A method and system for tracking and assessing the evolution of high-temperature heat waves based on a topology-sensing graph spectral framework.
By using a topology-aware graph spectrum framework, the problem of defining the identity of high-temperature heat wave events and separating topology diagnosis is solved, enabling repeatable tracking and multi-dimensional assessment of high-temperature heat wave events. This overcomes the fragmentation of events and differences in scale indicators in existing technologies, and improves the accuracy and applicability of the assessment.
Patent Information
- Application Number
- CN202611520878.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-29
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies cannot simultaneously define the identity and diagnose the topological context of high-temperature heat waves within the same framework, resulting in event fragmentation and differences in scale index representation, making it difficult to fully characterize the expansion, contraction, propagation, and reorganization processes of high-temperature heat waves.
By adopting a topology-aware graph-based hierarchical framework, a high-temperature threshold transcendence mask is constructed, spatial objects in the heat wave are identified, candidate directed edges are constructed and continuity scores are calculated to determine the backbone edges and candidate edges, thereby separating event identity definition from topology diagnosis and preserving the split and merged topology context.
It enables repeatable tracking of high-temperature heat wave events, avoids event fragmentation, improves the physical rationality and cross-regional applicability of disaster exposure assessment, can accurately identify splitting and merging processes, and provides multi-dimensional event assessment indicators.
Smart Images

Figure CN123115585A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of meteorological disaster monitoring, climate extreme event analysis, geospatial information processing and spatiotemporal map data mining, and in particular to a method and system for tracking and evaluating the evolution of high-temperature heat waves based on a topological sensing map spectrum framework. Background Technology
[0002] Heat waves are extreme climate events that have significant impacts on human health, ecosystems, agriculture, energy systems, and socioeconomic activities. As the frequency, intensity, and duration of extreme heat vary across many regions, relying solely on single-grid frequency, duration, or intensity indicators is insufficient to fully characterize the expansion, contraction, propagation, deformation, and reorganization of a spatially coherent high-temperature system throughout its lifecycle.
[0003] Existing event tracking methods typically describe event continuity using three-dimensional spatiotemporal connectivity, one-to-one connections of overlapping objects, or centroid trajectories. While these methods can effectively characterize overall continuity and propagation, one-to-one connections may lead to event fragmentation when daily heatwave objects split or merge, and centroid trajectories may lose spatial footprint information when objects undergo significant deformation. On the other hand, while pure connectivity representation can maintain overall connectivity, it may not explicitly distinguish which connections are used to define event identity and which connections only reflect local branches or convergences.
[0004] Existing technologies cannot separate "event identity definition" and "topological context diagnosis" within the same framework. It is difficult to simultaneously define the event footprint with repeatable and deterministic backbone connections, while retaining other candidate connections that meet the spatial continuity condition to explicitly identify splits, mergers, and their combined evolution. This leads to differences in the way event connections and identity are defined, resulting in event fragmentation and differences in the representation of event scale indicators. Summary of the Invention
[0005] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for tracking and evaluating the evolution of high-temperature heat waves based on a topology-aware graph spectral framework.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0007] This invention provides a method for tracking and evaluating the evolution of high-temperature heat waves based on a topology-aware graph spectral framework, comprising the following steps:
[0008] S1. Input the daily maximum temperature raster data of the target study area, calculate the daily high temperature threshold, construct a high temperature threshold transcending mask, and apply a target area spatial mask to complete the clipping.
[0009] S2. Perform eight-neighbor connected component labeling on the high temperature threshold exceeding the mask after clipping, identify daily heat wave spatial objects, remove objects smaller than the preset object size threshold, construct graph nodes for the retained heat wave spatial objects and record object attributes.
[0010] S3. Construct candidate directed edges only between heat wave spatial objects corresponding to two strictly adjacent calendar days, calculate source overlap rate, target overlap rate, and spatial intersection-union ratio, filter and obtain retained candidate edges according to preset retention conditions, and calculate the continuity score of retained candidate edges based on source overlap rate, target overlap rate, and spatial intersection-union ratio.
