A subalpine meadow degradation attribution analysis method fusing a knowledge graph

Abstract

The application relates to the technical field of information processing, in particular to a sub-alpine meadow degradation cause analysis method fusing a knowledge graph, ecological indexes of a meadow are collected, field data is regularized and checked, and missing values are supplemented, ecological element data is integrated, a core index is extracted to mark a change direction to generate a code, a correlation triple is built by relying on grazing vegetation parameters to divide degradation grades, a node is screened and checked in combination with coordinates to remove repeated paths, a continuous degradation chain is combed, a driving transmission result class index is distinguished, and a structured degradation cause result is output. According to the application, ecological data is regularized by unifying a data format and a supplementing rule, exclusive codes are generated by relying on index changes, degradation grades are divided by combining vegetation grazing parameters, element correlation topologies are built by relying on locations, a degradation chain is combed by removing redundant paths, a driving transmission result level is divided, a progressive cause system is built, and a time sequence evolution and a cause and effect transmission analysis dimension are expanded to adapt to a field ecological cause judgment scene.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a method for attribution analysis of subalpine meadow degradation that integrates knowledge graphs. Background Technology

[0002] The field of information processing technology includes knowledge graph construction and data mining analysis. Its core content includes the collection, organization and association of multi-source data, and the structured processing and semantic representation of data such as ecological environment, plant community, soil enzyme activity and microbial quantity. The technology field as a whole includes data modeling, relational network construction, data query and analysis methods, as well as multi-dimensional information management and reasoning based on knowledge graphs.

[0003] One method for attribution analysis of subalpine meadow degradation, which integrates knowledge graphs, involves establishing a knowledge graph for multi-source ecological data on subalpine meadow degradation. This graph uses nodes and edges to represent elements such as plant communities, soil properties, and microbial quantities and their relationships. It covers plant community structure analysis, soil enzyme activity statistics, microbial quantity measurement, and environmental factor organization. Furthermore, it completes data organization, information extraction, and relationship reasoning by constructing semantic association networks and data attribution analysis.

[0004] Existing technologies only complete basic structured processing and semantic association building of multi-source ecological data, and can only achieve shallow reasoning of basic relationships between ecological elements. The overall data processing lacks standardized and regular processes, and the original data is missing and format deviations cannot be effectively corrected. There is no unified standard for defining the state of ecological elements, and it is impossible to build correlations by combining spatial location and dynamic changes of indicators. It is difficult to identify the progressive evolution path and causal hierarchy of ecological degradation. In practical applications, it can only output static ecological data correlation results, and cannot accurately deconstruct the driving transmission logic of meadow degradation, making it difficult to meet the practical application needs of refined degradation attribution analysis. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a method for attribution analysis of subalpine meadow degradation that integrates knowledge graphs. The technical solution is as follows: A method for attribution analysis of subalpine meadow degradation integrating knowledge graphs includes the following steps: S1: Collect various ecological indicators from subalpine meadow plots, perform unified field name replacement and data type validation, replace missing records with similar indicators from plots with the smallest latitude and longitude difference, and splice them in plot number order to obtain an integrated ecological element record set. S2: Extract the records of plant species number, catalase activity, and microbial quantity from the ecological element integration record set, mark the change direction of corresponding indicators in adjacent sample plots and generate combined codes; combine grazing intensity index and vegetation coverage to classify degradation level categories, generate node status identifiers and edge relationship triplet, and obtain a set of degradation state node identifiers. S3: Call the node ID, edge relationship triplet and sample plot latitude and longitude coordinates in the degraded state node identifier set, calculate the latitude and longitude difference of node pairs and filter candidate nodes. After verifying the consistency of the change direction of three indicators, namely the number of plant species, catalase activity and the number of microorganisms, establish node connections, remove duplicate path segments, and obtain the path expression sequence of the degradation process. S4: Compare the change direction of adjacent nodes one by one along the degradation process path expression sequence, disconnect nodes with inconsistent directions, retain multi-node continuous chain segments as valid chains and sort them to generate a degradation causal progressive chain sequence.

[0006] As a further embodiment of the present invention, the field names in S1 are uniformly replaced with synonymous fields of multi-source ecological indicator data mapped to preset standard field names; Data type validation includes validating the value range of numeric fields and validating the format of text fields; When replacing missing records, all sample plot records are traversed, and the latitude and longitude differences between the missing index sample plot and the remaining complete sample plots are calculated. The sample plot with the smallest sum of latitude and longitude differences is selected to fill the missing position with the corresponding index. After filling, the records are spliced ​​one by one in ascending order of sample plot number.

