often adopt the method of periodically rerunning the entire classification process, or manually or semi-automatically updating local changes through simple rules. At the same time, at the data processing level, there is a lack of an automated patch integration and correction mechanism that is deeply coupled with the ecosystem classification logic to address mapping "noise" such as small patches and narrow patches caused by data collection, updating, or model errors.

[0003] Chinese Patent Publication No. CN119313026A discloses a method, apparatus, equipment, and mapping system for ecological zoning of plateau transportation corridors. The scheme acquires meteorological station data, digital elevation model data, and land use type data within the study area based on the vector range of the plateau transportation corridor ecological zoning study area. Based on the above data, the surface relief and ecosystem type of the study area are obtained. A method for interpolating missing temperature data from meteorological stations is proposed, and the missing temperature data is imputed. Based on the imputed temperature data, surface relief, and ecosystem type, and combined with a preset mapping relationship, three zoning indicators for the study area—plateau climate zone type, plateau landform type, and road area ecological type—are determined and created. These three indicators are spatially coupled to obtain coupling results. Based on the coupling results, multi-indicator-patch clustering analysis is performed to obtain clustering results. Based on the clustering results, the study area is ecologically zoned. It is evident that existing ecosystem mapping methods, facing massive, multi-source data streams, lack lightweight, high-performance automatic change detection, making it difficult to achieve real-time discovery and response to change events, resulting in low update monitoring efficiency and low classification accuracy of ecosystem classification maps. Summary of the Invention

[0004] To address this, the present invention provides an ecosystem mapping method and system that links multiple elements of natural resources, in order to overcome the problems in existing technologies that lack lightweight, high-performance automatic change detection when faced with massive, multi-source data streams, resulting in low update monitoring efficiency and low classification accuracy of ecosystem classification maps.

[0005] To achieve the above objectives, on the one hand, the present invention provides an ecosystem mapping method involving multiple elements of natural resources, comprising:

[0006] Collect multi-source natural resource survey and monitoring data, perform semantic cleaning and fusion, and construct a natural resource data base.

[0007] Based on an artificial intelligence coupling model, the natural resource data base is mapped to an initial ecosystem classification map;

[0008] Continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events;

[0009] The detected data change events are classified into a first category of events and a second category of events;

[0010] For the second type of event, a change intensity index corresponding to the change event is calculated based on multi-source change information. The change intensity index is then compared and processed with the type stability threshold of the corresponding ecosystem type.

[0011] If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered;

[0012] If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base.

[0013] The initial ecosystem classification map is updated based on the revised natural resource data base to generate the target ecosystem classification map.

[0014] Furthermore, based on an artificial intelligence coupling model, mapping the natural resource data base to an initial ecosystem classification map includes:

[0015] For each patch object in the natural resource data base, extract and fuse its multi-source attribute data to construct a unified multi-dimensional feature vector.

[0016] The multi-dimensional feature vector is input into a pre-trained artificial intelligence coupling model; the artificial intelligence coupling model is configured to perform the following operations:

[0017] Through its internal attention mechanism layer, the features in the multi-dimensional feature vector are weighted to generate a weighted feature representation;

[0018] The weighted feature representation is input into a fully connected classification network to calculate the probability that the patch object belongs to each preset ecosystem type;

[0019] For each patch object, the ecosystem type with the highest probability value is selected from the probabilities output by the fully connected classification network as the final ecosystem classification result for the patch object, and the highest probability value is used as the confidence level of the classification result.

[0020] Based on the final ecosystem classification results, corresponding confidence levels, and spatial geometric information of all map objects, the initial ecosystem classification map is generated and output.

[0021] Furthermore, the process of classifying the detected data change events includes:

[0022] If the data change event is associated with a legitimate administrative approval document, it is classified as a first-category event, and the type update of the initial ecosystem classification map is directly triggered.

[0023] If the data change event is not associated with a legitimate administrative approval document, it is classified as a Category II event.

[0024] Furthermore, the process of continuously monitoring the update stream of the multi-source natural resource survey and monitoring data includes:

[0025] A dynamic data fingerprint is established and maintained for each data source. The data fingerprint is a unique identifier calculated based on the data content and is used to characterize the complete data state of the data source at a specific point in time.

[0026] By periodically recalculating the data fingerprint and comparing it with the previously stored fingerprint value, the update stream of each data source is continuously monitored.

[0027] When the current data fingerprint of any data source is inconsistent with the previous fingerprint, it is automatically determined that the data source has changed, and the corresponding data change event is extracted and generated.

[0028] The multi-source natural resource survey and monitoring data includes several of the aforementioned data sources.

[0029] Furthermore, the process of calculating the change intensity index corresponding to the change event based on multi-source change information includes:

[0030] Calculate the area adjustment coefficient based on the changed area;

[0031] Calculate the time urgency coefficient based on the type of emergency being monitored;

[0032] The change intensity index is calculated based on the inherent difficulty of type conversion, the area adjustment coefficient, and the time urgency coefficient.

[0033] Furthermore, triggering the patch aggregation and correction process of the natural resource data base to obtain the process of correcting the natural resource data base includes:

[0034] Based on the change areas provided by the second type of events, change patches of each ecosystem that have changed are extracted from the initial ecosystem classification map;

[0035] Based on each of the aforementioned ecosystem change patches, the corresponding original patches are extracted from the natural resource data base and used as patches to be processed.

[0036] For each of the aforementioned unprocessed map features, a graph traversal is performed in the natural resource data base knowledge graph to determine the target map features to be merged, and map feature aggregation is then performed.

[0037] Calculate the real-time total area change rate of the target merged patches, and verify the aggregation results based on the comparison between the standard total area change rate range and the real-time total area change rate;

[0038] The aggregated correction results that pass the verification will be synchronously updated to the natural resource data base and the natural resource data base knowledge graph, generating the corrected natural resource data base and the corrected natural resource data base knowledge graph.

[0039] Furthermore, the process of identifying target merged patches and performing patch aggregation includes:

[0040] The type of the patch to be processed is determined by comparing area thresholds, average width, local morphological clustering analysis, and automatic topological rule checking, resulting in Class I, Class II, Class III, and Class IV problem patches.