[0011] S4. Determinely sort the candidate outgoing edges of each source node and the candidate incoming edges of each target node according to their continuity scores; define the candidate edge that is both the optimal outgoing edge of the source node and the optimal incoming edge of the target node as the backbone edge, construct a one-to-one directed backbone spectrum along the backbone edge, and define the event footprint with the backbone spectrum; perform topology diagnosis and identification of split nodes and merged nodes based on the in-degree and out-degree of all candidate edges.
[0012] S5. Based on the event footprints corresponding to the main spectrum, calculate the event evaluation index, classify the heat wave event evolution type according to the topological diagnosis results, and complete the high temperature heat wave evolution tracking and evaluation output.
[0013] Furthermore, in step S1, the daily high temperature threshold is the temperature percentile threshold under the calendar window centered on the target day within the reference climate period; the high temperature threshold exceeding the mask is generated by the indicator function judging that the highest temperature of the grid cell exceeds the temperature percentile threshold.
[0014] Furthermore, in step S2, the recorded object attributes include node identifier, date, spatial mask, centroid, number of mesh cells, and physical area.
[0015] Furthermore, in step S3, the candidate directed edge retention condition is: the two objects have at least one shared grid cell and satisfy the maximum directional overlap rate is greater than or equal to the directional overlap threshold, or the spatial intersection-union ratio is greater than or equal to the IoU threshold; the continuity score is obtained by multiplying the spatial intersection-union ratio, source direction overlap rate, and target direction overlap rate by their respective weights and summing them, with the sum of each weight being 1.
[0016] Furthermore, in step S4, the deterministic sorting rule for candidate edges prioritizes continuity score, followed by spatial intersection-union ratio, and then the number of spatial intersection grid cells; when all multiple indicators are the same, the order in which candidate edges are generated is used as the deciding rule; when screening the main lineage, the main lineage that reaches the preset minimum duration of days and has consecutive node dates is retained; main lineage nodes with a candidate out-degree greater than 1 are judged as split nodes, and main lineage nodes with a candidate in-degree greater than 1 are judged as merged nodes.
[0017] Furthermore, in step S5, the event evaluation indicators include duration, maximum physical area, cumulative area-daily exposure, topology transformation number, and topology complexity; the heat wave event evolution types are divided into four categories: topology preservation, splitting, merging, and split-merging; the topology transformation number is the sum of the number of splitting node roles and the number of merging node roles; the topology complexity is the topology transformation number divided by the total number of main lineage nodes; and the cumulative area-daily exposure is the sum of the physical areas of the event footprint grid cells on each date within the event lifecycle.
[0018] This invention also provides a high-temperature heat wave evolution tracking and assessment system based on a topology-aware graph spectral framework, comprising:
[0019] Data Input and Mask Construction Module: Used to input daily maximum temperature raster data of the target study area, calculate daily high temperature threshold, construct a high temperature threshold transcending mask, and apply a spatial mask of the target area to complete the clipping;
[0020] Heatwave object recognition module: used to perform eight-neighbor connected component labeling on the high temperature threshold exceeding the mask after clipping, identify daily heatwave spatial objects, remove objects smaller than the preset object size threshold, construct graph nodes for the retained heatwave spatial objects and record object attributes;
[0021] Candidate edge construction and scoring module: This module is used to construct candidate directed edges only between heat wave spatial objects corresponding to two strictly adjacent calendar days, calculate the source overlap rate, target overlap rate, and spatial intersection-union ratio, filter and obtain retained candidate edges according to preset retention conditions, and calculate the continuity score of the retained candidate edges based on the source overlap rate, target overlap rate, and spatial intersection-union ratio.
[0022] The backbone spectrum reconstruction and topology diagnosis module is used to deterministically sort the candidate outgoing edges of each source node and the candidate incoming edges of each target node based on their continuity scores; it defines the candidate edges that are simultaneously the optimal outgoing edges of the source node and the optimal incoming edges of the target node as backbone edges, constructs a one-to-one directed backbone spectrum along the backbone edges, and defines the event footprint using the backbone spectrum; and performs topology diagnosis based on the in-degree and out-degree of all candidate edges to identify split nodes and merged nodes.