[0007] As a further aspect of the present invention, in S2, the change direction markers are compared one by one with the corresponding index values ​​of adjacent plots in ascending order of plot number. A positive symbol is marked for an increase in value, a negative symbol for a decrease in value, and a zero symbol for a balance. The symbols are then combined in a fixed order of plant species number, catalase activity, and microbial quantity to generate a combined code.

[0008] As a further aspect of the present invention, in step S2, the degradation level classification is based on the preset grazing intensity index range and vegetation coverage range to divide the level gradients, and the two types of gradients are matched one by one to determine the degradation level category corresponding to the individual plots based on the matching results. Then, the degradation level categories are matched with the node status identifiers item by item.

[0009] As a further aspect of the present invention, in step S3, all pairwise node combinations within the set of degraded state node identifiers are traversed, and the absolute values ​​of the longitude difference and latitude difference for each pair of nodes are calculated. The average difference is calculated as the mean of the longitude difference and the mean of the latitude difference for all node pairs. Node pairs whose longitude difference and latitude difference are both less than the corresponding average difference are determined as candidate nodes.

[0010] As a further aspect of the present invention, the consistency verification of the change direction in S3 is to compare the direction symbols of three types of indicators—the number of plant species, catalase activity, and the number of microorganisms—for each candidate node pair. If the three types of symbols are completely consistent, the direction is determined to be consistent. Nodes that meet the conditions are connected by edges in the order of their sample plot numbers; Extract the ordered sequence of node identifiers for each path. Only one path with completely overlapping sequences is retained, and the rest of the duplicate path segments are removed.

[0011] As a further aspect of the present invention, in S4, the node order of the sequence expressed along the degradation process path is compared group by group with all the index direction signs of adjacent nodes. If any type of index direction sign is opposite, it is determined that the directions are inconsistent, and the connection between adjacent nodes is broken. The number of nodes in each chain segment after disconnection is counted. Chain segments with more than 1 node are marked as valid chains. All valid chains are arranged in ascending order according to the sample plot number of the starting node.

[0012] As a further aspect of the present invention, the method further includes: S5: Extract various indicator change records within the degeneration causal progression chain sequence, mark the initial driving item, transmission effect item and result performance item according to the node order, bind the corresponding degeneration level category and group them, and output the degeneration causal attribution structured result; In S5, the nodes of the degenerative causal progression chain sequence are traversed sequentially, and the records of changes in the number of plant species, changes in catalase activity, and changes in the number of microorganisms within the chain are assigned consecutive position numbers according to their order of appearance. The indicator category with the smallest number is marked as the initial driving item, the one with the middle number is marked as the transmission effect item, and the one with the largest number is marked as the result performance item.

[0013] As a further aspect of the present invention, the ecological indicators collected by S1 specifically include the latitude and longitude coordinates of the sample plot, altitude, slope value, aspect angle, grazing intensity index, precipitation, soil pH value, organic matter content, catalase activity, urease activity, number of microorganisms, number of plant species, and vegetation coverage.

[0014] As a further aspect of the present invention, each sample plot in S2 generates a unique node identifier ID, and the structure of the edge relationship triplet is node identifier ID, ecological association relationship, and degradation level category; In S5, the causal attribution structured results of degradation are grouped and aggregated according to the degradation level category to form a set of multi-factor coupling action paths as the final output.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this invention, a unified ecological data field format and gap-filling calibration rules are used to complete the overall data regularization. A unique coding system is generated based on the dynamic change characteristics of multiple core ecological indicators. Degradation level dimensions are defined by linking vegetation and grazing characteristics. A topological structure of element association is built by combining spatial location matching rules. Redundant paths and contradictory data structures are screened out, and a continuous and complete degradation evolution chain is formed. The internal driving transmission and result levels of the ecosystem are distinguished, and a dynamic and progressive attribution analysis system is constructed. The analysis dimensions of the temporal evolution and causal transmission of meadow degradation process are strengthened, which is suitable for in-depth attribution judgment work in complex field ecological scenarios. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the difference between them, their intended meanings are consistent. Similarly, the terms "of," "correlation (ponding)," and "correlation (ponding)" may sometimes be used interchangeably. It should be noted that, without emphasizing the difference between them, their intended meanings are consistent.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] This solution is primarily applied to the field of ecological monitoring in high-altitude mountainous areas, and is adapted to the analysis of field survey data from subalpine meadows with different degradation gradients. It supports grassland management units and ecological research institutions in integrating and analyzing multi-source ecological data on grazing disturbance, soil physicochemical properties, soil enzymes, microorganisms, and plant communities, accurately attributing multi-factor coupling to meadow degradation. It is also suitable for regional meadow degradation risk assessment, vegetation restoration measure deduction, and structured mining of long-term ecological observation datasets. Furthermore, it can be integrated with an ecological information management platform to achieve visualized retrieval of degradation causal paths, providing quantitative and interpretable data support for alpine meadow ecological restoration and grassland management decisions.