[0041] Select the patch aggregation method according to the type and priority of the patches to be processed;

[0042] Perform image aggregation according to the corresponding image aggregation method;

[0043] The priority order is: fourth aggregation method > third aggregation method > second aggregation method > first aggregation method;

[0044] The image patch aggregation methods include a first aggregation method, a second aggregation method, a third aggregation method, and a fourth aggregation method;

[0045] The first aggregation method is to merge the patch to be processed into the target merged patch, and update the attribute information according to the patch attribute type;

[0046] The second aggregation method is to assign the patch to be processed to the target merged patches on both sides of the adjacent side;

[0047] The third aggregation method is to perform a process that simplifies the boundary contour;

[0048] The fourth aggregation method is to perform gap processing or overlap processing based on the problem type.

[0049] Furthermore, the process of verifying the aggregation results based on the comparison between the standard total area change rate range and the real-time total area change rate includes:

[0050] Compare the standard total area change rate range with the real-time total area change rate:

[0051] If the real-time total area change rate is within the standard total area change rate range, the test is passed and the aggregation correction result is obtained;

[0052] If the real-time total area change rate is not within the standard total area change rate range, the sign of the real-time total area change rate shall be determined as follows:

[0053] If the real-time total area change rate is negative, the target merged patches are reselected and merged by adding strong constraints.

[0054] If the real-time total area change rate is positive, execute the segmentation and release strategy to reselect target patches for merging.

[0055] Furthermore, the process of calculating the real-time total area change rate of the target merged patches is as follows:

[0056] The inspection area is determined based on the minimum bounding radius and buffer zone of each patch to be processed;

[0057] Extract the original total area of ​​each category within the test area from the natural resource data base before aggregation;

[0058] Extract the new total area of ​​each category within the test area from the aggregated and corrected natural resource data base;

[0059] Calculate the percentage of the new total area that exceeds the original total area to obtain the real-time total area change rate.

[0060] On the other hand, the present invention also provides an ecosystem mapping system that links multiple elements of natural resources, and the above-mentioned ecosystem mapping method that links multiple elements of natural resources includes:

[0061] The data fusion and processing module collects multi-source natural resource survey and monitoring data, performs semantic cleaning and fusion, and constructs a natural resource data base.

[0062] The classification and mapping module, which is connected to the data fusion and processing module, is used to map the natural resource data base into an initial ecosystem classification map based on an artificial intelligence coupling model.

[0063] The change detection module is connected to the data fusion processing module and the classification mapping module respectively, and is used to continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events.

[0064] An event classification module, which is connected to the data fusion processing module and the change detection module respectively, is used to classify detected data change events into a first category of events and a second category of events;

[0065] A response correction module, connected to the event classification module, calculates a change intensity index corresponding to the second type of event based on multi-source change information, and compares and processes the change intensity index with the type stability threshold of the corresponding ecosystem type.

[0066] If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered;

[0067] If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base.

[0068] The linkage update module is connected to the event classification module, the response correction module, the data fusion processing module, and the classification mapping module, respectively, and is used to update the initial ecosystem classification map according to the corrected natural resource data base to generate the target ecosystem classification map.

[0069] Compared with existing technologies, the beneficial effects of this invention are as follows: by deeply semantically cleaning and fusing survey and monitoring data from multiple departments such as land, forestry and grassland, planning, and farmland protection, a logically consistent natural resource data base is constructed. This achieves deep fusion and coordinated updating of multi-source heterogeneous natural resource data under a unified ecological semantic framework, improving the quality and accuracy of the data base in ecological mapping. Furthermore, an artificial intelligence model simulating the concept of "micro-macro mechanical coupling" is used. This model assesses the importance of various ecological characteristics through an attention mechanism and makes comprehensive decisions through a fully connected network, mapping the data base with high precision to an initial ecosystem classification map with confidence, thus improving the scientific rigor and accuracy of ecosystem classification. To achieve dynamic updates, a technology based on "data fingerprint" comparison is used to continuously monitor multi-source data streams and automatically capture change events. For detected changes, the system performs intelligent classification and decision-making: if the change is accompanied by legal approval documents, a type update is directly triggered; if it is a natural or unapproved change, the change intensity index is calculated and compared with a type stability threshold to determine whether it is a "qualitative change" in ecological type or a "quantitative change" in state. In response to quantitative changes, the system initiates a knowledge graph-based patch aggregation and correction process. This process can intelligently identify problematic patches such as small patches and narrow patches, optimize and merge them based on semantic proximity and spatial rules, and pass the area change rate test to finally generate a corrected data base and classification map. Through knowledge graph-based patch aggregation, while ensuring data topology and area balance, the system achieves automated and intelligent comprehensive summarization of mapping results. Ultimately, it efficiently and accurately produces and maintains a highly up-to-date "single ecological map," providing strong technical support for national and regional ecological protection, system governance, and macro-level decision-making. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating the ecosystem mapping method involving multiple elements of natural resources according to an embodiment of the present invention.

[0071] Figure 2 This is a flowchart illustrating the process of continuously monitoring the update stream of the multi-source natural resource survey and monitoring data in an embodiment of the present invention.

[0072] Figure 3 This is a flowchart illustrating the process of calculating the change intensity index corresponding to a change event in an embodiment of the present invention.

[0073] Figure 4 This is a schematic diagram of the structure of an ecosystem mapping system that links multiple elements of natural resources, according to an embodiment of the present invention. Detailed Implementation

[0074] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0075] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0076] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0077] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0078] Please see Figure 1 As shown, it is a flowchart illustrating the ecosystem mapping method involving multiple elements of natural resources according to an embodiment of the present invention. The present invention provides an ecosystem mapping method involving multiple elements of natural resources, including:

[0079] Step S1: Collect multi-source natural resource survey and monitoring data, perform semantic cleaning and fusion, and construct a natural resource data base.

[0080] Step S2: Based on the artificial intelligence coupling model, the natural resource data base is mapped to an initial ecosystem classification map;

[0081] Step S3: Continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events;

[0082] Step S4: Classify the detected data change events into a first category of events and a second category of events;

[0083] Step S5: For the second type of event, calculate the change intensity index corresponding to the change event based on multi-source change information, and compare and process the change intensity index with the type stability threshold of the corresponding ecosystem type.

[0084] If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered;

[0085] If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base.

[0086] Step S6: Update the initial ecosystem classification map based on the modified natural resource data base to generate the target ecosystem classification map.