[0023] Event assessment and output module: Based on the event footprints corresponding to the main spectrum, it calculates event assessment indicators, classifies heat wave event evolution types according to topological diagnosis results, and completes high-temperature heat wave evolution tracking and assessment output.
[0024] Compared with the prior art, the technical solution disclosed in this invention has the following beneficial effects:
[0025] 1. This invention uses a two-layer representation mechanism of candidate directed edges and backbone edges. The backbone edges are used to define the unique identity of an event, while the remaining candidate edges serve as topological context for diagnosing splits, merging, and evolving topologies. This enables the event identity definition and the branch convergence topological context to be preserved simultaneously within the same object graph framework, overcoming the deficiency of existing technologies where event identity and topological evolution cannot be represented synchronously.
[0026] 2. This invention limits the establishment of candidate edges to strictly adjacent calendar day objects, and at the same time uses a deterministic sorting rule of multi-dimensional indicators to filter the backbone edges, thereby improving the repeatability of tracking results, avoiding the generation of non-continuous object connections across time, and alleviating the problem of event fragmentation.
[0027] 3. This invention is based on the approximate calculation of the physical area of the grid cell on the spherical Earth, and uses this to calculate the cumulative area-daily exposure, so that the contribution of grids at different latitudes to the exposure has a clear physical dimension, thereby improving the physical rationality of disaster exposure assessment results;
[0028] 4. The core process of this invention relies only on heat wave spatial mask, temperature threshold information, mesh geometry information, and object spatial relationships, without relying on specific fitting parameters, and has portability across regions, datasets, and multiple business scenarios. Attached Figure Description
[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0030] Figure 1 A schematic diagram of the process for tracking and evaluating the evolution of high-temperature heat waves based on a topology-aware graph spectral framework, provided in an embodiment of the present invention;
[0031] Figure 2 Figures (a) to (d) are schematic diagrams of the four typical evolutionary scenarios of propagation, splitting, merging, and splitting-merging provided in the embodiments of the present invention, and their object graph connection structures.
[0032] Figure 3 Figures (a) to (f) are schematic diagrams of deterministic stress testing provided in the embodiments of the present invention, including schematic diagrams of test scenarios such as slow expansion, time fluctuation, adjacent independent objects, boundary disturbance, translation with spatial overlap and gradual decay;
[0033] Figure 4 The number and proportion of events of different topological evolution types in the main spectrum of high temperature heat waves in the Yangtze River Basin under the reference parameter configuration provided in the embodiments of the present invention;
[0034] Figure 5 Figures (a) to (b) show the event area-daily exposure characteristics and typical split-merge event evolution processes under different heat wave tracking representations provided in the embodiments of the present invention. Detailed Implementation
[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0036] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0037] like Figures 1-5 As shown, this embodiment of the invention provides a method for tracking and evaluating the evolution of high-temperature heat waves based on a topology-aware graph spectral framework, including the following steps:
[0038] Step S1: Construct a high-temperature threshold exceeding the mask for the target region
[0039] For calendar day t and grid cell x, a preset percentile threshold T is calculated within a calendar window centered on the target day, based on the daily maximum 2-meter temperature during the reference climatic period. 90 [d(t),x]. Taking the 90th percentile and the 5-day central window as an example, the high temperature threshold exceedance indicator is defined as:
[0040] (1)
[0041] In the formula, This is an indicator function; it takes the value 1 if the condition within the parentheses is true, and 0 otherwise. max (t,x) represents the daily maximum 2-meter temperature of grid cell x on day t; d(t) represents the corresponding calendar day. The threshold parameters can be preset according to business or regional climate characteristics, and the 90% and 5-day windows should not be understood as the only available values in this embodiment of the invention.
[0042] The target area boundary is transformed to a coordinate reference system consistent with the meteorological data, and the threshold exceedance mask is spatially clipped before object recognition to obtain a daily high temperature threshold exceedance mask that only covers the target study area.