[0023] Please see Figure 1 This invention provides a technical solution: a method for attribution analysis of subalpine meadow degradation integrating knowledge graphs, comprising the following steps: S1: Collect various ecological indicators from subalpine meadow plots, perform unified field name replacement and data type validation, replace missing records with similar indicators from plots with the smallest latitude and longitude difference, and splice them in plot number order to obtain an integrated ecological element record set. S2: Extract the records of plant species number, catalase activity, and microbial quantity from the ecological element integration record set, mark the change direction of corresponding indicators in adjacent sample plots and generate combined codes; combine grazing intensity index and vegetation coverage to classify degradation level categories, generate node status identifiers and edge relationship triplet, and obtain a set of degradation state node identifiers. S3: Call the node ID, edge relationship triplet and sample plot latitude and longitude coordinates in the degraded state node identifier set, calculate the latitude and longitude difference of node pairs and filter candidate nodes. After verifying the consistency of the change direction of three indicators, namely the number of plant species, catalase activity and the number of microorganisms, establish node connections, remove duplicate path segments, and obtain the path expression sequence of the degradation process. S4: Compare the change direction of adjacent nodes one by one along the degradation process path expression sequence, disconnect nodes with inconsistent directions, retain multi-node continuous chain segments as valid chains and sort them to generate a degradation causal progressive chain sequence; S5: Extract various indicator change records within the degeneration causal progression chain sequence, mark the initial driving item, transmission effect item and result performance item according to the node order, bind the corresponding degeneration level category and group them, and output the degeneration causal attribution structured result.

[0024] To clearly illustrate the complete implementation process of this knowledge graph-based degradation attribution analysis method, the following section provides a comprehensive description of its implementation using a real-world subalpine meadow field sample dataset. Following the standardized processing sequence defined in the claims, the entire process—including multi-source data cleaning, graph node construction, spatial path selection, causal chain derivation, and multi-factor coupling attribution—is broken down step-by-step into detailed operational rules. The quantitative judgment criteria, parameter calculation logic, and data output specifications for each step are clearly defined, thus fully reconstructing the entire process of implementing the original technical solution.