[0087] In this embodiment, the data sources include land change survey data, forest, grassland and wetland comprehensive monitoring data, land spatial planning data, and cultivated land quality monitoring data. These systems generate massive amounts of multi-element (covering forest land, grassland, water area, construction land, cultivated land, etc.) and multi-attribute (land type, area, ownership, quality, etc.) geospatial data every year, forming a "data goldmine" for land cover monitoring. The process of triggering the type update of the initial ecosystem classification map includes: in response to the first event type, updating the ecosystem type of the corresponding area in the initial ecosystem classification map.

[0088] By deeply semantically cleaning and integrating survey and monitoring data from multiple departments such as land resources, forestry and grassland, planning, and farmland protection, a logically consistent natural resource data base is constructed. This achieves deep integration and coordinated updating of multi-source heterogeneous natural resource data under a unified ecological semantic framework, improving the quality and accuracy of the data base in ecological mapping. An artificial intelligence model simulating the concept of "micro-macro mechanical coupling" is used. This model assesses the importance of various ecological characteristics through an attention mechanism and makes comprehensive decisions through a fully connected network, mapping the data base with high precision into an initial ecosystem classification map with confidence, thus improving the scientific rigor and accuracy of ecosystem classification. To achieve dynamic updates, a technology based on "data fingerprint" comparison is used to continuously monitor multi-source data streams and automatically capture change events. For detected changes, the system performs intelligent classification and decision-making: if the change is accompanied by legal approval documents, a type update is directly triggered; if it is a natural or unapproved change, the change intensity index is calculated and compared with a type stability threshold to determine whether it is a "qualitative change" in ecological type or a "quantitative change" in state. In response to quantitative changes, the system initiates a knowledge graph-based patch aggregation and correction process. This process can intelligently identify problematic patches such as small patches and narrow patches, optimize and merge them based on semantic proximity and spatial rules, and pass the area change rate test to finally generate a corrected data base and classification map. Through knowledge graph-based patch aggregation, while ensuring data topology and area balance, the system achieves automated and intelligent comprehensive summarization of mapping results. Ultimately, it efficiently and accurately produces and maintains a highly up-to-date "single ecological map," providing strong technical support for national and regional ecological protection, system governance, and macro-level decision-making.

[0089] Specifically, based on an artificial intelligence coupling model, mapping the natural resource data base to an initial ecosystem classification map includes:

[0090] Feature vector construction steps: For each patch object in the natural resource data base, extract and fuse its multi-source attribute data to construct a unified multi-dimensional feature vector;

[0091] Model processing steps: The multi-dimensional feature vector is input into a pre-trained artificial intelligence coupling model; the artificial intelligence coupling model is configured to perform the following operations:

[0092] Through its internal attention mechanism layer, the features in the multi-dimensional feature vector are weighted to generate a weighted feature representation;

[0093] The weighted feature representation is input into a fully connected classification network to calculate the probability that the patch object belongs to each preset ecosystem type;

[0094] Classification decision steps: For each patch object, select the ecosystem type with the highest probability value from the probabilities output by the fully connected classification network as the final ecosystem classification result for the patch object, and use the highest probability value as the confidence level of the classification result;

[0095] Map generation steps: Based on the final ecosystem classification results, corresponding confidence levels, and spatial geometric information of all map objects, generate and output the initial ecosystem classification map.

[0096] In this embodiment, the multi-dimensional features of each patch are treated as "microscopic single-layer material properties." An attention mechanism (simulating constitutive relations) is used to evaluate the importance of each feature. Finally, a fully connected classification network (simulating macroscopic performance prediction of laminates) integrates all information to output a high-precision ecosystem classification and confidence level. In the feature vector construction step, a natural resource data base is input. Each patch in this base is associated with multi-source attribute data from land, forestry, remote sensing, geography, and planning. For each patch, key features are extracted from its attribute table and standardized and vectorized to form a unified mathematical representation. For example, the feature vector F of a patch can be constructed as follows:

[0097] F = [Land type code: embedding vector, vegetation coverage: numerical value, average tree height: numerical value, dominant tree species: embedding vector, altitude: numerical value, slope: numerical value, annual average NDVI: numerical value, whether it is located within the ecological red line: binary identifier];

[0098] The output is a feature vector of dimension d, representing all relevant quantitative information for ecological classification of the patch. In the model processing step, the pre-trained AI coupling model is a deep learning network, its core consisting of a multi-head self-attention layer and a fully connected classification layer connected sequentially. The function of the attention mechanism layer (micro-weighted) is as follows: this layer receives the feature vector F, internally learns the dependencies between different features through self-attention calculation, and automatically assigns appropriate attention weights αi to each feature dimension. The magnitude of the weight αi dynamically represents the impact of the feature (e.g., "average tree height") on the final ecosystem within the context of the current specific patch. The contribution importance of the overall classification is determined by the following operation: F' = Attention(F) = Σ(αi×fi), where fi is a component of the feature vector F. This step transforms the original features into a "weighted feature representation F'", simulating the difference in the overall contribution of different "material single-layer" performances (i.e., "constitutive relation"). The fully connected classification network (macro-prediction) receives the weighted feature representation F'. This network consists of multiple fully connected layers and nonlinear activation functions. Its role is to perform nonlinear transformation and deep fusion on all weighted feature information. The specific operation is as follows: the last layer of the network usually uses the Softmax activation function. The final output is normalized into a probability distribution vector P=[p1,p2,...,pk], where k is the total number of preset ecosystem types, such as: broadleaf forest, coniferous forest, shrubland, grassland, urban ecosystem, etc.; pj represents the predicted probability that the map patch belongs to the j-th ecosystem type. In the classification decision step, the decision logic is as follows: for the output probability vector P, the argmax function is used to find the index j with the highest probability: j=argmax(P). The ecosystem type corresponding to index j is determined as the final ecosystem classification result for the map patch. At the same time, the maximum value pj is used as the confidence level of this classification. For example, if P(broadleaf forest) = pj, then the probability distribution vector P(broadleaf forest) is determined as pj. If P(shrubland) = 0.92 and P(shrubland) = 0.07, then the classification result is "broadleaf forest" with a confidence level of 0.92. A high confidence level indicates that the model has a high degree of certainty in its judgment, while a low confidence level result can be marked for manual verification. In the map generation step, the spatial geometric information (such as boundary coordinates), the final ecosystem classification result and its confidence level of all map features are collected. Then, using Geographic Information System (GIS) technology, the above information is used to generate a thematic vector map. In this map feature, the graphic is rendered according to its classification result (such as different colors), and its attribute table contains at least two fields: "ecosystem type" and "classification confidence level". This is the initial ecosystem classification map.