[0043] This step converts the raw meteorological input data into a binary high-temperature spatial mask, limits the research scope, and provides standardized input for subsequent heat wave object extraction. The threshold parameters can be adapted to the climate characteristics of different regions.
[0044] Step S2: Identify daily connected heatwave objects and construct graph nodes
[0045] For each day's target area high-temperature threshold exceeding the mask output in step S1, eight-neighbor connected components are used for labeling. Spatially connected high-temperature grids on the same day are merged into a single heat wave spatial object; connected objects with an area smaller than a preset minimum object area threshold are discarded. Graph nodes are created for each retained object, and attributes such as node identifier, date, spatial mask, number of grid cells, centroid location, and physical area are recorded. The reference implementation uses 16 grid cells as the minimum object area threshold.
[0046] This enables the extraction and attribute archiving of spatial objects related to daily heat waves, filtering out small noise patches, and transforming raster data into an object-oriented graph node data structure, providing a set of nodes for building spatiotemporal connections between objects.
[0047] Step S3: Construct candidate directed edges for adjacent days and calculate continuity scores.
[0048] The daily set of object nodes output in step S2 is used as input. For source object A on day t... i And target object A on day t+1 j A candidate relationship is only allowed if the dates of two objects differ by exactly one calendar day. Calculate separately:
[0049] (2)
[0050] (3)
[0051] (4)
[0052] In the formula, n(·) represents the number of grid cells in the set; O src Used to measure the proportion of a source object that is carried over to the target object, O dst IoU is used to measure what proportion of the target object comes from the source object, while IoU is used to measure the overall spatial consistency between the two objects.
[0053] The candidate edge retention condition is defined as follows:
[0054] (5)
[0055] In the formula, θ O θ is the directional overlap threshold. I This is the IoU threshold. The reference configuration uses θ. O =0.30、θ I =0.10, and requires sharing at least 1 mesh cell. This "Directional Overlap or IoU" condition can simultaneously accommodate object growth, decay, and shape deformation.
[0056] Calculate the continuity score for all retained candidate edges:
[0057] (6)
[0058] In the formula, w I w s and w d These are the weights for spatial IoU, source-direction overlap rate, and target-direction overlap rate, respectively. The reference configuration uses w. I =0.50、w s =w d =0.25; These coefficients are used for candidate edge sorting, not for fitting physical parameters.
[0059] This step generates all candidate connections between objects with spatial continuity, without imposing one-to-one mandatory constraints, and fully preserves all potential spatiotemporal associations between objects. It can express the main continuity relationship and also preserve the one-to-many and many-to-one connections corresponding to splitting and merging, serving as the original context for topology diagnosis.
[0060] Step S4: Determine the optimal bidirectional backbone lineage and diagnose the split-merge topology.
[0061] The set of candidate directed edges output in step S3 and their continuity scores are used as input. For each source node, the continuity score is calculated as follows: ij The optimal outgoing edge is determined in the order of "high to low, IoU high to low, and number of spatial intersection grids high to low"; the same rule is used to determine the optimal incoming edge for each target node. When the above three indicators are still completely identical, a pre-fixed candidate edge generation order is used as the final deciding rule to ensure that the results are repeatable. Only when a candidate edge is both the optimal outgoing edge of its source node and the optimal incoming edge of its target node is it marked as a backbone edge.
[0062] Starting from a node without a trunk predecessor, trace daily along the trunk successor direction to form a one-to-one directed trunk lineage. Only lineages that reach a preset minimum duration of days and whose adjacent node dates are consecutive are retained; the reference configuration is at least 3 consecutive days. The final event label cube consists only of spatial masks of the trunk lineage nodes; other candidate edges do not add event footprint members.
[0063] The complete candidate edge context is preserved around the nodes of the main lineage: main nodes with a candidate out-degree greater than 1 are denoted as split nodes, and main nodes with a candidate in-degree greater than 1 are denoted as merge nodes. This categorizes events into topology-preserving, split, merge, and split-merge types. It should be noted that split / merge is a node role diagnosis; a node can simultaneously hold both split and merge roles.