[0025] In Example 1, during the multi-source ecological element data integration and processing in step S1, a unified replacement operation for field names of multi-source ecological indicator data is first performed. This operation maps synonymous fields with the same meaning but different names in the output data of field sampling equipment, laboratory testing instruments, and remote sensing monitoring platforms to standard field names pre-set by the system, eliminating data matching obstacles caused by inconsistent field names among multiple data sources. Subsequently, data type verification is performed, which is divided into two categories: first, verification of the reasonable value range of numerical fields such as altitude, soil enzyme activity, and microbial quantity. Based on the basic research results of subalpine meadow ecology, the normal value range of each indicator is pre-defined, and abnormal values ​​exceeding the range of ecological common sense are directly eliminated. Second, the format of text fields such as plot number and sampling date is validated, unifying text encoding and character writing standards, and deleting non-compliant text records such as garbled characters and missing prefixes. For plot records with missing indicators, a missing record replacement process is performed. All plot record entries are completely traversed, and the longitude and latitude differences between the target plot with missing indicators and other plots with complete indicators are calculated line by line. The sum of these differences is then calculated, and the nearest complete plot with the smallest sum of longitude and latitude differences is selected. The corresponding indicator value of that plot is extracted and filled into the missing position. The complete calculation logic is described as follows: Let the target missing plot number be... , The corresponding longitude value is Latitude values ​​are ; The complete control plot number for any indicator is , The corresponding longitude value is Latitude values ​​are ; Absolute value of longitude difference between single sample plots Absolute value of latitudinal difference between single sample plots The sum of the latitude and longitude differences of a single sample plot In the formula, The longitude of the sample plot represents the missing indicator. The latitude of the sample plot representing the missing indicator. Representing the A complete control plot longitude, Representing the Latitude of complete control plot This represents the absolute value of the difference in longitude between the two sets of sample plots. This represents the absolute value of the latitude difference between the two sets of sample plots. This represents the sum of the latitude and longitude differences between the two groups of sample plots. Iterate through all complete sample plots, calculating the sum for each group. Numerical values, compare all Size, select For the control plot with the smallest value, the detection value of the corresponding missing indicator is extracted and filled into the blank field of the target plot. After all missing values ​​are filled, all plot records are spliced ​​one by one according to the ascending order of plot number to generate a unified and standardized integrated record set of ecological elements. The ecological indicators collected in this step S1 through field mapping, household grazing surveys, soil physicochemical laboratory testing, microbial isolation and culture counting, vegetation quadrat field surveys, and UAV remote sensing inversion specifically include: plot latitude and longitude coordinates, altitude, slope value, slope aspect angle, grazing intensity index, precipitation, soil pH value, organic matter content, catalase activity, urease activity, microbial quantity, plant species number, and vegetation cover. The above indicators are used to characterize the spatial topographic base conditions of the sample plot, the intensity of human grazing disturbance, the soil acid-base and nutrient base environment, the soil microbial enzymatic metabolic activity, the scale of the underground microbial community, the richness of the aboveground plant community, and the development status of the surface vegetation cover.

[0026] Example 2: Based on the construction of the multi-source ecological element integrated record set completed in Example S1, this example further performs the node encoding generation, meadow degradation level determination and map triplet construction operations in step S2, and completes the quantification of ecological indicator change characteristics and the construction of knowledge graph basic units based on the unified and regular sample plot dataset.

[0027] First, the work of marking and generating combination codes for the direction of indicator changes was carried out. This operation strictly followed the ascending order of the sample plot numbers to retrieve the corresponding indicator values ​​of two adjacent consecutive sample plots one by one. The values ​​of three core ecological indicators—plant species number, catalase activity, and microbial quantity—were compared separately. After the comparison of each type of indicator, a unique symbol was assigned according to the trend of the value change. If the indicator value increased compared to the previous value, it was marked with a positive symbol +; if the indicator value decreased compared to the previous value, it was marked with a negative symbol -; if the indicator values ​​of the two sample plots were completely equal, it was marked with a zero symbol 0. Then, following a fixed order of priority, plant species number, catalase activity, and microbial quantity, the symbols corresponding to the three types of indicators were concatenated in sequence. After the concatenation was completed, a combination code corresponding to the linkage change characteristics of the indicators of adjacent sample plots was obtained. This code can intuitively quantify the synchronous fluctuation pattern of multiple soil-vegetation indicators and provide a standardized judgment basis for subsequent node connection consistency verification.

[0028] Secondly, the classification and matching of degradation levels for individual plots were completed. In the classification process, two key anthropogenic and vegetation characterization indicators, namely grazing intensity index and vegetation coverage, were retrieved from the S1 dataset. Based on the existing research results on the degradation classification of subalpine meadows, specific value ranges for grazing intensity index and vegetation coverage were pre-defined. Based on their respective ranges, the two indicators were divided into multiple grade gradients. Then, the grazing intensity gradient and vegetation coverage gradient were cross-matched one by one. Each matching combination corresponds to a unique and fixed meadow degradation level category. This level category can comprehensively reflect the degree of synergistic degradation between anthropogenic grazing disturbance and surface vegetation development. After the classification was completed, the degradation level category matched for each plot was bound to the specific node status identifier of that plot item by item, so as to achieve complete labeling of the degradation attributes of the node itself.

[0029] Finally, the standardized generation of basic units of the knowledge graph was completed. A set of unique node identifiers was assigned to each independent sample plot record in the dataset to ensure that nodes of different sample plots can be accurately distinguished and located in the knowledge graph. At the same time, edge relationship triples were constructed in a standardized manner. The fixed structure of the triples is as follows: node identifier ID, ecological association relationship, and degradation level category. The ecological association relationship is used to represent the inherent association logic of vegetation, soil, and microbial elements between the current sample plot node and other nodes. The triple fully includes three core information types: single node identity, association attribute, and degradation level. By summarizing all sample plot triples, a complete set of degraded state node identifiers can be formed, providing standardized basic materials for subsequent S3 candidate node selection and path sequence construction.