[0099] For example, suppose we process map patch number A-101:

[0100] Feature vector F: After extraction, some key feature values ​​are [Land type: 0301, Dominant tree species: Oak, Average tree height: 8.2 meters, Vegetation coverage: 68%, Annual average NDVI: 0.74] (numericalized); the attention layer assigns a high weight to "Average tree height (8.2 meters)" (e.g., α=0.6) because it is key to distinguishing between trees and shrubs; "Dominant tree species (Oak)" is second (α=0.3); "Vegetation coverage" has a relatively low weight (α=0.1). The fully connected network calculates probabilities based on weighted features: P = {broadleaf forest: 0.91, shrubland: 0.08, grassland: 0.01}; the type corresponding to the highest probability of 0.91 is selected, so the final result is "broadleaf forest ecosystem" with a confidence level of 0.91; patch A-101 is assigned the color representing "broadleaf forest" on the map, and its attribute table records {type: broadleaf forest, confidence level: 0.91}; after performing this operation on all patches, a complete classification map is generated.

[0101] Specifically, the process of classifying the detected data change events includes:

[0102] If the data change event is associated with a legitimate administrative approval document, it is classified as a first-category event, and the type update of the initial ecosystem classification map is directly triggered.

[0103] If the data change event is not associated with a legitimate administrative approval document, it is classified as a Category II event.

[0104] See Figure 2 As shown, it is a flowchart illustrating the continuous monitoring of the update stream of the multi-source natural resource survey and monitoring data in an embodiment of the present invention;

[0105] Specifically, the process of continuously monitoring the update stream of the multi-source natural resource survey and monitoring data includes:

[0106] Step S301: Establish and maintain a dynamic data fingerprint for each data source. The data fingerprint is a unique identifier calculated based on the data content and is used to characterize the complete data state of the data source at a specific point in time.

[0107] Step S302: By periodically recalculating the data fingerprint and comparing it with the previously stored fingerprint value, the update stream of each data source is continuously monitored.

[0108] Step S303: When the current data fingerprint of any data source is inconsistent with the previous fingerprint, it is automatically determined that the data source has changed, and the corresponding data change event is extracted and generated.

[0109] The multi-source natural resource survey and monitoring data includes several of the aforementioned data sources.

[0110] In this embodiment, the data fingerprint is a fixed-length string generated based on a cryptographic hash function. Its calculation input is a combination of all core data content or key fields of the data source at a certain moment, ensuring that even if the data changes slightly, the fingerprint value will produce a significant difference, i.e., the avalanche effect. The cryptographic hash function can be SHA-256 or MD5. Since the fingerprint value uniquely corresponds to a specific data state, it is extremely sensitive to any addition, deletion, or modification of data content, ensuring that no detection is missed. Compared with record-by-record comparison, fingerprint comparison is a string comparison with O(1) complexity, which greatly improves the monitoring efficiency, especially suitable for massive data. Continuous monitoring of the multi-source natural resource survey monitoring The specific process of updating the test data flow is as follows: The monitoring service creates a fingerprint record for each access data source (such as a land change survey database table), which includes the data source ID, the last known fingerprint, and the last check time. It starts a monitoring poll at a set time interval (such as every 5 minutes): it calls the dedicated interface of each data source to obtain a snapshot of all its current data, and calculates its SHA-256 hash value as the current fingerprint. It compares the current fingerprint with the last known fingerprint recorded by the data source. If they do not match, it immediately triggers a data change event and updates the last known fingerprint to the current fingerprint; if they match, it records this check and waits for the next round of monitoring query cycle.

[0111] The process of continuously monitoring the update stream of the multi-source natural resource survey and monitoring data includes a data fingerprint generation algorithm and a monitoring and comparison strategy. The data fingerprint generation algorithm includes the generation of vector patch data, raster / image data, and attribute database tables. The generation process of vector patch data is as follows: for each patch, its spatial geometric hash (such as the SHA-256 value of the boundary coordinate sequence) and key attribute summary (such as the hash of the concatenated string of land use code, area, and ownership) are extracted, and the two are hashed again to generate the fingerprint of the patch. The table-level fingerprint is generated by hashing the fingerprints of all patches in sequence. The generation process of raster / image data is as follows: the hash value of the concatenated string of image metadata (row and column number, resolution, time) and statistical values ​​(mean and variance of all pixel values) is calculated. The generation process of attribute database tables is as follows: after sorting by primary key, the values ​​of key fields (such as identifier code, change time, and land use type) are concatenated to generate row-level fingerprints, which are then aggregated into table-level fingerprints. The monitoring and comparison strategy includes periodic incremental monitoring and change event extraction. Periodic incremental monitoring involves calculating the fingerprint of each data source table every 5 minutes. Instead of recalculating the entire table, it only calculates the aggregate fingerprint of records added or modified since the last monitoring and compares it with the baseline fingerprint. Change event extraction involves extracting the specific patch identifier, change type (add, delete, modify), and key attributes before and after the change through SQL query or spatial comparison when the fingerprint changes, and encapsulating them into standardized data change events.

[0112] See Figure 3 As shown, it is a flowchart illustrating the process of calculating the change intensity index corresponding to a change event in an embodiment of the present invention.

[0113] Specifically, the process of calculating the change intensity index corresponding to the change event based on multi-source change information includes:

[0114] Step S5001: Calculate the area adjustment coefficient based on the changed area;

[0115] Step S5002: Calculate the time urgency coefficient based on the monitoring emergency type;

[0116] Step S5003: Calculate the change intensity index based on the inherent difficulty of type conversion, area adjustment coefficient, and time urgency coefficient.

[0117] The formula for calculating the intensity of change index in this embodiment is:

[0118] ;

[0119] Wherein, DF: Change Intensity Index (typically ranging from 0 to 1 or 0 to 100).

[0120] : The i-th change factor;

[0121] : The normalized value (0-1) of the i-th factor;

[0122] The weight of the i-th factor ;

[0123] Ga: Area adjustment coefficient;

[0124] Gtime: Time urgency factor;

[0125] The formula for calculating the type stability threshold is:

[0126] AE = ICD × (1 + α × ECF + β × PCF + γ × CST), where:

[0127] AE: Type stability threshold (value range is usually >0, a baseline value can be set).