[0064] By uniquely identifying the event identity and event footprint through the bidirectional optimal backbone edge, the same heat wave event has a definite and repeatable life cycle; at the same time, the remaining candidate edges are fully preserved as topological context, and the split and merge topological roles are explicitly identified, realizing the separation of event identity definition and topological diagnosis, and solving the defects of traditional methods that are either fragmented or lose branch convergence information.
[0065] Step S5: Calculate event-scale indices and conduct topological evolution assessment.
[0066] The main lineage event footprint, split / merge node roles, and complete candidate edge topology information output from step S4 are used as input. For each retained event, the duration, maximum daily physical area, average daily physical area, grid-to-daily exposure, cumulative area-to-daily exposure, number of topology transformations, and topology complexity are calculated. Based on whether split and / or merge node roles are included, the events are categorized as topology-preserving, split, merge, and split-merge. The total number of topology transformation roles is defined as follows:
[0067] (7)
[0068] Topological complexity is defined as:
[0069] (8)
[0070] In the formula, N split and N merge N represents the number of splitting node roles and merging node roles on the main lineage, respectively. primary This represents the total number of nodes in the main phylogenetic tree. Nodes that have both splitting and merging roles are counted in both roles, so C is the normalized count of topology transformation roles, not the percentage of unique transformation nodes.
[0071] For regular latitude and longitude grids, the physical area of the grid cells can be calculated based on the grid boundary under the spherical Earth approximation:
[0072] (9)
[0073] In the formula, A x R is the physical area of grid cell x; R is the Earth's radius, approximately 6371 km for the reference implementation; φ N φ S These are the latitudes of the north and south boundaries of the grid cell, λ. E , λ W The longitudes of the eastern and western boundaries are respectively, and the angles are in radians. The cumulative area-daily exposure of event E is defined as:
[0074] (10)
[0075] In the formula, T EE is the set of dates covered by the event. t This represents the main phylogenetic physical footprint of the event on day t. This indicator is equivalent to summing the daily physical area over the event's lifetime, in km². 2 • day; it can also record auxiliary indicators such as duration, maximum daily area, average daily area, and grid-day exposure.
[0076] It enables multi-dimensional quantitative assessment of events, characterizing event duration, spatial scale, total disaster exposure, and evolutionary complexity, distinguishing between event frequency statistics and exposure statistics, and supporting comprehensive analysis of regional heat wave disasters.
[0077] The specific implementation method is as follows:
[0078] Example 1: Verification of topology recovery under controlled synthesis truth conditions
[0079] To verify the ability of this invention to identify and recover different thermal wave evolution topologies, a controlled synthetic experiment with known evolutionary truth values was constructed. For example... Figure 2 As shown, the typical heat wave evolution includes four scenarios: propagation, splitting, merging, and split-merging. Propagation indicates that adjacent date objects maintain a continuous correspondence, splitting and merging represent local one-to-many and many-to-one candidate connection relationships, respectively, and split-merging indicates that the same event spectrum lifecycle contains both splitting and merging node roles.
[0080] In controlled synthesis truth experiments, except Figure 2 In addition to the four scenarios shown, an expansion scenario is further set up to test the spectral continuity when the spatial extent of the object changes without splitting or merging. Therefore, the synthetic truth test actually includes five preset scenarios: propagation, expansion, splitting, merging, and split-merging. Each scenario generates 100 valid realizations, for a total of 500 valid realizations; among them, propagation and expansion are both classified into the topology-preserving class.
[0081] After tracing using the method of this embodiment, the correct recovery rates for four types of event paths—topology preservation, splitting, merging, and split-merging—were 100%, 97%, 100%, and 80%, respectively. The results show that this invention can recover a pre-defined primary evolutionary topology under controlled conditions with known truth values. In the synthetic experimental configuration of this embodiment, the class recovery rate of the composite split-merging structure is lower than that of the topology preservation, single splitting, and single merging structures, indicating that the composite topology involves more complex local object connections and candidate edge combination relationships.