[0030] Example 3: Based on Example S2, which generates a set of degraded state node identifiers, a unique identifier ID for each node, edge relationship triples, sample plot latitude and longitude coordinates, and symbol codes for changes in three types of indicators, this example further performs step S3, which involves candidate node screening, node edge connection construction, and duplicate path cleanup. Based on the dual screening rules of spatial constraints and indicator changes, a standardized degraded process path expression sequence is generated.

[0031] First, all node IDs, edge relation triples, and sample plot latitude and longitude coordinates bound to each node are retrieved from the degraded state node identifier set. Then, all pairwise node combinations in the degraded state node identifier set are traversed to perform spatial distance quantification calculations. The complete calculation logic is explained below. Let any pair of nodes to be calculated be node A and node B, with node A bound to a longitude value. ,latitude Node B is bound to longitude value latitude value The absolute value of the longitude difference between a single set of nodes The absolute value of the latitude difference between a single set of nodes ; in is the longitude of the sample plot corresponding to node A, is the latitude of the sample plot corresponding to node A, is the longitude of the sample plot corresponding to node B, is the latitude of the sample plot corresponding to node B, AbsDLon represents the absolute value of the longitude difference between the two groups of nodes, and AbsDLat represents the absolute value of the latitude difference between the two groups of nodes. All pairwise node combinations in the set are traversed in sequence to calculate the corresponding AbsDLon and AbsDLat for each group respectively. After the difference calculation of all node pairs is completed, the sum of AbsDLon of all node pairs is divided by the total number of node pairs to obtain the average longitude difference AvgDLon, and the sum of AbsDLat of all node pairs is divided by the total number of node pairs to obtain the average latitude difference AvgDLat. Only when the AbsDLon of a certain group of node pairs is less than AvgDLon and the AbsDLat of this group of node pairs is less than AvgDLat, can this group of node pairs be determined as spatially adjacent candidate nodes. This spatial screening logic is set based on the existing research conclusions on the spatial heterogeneity of subalpine meadow areas. Sample plots with close spatial distances are more likely to form a continuous degradation conduction process, which can eliminate irrelevant node combinations with excessive spatial span and no ecological linkage effect.

[0032] Next, a change direction consistency check is performed on all the screened candidate node pairs. In the check process, the direction signs of the three indicators corresponding to each candidate node pair, namely the number of plant species, catalase activity, and microbial quantity, are retrieved one by one. Only when the signs corresponding to the three groups of indicators are completely identical, can the candidate node pair be determined to meet the constraint condition of consistent change direction. This consistency check rule can ensure that vegetation, soil enzymes and microorganisms between connected nodes fluctuate in the same direction synchronously, avoiding the establishment of invalid associations between nodes with no synergistic change law. For qualified node pairs that have passed both the spatial screening and the sign consistency check, the graph edge connections between nodes are built strictly in the order of sample plot numbers from smallest to largest, so as to ensure that the arrangement of path nodes conforms to the spatial sampling sequence of field sample plots.

[0033] Finally, the ordered sequence of node identifiers corresponding to each path is extracted from all the connected node combinations, and the ordered node sequences of all paths are compared for coincidence one by one. If there are two or more paths with completely coincident arrangement sequences of node identifiers, only any one of the original paths is retained, and all other duplicate path segments are removed simultaneously. After completing the duplicate path cleaning, all remaining effective ordered node connection sequences are integrated, and a standardized degradation process path expression sequence is finally obtained. This sequence provides complete and ordered basic path materials for the chain breaking processing in subsequent S4 and the extraction of effective degradation causal progressive chains; Example 4: Based on Example S3, which completes the removal of duplicate path segments and outputs a standardized degradation process path expression sequence, this example further performs step S4, which involves node direction comparison, link truncation, effective chain screening and sorting. It splits continuous paths based on the matching rules of three types of ecological indicator change symbols to generate a degradation causal progression chain sequence with complete transmission logic.