[0128] ICD: The inherent difficulty of type conversion (benchmark value)

[0129] ECF: Ecological Constraint Factor

[0130] PCF: Policy Constraint Factor

[0131] CST: Gap in compliance with classification criteria

[0132] α, β, γ: adjustment coefficient, α=0.3, β=0.5, γ=0.2;

[0133] In this embodiment, the change intensity index is the driving force, which characterizes the potential value of ecological transformation and quantifies the dimensionless value of the intensity of the event's promotion of ecosystem type transformation; the type stability threshold is the activation energy, which is a critical impedance value determined comprehensively based on factors such as the natural stability of the target ecosystem type, the definition of classification criteria, and regional management policies, and is used to determine whether the type will be changed by external change events.

[0134] Specifically, the process of triggering the patch aggregation and correction process of the natural resource data base to obtain the corrected natural resource data base includes:

[0135] Based on the change areas provided by the second type of events, change patches of each ecosystem that have changed are extracted from the initial ecosystem classification map;

[0136] Based on each of the ecosystem change patches, the corresponding original patches are extracted from the natural resource data base and used as patches to be processed.

[0137] For each of the aforementioned unprocessed map features, a graph traversal is performed in the natural resource data base knowledge graph to determine the target map features to be merged, and map feature aggregation is then performed.

[0138] Calculate the real-time total area change rate of the target merged patches, and verify the aggregation results based on the comparison between the standard total area change rate range and the real-time total area change rate;

[0139] The aggregated correction results that pass the verification will be synchronously updated to the natural resource data base and the natural resource data base knowledge graph, generating the corrected natural resource data base and the corrected natural resource data base knowledge graph.

[0140] The knowledge graph of the natural resource data base includes entities, attributes, and relationships. Entities are centered around each map patch in the natural resource data base, and extend to entities such as "administrative units" and "ecological functional zones." Map patch entity attributes include land type, area, perimeter, center coordinates, administrative region, ecological function type, and change timestamp. Multiple semantic relationships are defined, such as adjacent (spatial topology), belonging (administrative affiliation), having similar ecological functions (based on attribute rules), and historical version association (time series). The construction process of the natural resource data base knowledge graph is as follows: During system initialization, the data base is automatically traversed, and map patches and their attributes are instantiated as nodes and attributes of the knowledge graph; adjacent semantic relationships are established through spatial analysis (such as determining boundary intersections); and semantic relationships are established through a rule engine (such as matching land type codes with ecological function categories).

[0141] Map patch aggregation includes semantic proximity calculation and target patch merging decision-making. Semantic proximity calculation comprehensively measures spatial proximity and semantic similarity. For example, it calculates whether two patches are spatially adjacent (if so, a bonus is awarded), and whether their land use type and ecological function attributes are the same or compatible (based on a preset land use type conversion rule table), and then calculates a weighted total score. Target patch merging decision-making involves traversing the knowledge graph along edges related to adjacency and similar ecological functions for the small patches to be processed, searching for the neighboring patches with the highest semantic proximity as the recommended merging targets.

[0142] Specifically, the process of identifying target mergeable patches and performing patch aggregation includes:

[0143] The type of the patch to be processed is determined by comparing area thresholds, average width, local morphological clustering analysis, and automatic topological rule checking, resulting in Class I, Class II, Class III, and Class IV problem patches.

[0144] Select the patch aggregation method according to the type and priority of the patches to be processed;

[0145] Perform image aggregation according to the corresponding image aggregation method;

[0146] The priority order is: fourth aggregation method > third aggregation method > second aggregation method > first aggregation method;

[0147] The image patch aggregation methods include a first aggregation method, a second aggregation method, a third aggregation method, and a fourth aggregation method;

[0148] The first aggregation method is to merge the patch to be processed into the target merged patch, and update the attribute information according to the patch attribute type;

[0149] The second aggregation method is to assign the patch to be processed to the target merged patches on both sides of the adjacent side;

[0150] The third aggregation method is to perform a process that simplifies the boundary contour;

[0151] The fourth aggregation method involves performing gap processing or overlap processing based on the problem type. In this embodiment, the priority of the map patch aggregation method is selected so that when the same map patch meets multiple conditions, the fourth method is executed first. Problem patches are categorized as follows: Category 1: small patches; Category 2: narrow patches; Category 3: narrow-necked patches; Category 4: topologically conflicting patches. The identification of small patches is based on the area thresholds of the map standard and scale. The identification process is as follows: determine the target scale for the current mapping or monitoring, for example, 1:10000; determine the minimum map area for different land types according to the specifications. For example, according to the "Land Use Status Classification" standard, on a 1:10,000 topographic map, residential patches should be no less than 4 mm², cultivated land no less than 6 mm², and other land types no less than 15 mm²; convert the map area to the actual area according to the scale, using the formula: Actual area threshold (square meters) = Map area threshold (mm²) × (Scale denominator)² × 10 -12 Calculate the actual area of ​​each map patch and compare it with the actual area threshold for the corresponding land type. If the area is less than the threshold, it is considered a small map patch. For example, if the target scale is 1:10000, the land type is cultivated land, and the minimum area on the map is 6 mm², the actual area threshold is calculated as: 6 × (10,000)² × 10 -12 =600 square meters. If the actual area of ​​a cultivated land patch is 450 square meters < 600 square meters, it is identified as a small patch. The identification of narrow and elongated patches is based on calculating the average width using area and perimeter. The specific process is as follows: calculate the area (S) and perimeter (L) of the patch, and use the equivalent rectangle method to calculate the average width (W). This method assumes that the patch can be equivalent to a rectangle with the same area and perimeter. The formula is: The average width threshold is determined based on data specifications and regional characteristics. It can be set to 2 meters or 5 meters, with 5 meters being the preferred setting. The calculated average width (W) is compared with the threshold; if it is less than the threshold, it is considered a narrow and elongated patch. For example, a patch with the following parameters: area S = 1200 square meters, perimeter L = 200 meters, average width W = (200- / 4=(200- / 4=(200- / 4≈(200-144.22) / 4≈13.95 meters. Since 13.95 meters > 5 meters, this patch is not a narrow patch. The identification of the thin-necked patch is based on the local morphological analysis of the constrained Delaunay triangulation (CD-TIN). The specific process is as follows: Construct a constrained Delaunay triangulation (CD-TIN) for the patch boundary, ensuring that the triangulation covers the interior of the patch and is consistent with the boundary. Calculate the ratio of the longer side to the shorter side or the minimum interior angle of each triangle to analyze the triangle shape. Set a triangle "narrowness" threshold: the ratio of the longer side to the shorter side > 5, or the minimum interior angle > 5. Triangles with a minor interior angle <10° are marked as "narrow triangles" and subjected to spatial clustering analysis. Spatially adjacent narrow triangles are grouped together. For each narrow triangle cluster, its cluster length and average width are calculated. A necking threshold is set, i.e., an aspect ratio threshold of 4. If the aspect ratio of a narrow triangle cluster exceeds the threshold, the patch is considered to have a "neck" at that location. For example, through triangulation analysis, five adjacent narrow triangles are clustered at a certain location in the patch. The cluster length (along the patch's direction) is approximately 45 meters, and the average width (cluster area / length) is approximately 8 meters. Calculation: 45 / 8 = 5.6. If the aspect ratio threshold is 4, since 5.6 > 4, the polygon is identified as having a "neck" at this location. The identification of topologically conflicting polygons is based on automated checks using GIS topology rules. The specific process is as follows: Establish topology check rules, including in-plane seamlessness and in-plane non-overlap. In-plane seamlessness checks for uncovered blank areas between adjacent polygons, while in-plane non-overlap checks for illegal overlapping areas between polygons. This is achieved by creating a topology dataset in GIS software (such as ArcGIS), loading polygon data, and adding... Add the above rules, set an appropriate topology tolerance (0.001 meters; gaps or overlaps smaller than this value will be ignored), run a topology check, and polygons that violate the rules will be automatically marked and an error record will be generated, namely "topology conflict polygons". For example, set the topology tolerance to 0.001 meters, and obtain the check results: there is a blank area of ​​0.8 square meters between polygon A and polygon B (>tolerance), the system reports a "gap" error, and both A and B are marked; polygons C and D overlap each other with an area of ​​1.5 square meters, the system reports an "overlap" error, and both C and D are marked.