[0082] In addition to the aforementioned synthetic truth tests, further deterministic stress tests were conducted. For example... Figure 3As shown, the tests included scenarios such as slow expansion, temporal fluctuations, adjacent independent objects, boundary perturbations, translations maintaining spatial overlap, and gradual decay. Under the set test conditions, slow expansion did not produce spurious splits, temporal fluctuations did not cause unexpected lineage interruptions, adjacent independent objects did not undergo spurious mergers, translations with boundary perturbations but still maintaining effective spatial overlap maintained lineage continuity, and gradual decay did not produce spurious branches or unexpected lineage breaks. These two types of experiments were used to evaluate the recovery capability of preset topology categories and the lineage stability under typical boundary conditions, respectively.
[0083] Example 2: Tracking and Comprehensive Assessment of the Evolution of High Temperature Heat Waves in the Yangtze River Basin
[0084] Taking the daily maximum 2-meter temperature data of the Yangtze River Basin from June to August 1979 to 2024 as an example, and using 1981 to 2010 as the reference climatic period, the 90th percentile threshold for each day was calculated using a 5-day central calendar window. The minimum object size threshold was set to 16 grid cells, the directional overlap threshold to 0.30, the IoU threshold to 0.10, and the continuity score weight w. I =0.50、w s =w d With reference parameters of 0.25 and a minimum duration of 3 days, the method of this invention identified and reconstructed a total of 654 heat wave main lineages that met the minimum duration condition.
[0085] like Figure 4 As shown, based on whether the backbone lineage includes splitting and merging node roles, among the 654 events, 277 are topology-preserving, 85 are splitting, 108 are merging, and 184 are split-merging, accounting for 42.4%, 13.0%, 16.5%, and 28.1% respectively. Therefore, there are 377 topology change events that include at least one splitting or merging node role, accounting for 57.6% of all events.
[0086] Further accumulation based on the area-daily exposure of each event reveals that topology change events involving at least one splitting or merging node contribute approximately 87.8% of the total area-daily exposure of all events, with splitting-merging events contributing approximately 66.3%. Under this dataset and reference parameter configuration, there is a significant difference between the proportion of topology change events and their cumulative spatial exposure contribution, indicating that event frequency and area-daily exposure represent different but complementary event-scale information.
[0087] Example 3: Comparison of event exposure characteristics and fragmentation impact under different tracking representations
[0088] To further illustrate the impact of differences in event tracking representations on event-scale spatial exposure and the representation of complex topological events, the graph hierarchy representation of this invention is compared with connected representations and centroid trajectory representations. For example... Figure 5 As shown in (a), the grid-to-daily exposure and physical area-to-daily exposure of the identified events by the three tracking representations all exhibit a wide event-scale distribution. The exposure distribution and total amount obtained by the graph spectral representation and the connected representation are generally similar, while the total grid-to-daily exposure and total physical area-to-daily exposure of the centroid trajectory representation are relatively low, indicating that different event tracking representations can have different effects on the cumulative exposure statistics at the event scale.
[0089] like Figure 5 As shown in (b), event 611 was further selected as a typical split-merge type heat wave event. This event lasted for 31 days, and the main lineage obtained by this invention contained 8 split node roles and 7 merge node roles, indicating that there were multiple local branching and convergence relationships within its life cycle. When tracking using the centroid trajectory comparison method, a total of 5 centroid trajectory event segments had non-zero spatiotemporal overlap with this main lineage. Using the main lineage footprint of this invention as a cross-method comparison reference, the best single centroid event segment had a spatiotemporal intersection-union ratio (ST-IoU) of 0.487; only to diagnose the fragmentation degree of the comparison method, the above 5 centroid event segments were subjected to diagnostic footprint union, and repeated grid-time units were counted only once, and the ST-IoU between them and this main lineage reached 0.975.