[0034] First, following the predetermined node arrangement order within the degradation process path expression sequence, adjacent nodes in each group are extracted for a full indicator direction sign comparison. During the comparison process, the direction signs of three types of indicators bound to each adjacent node are retrieved simultaneously: the number of plant species, catalase activity, and the number of microorganisms. If the positive, negative, or zero signs of any of the three types of indicators are opposite, it is directly determined that the change direction between adjacent nodes in this group is inconsistent, and a disconnection operation is performed at the map connection position of this group of adjacent nodes. This judgment logic is based on the existing ecological theory of the degradation chain transmission in subalpine meadows. Only node combinations where all core indicators change synchronously in the same direction have a continuous degradation-driven transmission relationship. A single indicator fluctuating in the opposite direction indicates that the degradation transmission chain is interrupted at this point, and there is no continuous causal relationship effect.

[0035] Secondly, after all the disconnection processes were completed, the number of nodes contained in each of the independent chain segments was counted. Only continuous chain segments with more than 1 node were marked as valid chains. Segments containing only a single node did not have a transmission relationship between ecological elements before and after the breakup and were therefore not valuable for causal analysis and were directly discarded. This was done to filter out isolated node segments that were not meaningful for analysis.

[0036] Finally, all the selected valid chains are collected, and the plot number corresponding to the starting node of each valid chain is extracted. All valid chains are then rearranged in ascending order of plot number value from smallest to largest. After sorting, all the ordered valid chains are integrated to generate a complete degenerate causal progressive chain sequence. This sequence has a clear logical hierarchy and continuous transmission relationship, which can directly provide a standardized chain analysis carrier for subsequent S5 multi-factor coupling attribution and the division of driving and result items.

[0037] Example 5: Based on Example S4, which completes link truncation, filtering, and orderly arrangement of all effective chain segments to obtain a complete degradation causal progression chain sequence, this example further executes step S5, which involves index hierarchy division, degradation level binding, and result aggregation. It distinguishes the three ecological action levels—driving, transmission, and result—based on the order of appearance of indicators within the chain, and integrates and outputs a structured degradation causal attribution result that can be directly used for ecological assessment.

[0038] First, all change records corresponding to the three types of indicators—number of plant species, catalase activity, and number of microorganisms—stored within the degradation causal progression chain sequence were extracted completely. Then, following the node arrangement order inherent in the degradation causal progression chain sequence, all indicator records within each chain were traversed segment by segment. For each record of changes in the number of plant species, catalase activity, and number of microorganisms within the chain, a continuous position number was assigned in strict accordance with the order in which the indicators first appeared in the chain transmission path. This numbering can intuitively reflect the sequential action of multiple ecological factors in the meadow degradation process. The sequential action is the core criterion for distinguishing causal levels, which is consistent with the classic degradation transmission mechanism in ecosystems: human interference first changes vegetation, then disturbs soil enzymes, and finally affects the microbial community.

[0039] Secondly, based on the continuous position numbers obtained from the allocation of various indicators, hierarchical labeling is completed. The indicator category with the smallest position number value within a single chain is uniformly marked as the starting driving item, representing the source influencing factor that triggers meadow degradation. The index category with the central position number is uniformly marked as the transmission effect item, representing the intermediate mediating factor that receives the source disturbance and transmits the degradation effect downward; The category of indicators with the largest position number is uniformly marked as the result performance item, representing the ecosystem response characterization formed by the transmission of degradation to the end. Through the three types of labels, the causal logic within each transmission chain can be clearly deconstructed, realizing the hierarchical analysis of multi-factor coupling relationships.

[0040] Finally, the entire causal chain with completed driver / transmission / result hierarchical marking is bound one by one to the degradation level category corresponding to the node in the chain. Then, all chains are grouped and aggregated according to different degradation level categories. All causal chains under the same degradation level are summarized into an independent subset. After the integration of all subsets, a complete set of multi-factor coupling action paths is formed. This set is the degradation causal attribution structured result finally output in step S5, which can directly support researchers and grassland management units to carry out subsequent application work such as gradient meadow degradation mechanism analysis and targeted restoration scheme simulation.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A method for attribution analysis of subalpine meadow degradation integrating knowledge graphs, characterized in that, Includes the following steps: S1: Collect various ecological indicators from subalpine meadow plots, perform unified field name replacement and data type validation, replace missing records with similar indicators from plots with the smallest latitude and longitude difference, and splice them in plot number order to obtain an integrated ecological element record set. S2: Extract the records of plant species number, catalase activity, and microbial quantity from the ecological element integration record set, mark the change direction of corresponding indicators in adjacent sample plots and generate combined codes; combine grazing intensity index and vegetation coverage to classify degradation level categories, generate node status identifiers and edge relationship triplet, and obtain a set of degradation state node identifiers. S3: Call the node ID, edge relationship triplet and sample plot latitude and longitude coordinates in the degraded state node identifier set, calculate the latitude and longitude difference of node pairs and filter candidate nodes. After verifying the consistency of the change direction of three indicators, namely the number of plant species, catalase activity and the number of microorganisms, establish node connections, remove duplicate path segments, and obtain the path expression sequence of the degradation process. S4: Compare the change direction of adjacent nodes one by one along the degradation process path expression sequence, disconnect nodes with inconsistent directions, retain multi-node continuous chain segments as valid chains and sort them to generate a degradation causal progressive chain sequence.

2. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, In S1, the field names are uniformly replaced with synonymous fields of multi-source ecological indicator data mapped to preset standard field names. Data type validation includes validating the value range of numeric fields and validating the format of text fields; When replacing missing records, all sample plot records are traversed, and the latitude and longitude differences between the missing index sample plot and the remaining complete sample plots are calculated. The sample plot with the smallest sum of latitude and longitude differences is selected to fill the missing position with the corresponding index. After filling, the records are spliced ​​one by one in ascending order of sample plot number.

3. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, In S2, the direction of change markers are compared one by one with the corresponding index values ​​of adjacent plots in ascending order of plot number. A positive symbol is marked for an increase in value, a negative symbol for a decrease, and a zero symbol for a balance. The symbols are then combined in a fixed order of plant species number, catalase activity, and microbial quantity to generate a combined code.

4. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, In S2, the degradation level classification is based on the preset grazing intensity index range and vegetation coverage range to divide the level gradients. The two gradients are matched one by one to determine the degradation level category corresponding to the individual plots. Then, the degradation level category is matched with the node status identifier item by item.

5. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, In step S3, all pairwise node combinations within the degraded state node identifier set are traversed, and the absolute values ​​of the longitude difference and latitude difference for each pair of nodes are calculated. The average difference is calculated as the mean of the longitude difference and the mean of the latitude difference for all node pairs. Node pairs whose longitude difference and latitude difference are both less than the corresponding average difference are determined as candidate nodes.

6. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, The consistency verification of the change direction in S3 is to compare the direction signs of three types of indicators of candidate node pairs: the number of plant species, catalase activity, and the number of microorganisms. If the three types of signs are completely consistent, the direction is considered to be consistent. Nodes that meet the conditions are connected by edges in the order of their sample plot numbers; Extract the ordered sequence of node identifiers for each path. Only one path with completely overlapping sequences is retained, and the remaining duplicate path segments are removed.

7. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, In S4, the node order of the sequence expressed along the degradation process path is compared group by group with all the index direction signs of adjacent nodes. If any index direction sign is opposite, it is determined that the directions are inconsistent and the connection between adjacent nodes is broken. The number of nodes in each chain segment after disconnection is counted. Chain segments with more than 1 node are marked as valid chains. All valid chains are arranged in ascending order according to the sample plot number of the starting node.

8. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, The method further includes: S5: Extract various indicator change records within the degeneration causal progression chain sequence, mark the initial driving item, transmission effect item and result performance item according to the node order, bind the corresponding degeneration level category and group them, and output the degeneration causal attribution structured result; In S5, the nodes of the degenerative causal progression chain sequence are traversed sequentially, and the records of changes in the number of plant species, changes in catalase activity, and changes in the number of microorganisms within the chain are assigned consecutive position numbers according to their order of appearance. The indicator category with the smallest number is marked as the initial driving item, the one with the middle number is marked as the transmission effect item, and the one with the largest number is marked as the result performance item.

9. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 1, characterized in that, The ecological indicators collected by S1 specifically include the plot's latitude and longitude coordinates, altitude, slope value, aspect angle, grazing intensity index, precipitation, soil pH value, organic matter content, catalase activity, urease activity, number of microorganisms, number of plant species, and vegetation coverage.

10. The method for attribution analysis of subalpine meadow degradation fused with knowledge graphs according to claim 8, characterized in that, In S2, each sample plot is generated with a unique node identifier ID, and the structure of the edge relationship triplet is node identifier ID, ecological association relationship, and degradation level category; In S5, the causal attribution structured results of degradation are grouped and aggregated according to the degradation level category to form a set of multi-factor coupling action paths as the final output.

Images