[0152] In this embodiment, the process of updating attribute information according to the attribute type of the patch is as follows: For continuous and summable numerical attributes, the updated attribute information is calculated according to the area weight. Applicable attributes include area, estimated yield, carbon storage, etc. The update process is: the attribute value of the new patch B' = the attribute value of the original patch B + the sum of the attribute values ​​of all merged small patches (A1, A2...); for example, the target forest land B has an area of ​​50 hectares and a carbon storage of 500 tons. A small grassland A is merged with it, with an area of ​​2 hectares and a carbon storage of 10 tons. The area of ​​the new map patch B' is 52 hectares, and its carbon storage is 510 tons. For discrete, descriptive, and exclusive attributes, inheritance is based on rules, requiring reliance on knowledge graphs or business rule bases. The main rules include "subject absorption" rules, "priority" rules, "status downgrade" rules, and "weighted average or aggregation" rules. The "subject absorption" rule prioritizes the attributes of the target map patch (subject), ignoring the attributes of smaller map patches. Applicable scenarios include land use code / name (the merged land use code is usually consistent with the subject), dominant function, and administrative division. For example, when merging a small patch of scattered grassland into adjacent forest land, the new map patch's "land use name" should inherit "arbor forest land" from the forest land, not "natural pasture land." The "priority" rule compares the priority or importance of the attributes of the merged parties, inheriting the higher priority. Applicable scenarios include protection level (e.g., nature reserve > general forest land) and approval status (approved). > Unapproved), ownership type (state ownership > collective ownership), for example, merging a small "general cultivated land" patch into a "permanent basic farmland" patch; since "permanent basic farmland" has a higher protection level, the "protection level" attribute of the new patch should inherit the level of "permanent basic farmland"; the "state downgrade" rule requires certain attributes to be set to null or a specific state after being merged, applicable scenarios are independent project numbers, special identifiers, indicating that after the small patch is merged, its independent project identity disappears, for example, a small patch belonging to "a certain ecological restoration project" is merged into a large natural forest, the "project number" attribute of the new patch may be set to null because it is no longer an independent artificial restoration plot; the "weighted average or aggregation" rule requires that for attributes such as "average slope" and "average vegetation index", a new average value needs to be calculated by area weighting, the specific process is new attribute value = (B area × B attribute value + Σ(A i Area × A i Attribute value)) / New total area.

[0153] Specifically, the process of verifying the aggregation results by comparing the standard total area change rate range with the real-time total area change rate includes:

[0154] Compare the standard total area change rate range with the real-time total area change rate:

[0155] If the real-time total area change rate is within the standard total area change rate range, the test is passed and the aggregation correction result is obtained;

[0156] If the real-time total area change rate is not within the standard total area change rate range, the sign of the real-time total area change rate shall be determined as follows:

[0157] If the real-time total area change rate is negative, the target merged patches are reselected and merged by adding strong constraints.

[0158] If the real-time total area change rate is positive, execute the segmentation and release strategy to reselect target patches for merging.