[0090] The above comparisons show that for complex heat wave processes containing obvious splitting and converging relationships, different event connection rules may lead to different event-level organizational results, further affecting the event continuity and the statistics of event-scale exposure. This invention uses a set of backbone edges to determine event identity and event footprints, and utilizes a complete set of candidate edges to diagnose splitting and merging topologies. Candidate edges not selected as backbone edges are used to preserve the topological context of local branching-converging relationships without increasing the event footprint members of the backbone lineage, thereby achieving separation of event identity determination and evolutionary topology diagnosis.
[0091] Based on the same idea, this invention also provides a high-temperature heat wave evolution tracking and assessment system based on a topology-aware graph spectral framework, including:
[0092] Data Input and Mask Construction Module: Used to input daily maximum temperature raster data of the target study area, calculate daily high temperature threshold, construct a high temperature threshold transcending mask, and apply a spatial mask of the target area to complete the clipping;
[0093] Heatwave object recognition module: used to perform eight-neighbor connected component labeling on the high temperature threshold exceeding the mask after clipping, identify daily heatwave spatial objects, remove objects smaller than the preset object size threshold, construct graph nodes for the retained heatwave spatial objects and record object attributes;
[0094] Candidate edge construction and scoring module: This module is used to construct candidate directed edges only between heat wave spatial objects corresponding to two strictly adjacent calendar days, calculate the source overlap rate, target overlap rate, and spatial intersection-union ratio, filter and obtain retained candidate edges according to preset retention conditions, and calculate the continuity score of the retained candidate edges based on the source overlap rate, target overlap rate, and spatial intersection-union ratio.
[0095] The backbone spectrum reconstruction and topology diagnosis module is used to deterministically sort the candidate outgoing edges of each source node and the candidate incoming edges of each target node based on their continuity scores; it defines the candidate edges that are simultaneously the optimal outgoing edges of the source node and the optimal incoming edges of the target node as backbone edges, constructs a one-to-one directed backbone spectrum along the backbone edges, and defines the event footprint using the backbone spectrum; and performs topology diagnosis based on the in-degree and out-degree of all candidate edges to identify split nodes and merged nodes.
[0096] Event assessment and output module: Based on the event footprints corresponding to the main spectrum, it calculates event assessment indicators, classifies heat wave event evolution types according to topological diagnosis results, and completes high-temperature heat wave evolution tracking and assessment output.
[0097] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the present invention to the necessity of employing the aforementioned specific details.
[0098] The block diagrams of devices, apparatuses, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.
[0099] It should also be noted that in the apparatus, device, and method of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of the present invention.
[0100] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.
[0101] It should be understood that the qualifying terms "first", "second", "third", "fourth", "fifth" and "sixth" used in the description of the embodiments of the present invention are only used to more clearly illustrate the technical solutions and are not intended to limit the scope of protection of the present invention.
[0102] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.
Claims
1. A method for tracking and assessing the evolution of high-temperature heat waves based on a topology-sensing graph phylogenetic framework, characterized in that, Includes the following steps: S1. Input the daily maximum temperature raster data of the target study area, calculate the daily high temperature threshold, construct a high temperature threshold transcending mask, and apply a target area spatial mask to complete the clipping. S2. Perform eight-neighbor connected component labeling on the high temperature threshold exceeding the mask after clipping, identify daily heat wave spatial objects, remove objects smaller than the preset object size threshold, construct graph nodes for the retained heat wave spatial objects and record object attributes. S3. Construct candidate directed edges only between heat wave spatial objects corresponding to two strictly adjacent calendar days, calculate source overlap rate, target overlap rate, and spatial intersection-union ratio, filter and obtain retained candidate edges according to preset retention conditions, and calculate the continuity score of retained candidate edges based on source overlap rate, target overlap rate, and spatial intersection-union ratio. S4. Determinely sort the candidate outgoing edges of each source node and the candidate incoming edges of each target node according to their continuity scores; define the candidate edge that is both the optimal outgoing edge of the source node and the optimal incoming edge of the target node as the backbone edge, construct a one-to-one directed backbone spectrum along the backbone edge, and define the event footprint with the backbone spectrum; perform topology diagnosis and identification of split nodes and merged nodes based on the in-degree and out-degree of all candidate edges. S5. Based on the event footprints corresponding to the main spectrum, calculate the event evaluation index, classify the heat wave event evolution type according to the topological diagnosis results, and complete the high temperature heat wave evolution tracking and evaluation output.
2. The method according to claim 1, characterized in that, In step S1, the daily high temperature threshold is the temperature percentile threshold under the calendar window centered on the target day within the reference climate period; the high temperature threshold exceeding the mask is generated by the indicator function judging that the highest temperature of the grid cell exceeds the temperature percentile threshold.
3. The method according to claim 1, characterized in that: In step S2, the recorded object attributes include node identifier, date, spatial mask, centroid, number of mesh cells, and physical area.
4. The method according to claim 1, characterized in that, In step S3, the candidate directed edge retention condition is: the two objects have at least one shared grid cell and satisfy the maximum directional overlap rate is greater than or equal to the directional overlap threshold, or the spatial intersection-union ratio is greater than or equal to the IoU threshold; the continuity score is obtained by multiplying the spatial intersection-union ratio, source direction overlap rate, and target direction overlap rate by their respective weights and summing them, and the sum of each weight is 1.
5. The method according to claim 1, characterized in that, In step S4, the deterministic sorting rule for candidate edges prioritizes continuity score, followed by spatial intersection-union ratio, and then the number of spatial intersection grid cells; when all multiple indicators are the same, the order in which candidate edges are generated is used as the deciding rule; when screening the main lineage, the main lineage that meets the preset minimum duration and the node dates are continuous is retained. A trunk node with a candidate out-degree greater than 1 is determined to be a split node, and a trunk node with a candidate in-degree greater than 1 is determined to be a merge node.
6. The method according to claim 1, characterized in that, In step S5, the event evaluation metrics include duration, maximum physical area, cumulative area-daily exposure, topology transformation number, and topology complexity; the heatwave event evolution types are divided into four categories: topology preservation, splitting, merging, and split-merging; the topology transformation number is the sum of the number of splitting node roles and the number of merging node roles; the topology complexity is the topology transformation number divided by the total number of mainline nodes; and the cumulative area-daily exposure is the sum of the physical areas of the event footprint grid cells on each date within the event's lifecycle.
7. A system for tracking and assessing the evolution of high-temperature heat waves based on a topology-sensing graph phylogenetic framework, characterized in that, include: Data Input and Mask Construction Module: Used to input daily maximum temperature raster data of the target study area, calculate daily high temperature threshold, construct a high temperature threshold transcending mask, and apply a spatial mask of the target area to complete the clipping; Heatwave object recognition module: used to perform eight-neighbor connected component labeling on the high temperature threshold exceeding the mask after clipping, identify daily heatwave spatial objects, remove objects smaller than the preset object size threshold, construct graph nodes for the retained heatwave spatial objects and record object attributes; Candidate edge construction and scoring module: This module is used to construct candidate directed edges only between heat wave spatial objects corresponding to two strictly adjacent calendar days, calculate the source overlap rate, target overlap rate, and spatial intersection-union ratio, filter and obtain retained candidate edges according to preset retention conditions, and calculate the continuity score of the retained candidate edges based on the source overlap rate, target overlap rate, and spatial intersection-union ratio. The backbone spectrum reconstruction and topology diagnosis module is used to deterministically sort the candidate outgoing edges of each source node and the candidate incoming edges of each target node based on their continuity scores; it defines the candidate edges that are simultaneously the optimal outgoing edges of the source node and the optimal incoming edges of the target node as backbone edges, constructs a one-to-one directed backbone spectrum along the backbone edges, and defines the event footprint using the backbone spectrum; and performs topology diagnosis based on the in-degree and out-degree of all candidate edges to identify split nodes and merged nodes. Event assessment and output module: Based on the event footprints corresponding to the main spectrum, it calculates event assessment indicators, classifies heat wave event evolution types according to topological diagnosis results, and completes high-temperature heat wave evolution tracking and assessment output.