[0159] When the real-time total area change rate is determined to be outside the standard total area change rate range and is negative, it indicates that too many small patches have been merged into other land types, resulting in an excessive reduction in the total area of ​​this type. In this case, the patch with the longest shared boundary or closest location to the patches in this type is selected as the target patch for merging. The source patch is redirected and merged into other patches in this type to compensate for the area loss. This search adds a strong constraint: the target patch must be limited to this type. The patch with the longest shared boundary or closest location to the target patch in this type is selected for merging, and the change rate is recalculated. If it is still below the lower limit, this process is repeated until the standard is met or no further adjustments are possible. For example, "cultivated land..." "The standard total area change rate range is [-1%, +1%]. The original total area of ​​cultivated land in the test area is 100.0 hectares. This is because the aggregation system merged three small cultivated land patches (areas of 0.3, 0.4, and 0.5 hectares, totaling 1.2 hectares) into the semantically adjacent 'rural roads' and 'ditches,' resulting in a new total cultivated land area of ​​100.0 - 1.2 = 98.8 hectares. The real-time total area change rate is (98.8 - 100.0) / 100.0 × 100% = -1.2%, which is considered a failure to pass the test (excessive loss). The 'cultivated land' land category is locked. Three merging operations that caused cultivated land loss were found. The system prioritizes reverting the '0' operation, sorting by area from largest to smallest." The operation of "merging 0.5 hectares of cultivated land patch into ditch" was performed, and a new target was found for this 0.5 hectare cultivated land patch, with the constraint "cultivated land only". The map recommended merging it into an adjacent large cultivated land patch. After correction, the new total cultivated land area = 98.8 + 0.5 = 99.3 hectares. The real-time total area change rate was then checked: (99.3 - 100.0) / 100.0 × 100% = -0.7%. The correction passed the check, and "cultivated land should be prioritized for merging within this category" was recorded as a high-priority rule in the natural resource data base knowledge graph. When the real-time total area change rate is determined to be outside the standard total area change rate range and the real-time total area change rate is positive, it indicates that because... If too many small patches of other land types are merged into the local type, or if merging within the local type results in an excessively large patch, leading to an artificially inflated area, then the land type with the excessive area (let's call it land type Y) is located, and all aggregation operations that significantly increase its area are identified. Special attention is paid to the operation that contributes the most to the area of ​​a single merge or the super-large patch formed after the merge. The source patches that constitute this patch are analyzed to determine which source patches belong to other land types. The parts that originally belonged to other land types are separated from this large patch, and the separated parts are re-merged with the original land type patches that are semantically closer. The real-time total area change rate is recalculated for determination. For example, the standard total area change rate range for "forest land" is [-2%, +4%].The original total area of ​​forest land within the test area was 200.0 hectares. Because the system merged two "Other Grassland" patches (areas of 1.5 and 2.0 hectares, totaling 3.5 hectares) and one "Orchard" patch (1.0 hectare) into an adjacent large forest land, the new total forest land area became 200.0 + 3.5 + 1.0 = 204.5 hectares. The real-time total area change rate was (204.5 - 200.0) / 200.0 × 100% = +2.25% < +4%, thus passing the test. The system then merged another "Other Grassland" patch (area 4.0 hectares) into the same large forest land, resulting in a new total forest land area of ​​204.5 + 4.0 = 208.5 hectares. The real-time total area change rate was (208.5 - 200.0) / 200.0 × 100% = +4.25%. If the expansion exceeds 4% (upper limit), the system fails the test (overexpansion exceeds the limit). The system determines that directly reverting to the last operation may not be the optimal solution and adopts a more refined segmentation and release strategy. It analyzes the composition of the forest patch and selects to separate the most recently merged 4.0 hectares of grassland from the forest. In the knowledge graph, it searches for a new target for this 4.0 hectares of grassland, recommending merging it into another adjacent "other grassland". After correction, the total forest area = 208.5 - 4.0204.5 hectares. The system then checks again, calculating the real-time total area change rate = (204.5 - 200.0) / 200.0 × 100% = +2.25% < +4%, indicating that the correction passed the test. The system learns the lesson: "When merging forest areas, be wary of incorporating large areas of grassland to avoid triggering the expansion limit."

[0160] Specifically, the process of calculating the real-time total area change rate of the merged target patches is as follows:

[0161] The inspection area is determined based on the minimum bounding radius and buffer zone of each patch to be processed;

[0162] Extract the original total area of ​​each category within the test area from the natural resource data base before aggregation;

[0163] Extract the new total area of ​​each category within the test area from the aggregated and corrected natural resource data base;

[0164] Calculate the percentage of the new total area that exceeds the original total area to obtain the real-time total area change rate.

[0165] In this embodiment, the minimum bounding range of each patch to be processed is the geometric envelope or convex hull of all patches to be processed. This range is the minimum bounding range of all patches that are directly merged, simplified, or split. The buffer zone is determined based on proximity. The buffer distance should be at least greater than the distance between all patches in the patch to be processed and their farthest neighbor. The buffer distance can be formed by taking 1-2 times the average minimum bounding rectangle width of all patches in the data base as the buffer distance. The minimum bounding range and the buffer zone are geometrically merged to form a continuous, possibly irregular polygonal area. The merged area is used to trim or intersect the complete natural resource data base, thereby obtaining a pure inspection area composed of complete patches.

[0166] See Figure 4 As shown, this is a schematic diagram of the structure of an ecosystem mapping system with multi-element linkage of natural resources according to an embodiment of the present invention. The present invention also provides an ecosystem mapping system with multi-element linkage of natural resources. The ecosystem mapping method with multi-element linkage of natural resources according to this embodiment includes:

[0167] The data fusion and processing module collects multi-source natural resource survey and monitoring data, performs semantic cleaning and fusion, and constructs a natural resource data base.

[0168] The classification and mapping module, which is connected to the data fusion and processing module, is used to map the natural resource data base into an initial ecosystem classification map based on an artificial intelligence coupling model.

[0169] The change detection module is connected to the data fusion processing module and the classification mapping module respectively, and is used to continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events.

[0170] An event classification module, which is connected to the data fusion processing module and the change detection module respectively, is used to classify detected data change events into a first category of events and a second category of events;

[0171] A response correction module, connected to the event classification module, calculates a change intensity index corresponding to the second type of event based on multi-source change information, and compares and processes the change intensity index with the type stability threshold of the corresponding ecosystem type.

[0172] If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered;

[0173] If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base.

[0174] The linkage update module is connected to the event classification module, the response correction module, the data fusion processing module, and the classification mapping module, respectively, and is used to update the initial ecosystem classification map according to the corrected natural resource data base to generate the target ecosystem classification map.

[0175] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0176] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for ecosystem mapping that integrates multiple natural resource elements, characterized in that, include: Collect multi-source natural resource survey and monitoring data, perform semantic cleaning and fusion, and construct a natural resource data base. Based on an artificial intelligence coupling model, the natural resource data base is mapped to an initial ecosystem classification map; Continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events; The detected data change events are classified into a first category of events and a second category of events; For the second type of event, a change intensity index corresponding to the change event is calculated based on multi-source change information. The change intensity index is then compared and processed with the type stability threshold of the corresponding ecosystem type. If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered; If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base. The initial ecosystem classification map is updated based on the revised natural resource data base to generate the target ecosystem classification map; Based on an artificial intelligence coupling model, mapping the natural resource data base to an initial ecosystem classification map includes: For each patch object in the natural resource data base, extract and fuse its multi-source attribute data to construct a unified multi-dimensional feature vector. The multi-dimensional feature vector is input into a pre-trained artificial intelligence coupling model; the artificial intelligence coupling model is configured to perform the following operations: Through its internal attention mechanism layer, the features in the multi-dimensional feature vector are weighted to generate a weighted feature representation; The weighted feature representation is input into a fully connected classification network to calculate the probability that the patch object belongs to each preset ecosystem type; For each patch object, the ecosystem type with the highest probability value is selected from the probabilities output by the fully connected classification network as the final ecosystem classification result for the patch object, and the highest probability value is used as the confidence level of the classification result. Based on the final ecosystem classification results, corresponding confidence levels, and spatial geometric information of all map objects, the initial ecosystem classification map is generated and output.

2. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 1, characterized in that, The process of classifying the detected data change events includes: If the data change event is associated with a legitimate administrative approval document, it is classified as a first-category event, and the type update of the initial ecosystem classification map is directly triggered. If the data change event is not associated with a legitimate administrative approval document, it is classified as a Category II event.

3. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 1, characterized in that, The process of continuously monitoring the update stream of the multi-source natural resource survey and monitoring data includes: A dynamic data fingerprint is established and maintained for each data source. The data fingerprint is a unique identifier calculated based on the data content and is used to characterize the complete data state of the data source at a specific point in time. By periodically recalculating the data fingerprint and comparing it with the previously stored fingerprint value, the update stream of each data source is continuously monitored. When the current data fingerprint of any data source is inconsistent with the previous fingerprint, it is automatically determined that the data source has changed, and the corresponding data change event is extracted and generated. The multi-source natural resource survey and monitoring data includes several of the aforementioned data sources.

4. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 1, characterized in that, The process of calculating the change intensity index corresponding to the change event based on multi-source change information includes: Calculate the area adjustment coefficient based on the changed area; Calculate the time urgency coefficient based on the type of emergency being monitored; The change intensity index is calculated based on the inherent difficulty of type conversion, the area adjustment coefficient, and the time urgency coefficient.

5. The ecosystem mapping method involving multiple natural resource elements according to claim 1, characterized in that, The process of triggering the patch aggregation and correction flow of the natural resource data base to obtain the corrected natural resource data base includes: Based on the change areas provided by the second type of events, change patches of each ecosystem that have changed are extracted from the initial ecosystem classification map; Based on each of the ecosystem change patches, the corresponding original patches are extracted from the natural resource data base and used as patches to be processed. For each of the aforementioned unprocessed map features, a graph traversal is performed in the natural resource data base knowledge graph to determine the target map features to be merged, and map feature aggregation is then performed. Calculate the real-time total area change rate of the target merged patches, and verify the aggregation results based on the comparison between the standard total area change rate range and the real-time total area change rate; The aggregated correction results that pass the verification will be synchronously updated to the natural resource data base and the natural resource data base knowledge graph, generating the corrected natural resource data base and the corrected natural resource data base knowledge graph.

6. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 5, characterized in that, The process of identifying target mergeable features and performing feature aggregation includes: The type of the patch to be processed is determined by comparing area thresholds, average width, local morphological clustering analysis, and automatic topological rule checking, resulting in Class I, Class II, Class III, and Class IV problem patches. Select the patch aggregation method according to the type and priority of the patches to be processed; Perform image aggregation according to the corresponding image aggregation method; The priority order is: fourth aggregation method > third aggregation method > second aggregation method > first aggregation method; The image patch aggregation methods include a first aggregation method, a second aggregation method, a third aggregation method, and a fourth aggregation method; The first aggregation method is to merge the patch to be processed into the target merged patch, and update the attribute information according to the patch attribute type; The second aggregation method is to assign the patch to be processed to the target merged patches on both sides of the adjacent side; The third aggregation method is to perform a process that simplifies the boundary contour; The fourth aggregation method is to perform gap processing or overlap processing based on the problem type.

7. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 5, characterized in that, The process of verifying the aggregation results based on the comparison between the standard total area change rate range and the real-time total area change rate includes: Compare the standard total area change rate range with the real-time total area change rate: If the real-time total area change rate is within the standard total area change rate range, the test is passed and the aggregation correction result is obtained; If the real-time total area change rate is not within the standard total area change rate range, the sign of the real-time total area change rate shall be determined as follows: If the real-time total area change rate is negative, the target merged patches are reselected and merged by adding strong constraints. If the real-time total area change rate is positive, execute the segmentation and release strategy to reselect target patches for merging.

8. The ecosystem mapping method based on the multi-element linkage of natural resources according to claim 5, characterized in that, The process of calculating the real-time total area change rate of the merged target patches is as follows: The inspection area is determined based on the minimum bounding radius and buffer zone of each patch to be processed; Extract the original total area of ​​each category within the test area from the natural resource data base before aggregation; Extract the new total area of ​​each category within the test area from the aggregated and corrected natural resource data base; Calculate the percentage of the new total area that exceeds the original total area to obtain the real-time total area change rate.

9. An ecosystem mapping system that integrates multiple natural resource elements, employing the ecosystem mapping method for integrating multiple natural resource elements as described in any one of claims 1-8, characterized in that, include: The data fusion and processing module collects multi-source natural resource survey and monitoring data, performs semantic cleaning and fusion, and constructs a natural resource data base. The classification and mapping module, which is connected to the data fusion and processing module, is used to map the natural resource data base into an initial ecosystem classification map based on an artificial intelligence coupling model. The change detection module is connected to the data fusion processing module and the classification mapping module respectively, and is used to continuously monitor the update stream of the multi-source natural resource survey and monitoring data, and automatically detect data change events. An event classification module, which is connected to the data fusion processing module and the change detection module respectively, is used to classify detected data change events into a first category of events and a second category of events; A response correction module, connected to the event classification module, calculates a change intensity index corresponding to the second type of event based on multi-source change information, and compares and processes the change intensity index with the type stability threshold of the corresponding ecosystem type. If the change intensity index is greater than the type stability threshold, the type update of the initial ecosystem classification map is triggered; If the change intensity index is less than or equal to the type stability threshold, the patch aggregation and correction process of the natural resource data base is triggered to obtain the corrected natural resource data base. The linkage update module is connected to the event classification module, the response correction module, the data fusion processing module, and the classification mapping module, respectively, and is used to update the initial ecosystem classification map according to the corrected natural resource data base to generate the target ecosystem classification map.

Images