A real-time interactive extreme climate disaster event correlation mining method

Abstract

The application discloses a real-time interactive extreme climate disaster event correlation mining method, which is from the perspective of knowledge aggregation, integrates the spatio-temporal data cube technology into the interactive mining process of the spatio-temporal correlation mode, and breaks through the key bottleneck that the complex mining constraints and the real-time interactive performance cannot be considered in the human-computer interaction. The method breaks through the limitation of the traditional Cube dimension orthogonal independence, unifies the three types of complex constraints, i.e. the spatio-temporal boundary randomness, the spatio-temporal coupling and the multi-dimensional rules, into the multi-dimensional query view in the data cube, thereby supporting the view control ability of the analyst based on the multi-dimensional and multi-scale Cube and the spatial online analytical processing operation, realizing the visualization exploration of the spatio-temporal non-stationary, multi-scale effect and spatio-temporal differentiation characteristics of the extreme climate disaster event correlation mode, overcoming the problem that the spatio-temporal correlation knowledge cannot be distributed and aggregated in theory, and establishing a multi-granularity spatio-temporal correlation mode distributed aggregation mechanism of the pre-aggregated knowledge query and the online mining cooperation.

Description

Technical Field

[0001] This invention belongs to the field of disaster prevention, specifically a real-time interactive method for mining the correlation of extreme climate disaster events. Background Technology

[0002] Against the backdrop of global warming, extreme weather events such as persistent droughts, heat waves, and wildfires are intensifying and exhibiting complex and combined correlations, leading to a synergistic amplification effect of multiple disasters. This poses significant challenges to national and regional disaster prevention and mitigation capabilities, and severely disrupts and threatens economic development and the security landscape. The inherent correlations and interactions among extreme climate disaster events urgently need to be explored, and have gradually become a new research hotspot. On the other hand, with the densification of ground observation networks, the successive launches of high spatiotemporal spectral Earth observation satellites, and the development of related geographic simulation systems, Earth data is experiencing explosive growth, far exceeding the capacity of traditional knowledge extraction. How to efficiently uncover the potential correlations and changing patterns of extreme climate disaster events from this massive amount of data is a key issue and challenge that Earth data science urgently needs to address in its work on multi-hazard coupling risk prevention and decision-making.

[0003] Because the distribution patterns of geographical events or phenomena exhibit spatiotemporal heterogeneity, scale effects, and multimodal characteristics, the exploration process often requires multiple rounds of progressive mining, observation, and analysis based on different spatiotemporal regions and rule constraints, inspired by multiple scientific hypotheses, while simultaneously integrating the analyst's domain knowledge, in order to find valuable patterns from numerous rules. This analyst-centered exploratory mining model typically needs to support arbitrary multidimensional constraints (including spatiotemporal and rule-based conditions) and places high demands on the real-time response of mining results; otherwise, it will seriously affect the continuity of the human-computer interaction process and the accuracy of the analysis. However, the above two efficiency requirements are a clear contradiction. In the context of Earth big data, this contradiction is particularly prominent and has become a major bottleneck restricting the mining of spatiotemporal correlation rules in the discovery of geoscientific correlation knowledge.

[0004] Introducing a data cube to organize multi-granular spatiotemporal association rules, transforming the exploratory mining of rule knowledge into interactive queries on the multidimensional view of the cube, and generating mining results under custom constraints in real time through multi-granular rule aggregation, is an ideal way to solve the aforementioned theoretical needs. In the field of exploratory analysis, the cube provides a powerful data engine for insightful analysis of massive multidimensional information by constructing multidimensional interactive models and multi-granular aggregation mechanisms in databases or memory. It has become a hot topic in spatiotemporal big data analysis and has been successfully used in the efficient organization and interactive analysis of Earth observation datasets. However, the existing research ideas and methodologies still have two problems: (1) The measures stored in the existing cube are mainly raw data or simple statistics, while spatiotemporal association rules are the potential knowledge contained in spatiotemporal data. The aggregation process represents the transformation of knowledge on the scale as the sample size changes. Therefore, there is a significant difference in the aggregation mechanism compared with traditional measures. (2) The exploratory mining of spatiotemporal association rules involves complex constraints. Besides the arbitrariness of spatiotemporal boundaries, spatiotemporal constraints often exhibit coupling (i.e., the time domain varies with spatial distribution, such as geographical differences in phenological periods or rainy seasons); furthermore, the diversity of state attributes of geographical phenomena leads to a significant increase in rule constraints. These complexities make modeling and representing the dimensions of Cube analysis difficult and will also significantly affect query performance. Current Cube-related research still has significant limitations in its applicability to the above problems and urgently needs breakthroughs. Summary of the Invention

[0005] This invention provides a real-time interactive method for mining associations of extreme climate disaster events, which addresses the problems in existing technologies where spatiotemporal constraints are often coupled, and the diversification of geographical phenomenon state attributes leads to a significant increase in rule constraints. These complexities make it difficult to model and express the dimensions of Cube analysis, and will also significantly affect query performance.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A real-time interactive method for mining associations of extreme climate disaster events, comprising:

[0008] Constructing multi-scale spatiotemporal units;

[0009] Based on global standard data, sequence extraction is performed on extreme climate disaster events to obtain extreme climate disaster event sequences;

[0010] Correlation rule metrics are obtained based on multi-scale spatiotemporal units and extreme climate disaster event sequences;

[0011] A spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics.

[0012] Extreme climate disaster events are mined using a spatiotemporal correlation cube model.

[0013] Preferably, a multi-scale spatiotemporal unit is constructed using the global discrete grid Google S2 as the spatial pyramid level, and the global spatiotemporal is divided into multiple levels using a natural time granularity hierarchical structure to form a multi-scale spatiotemporal unit.

[0014] Preferably, association rule metrics are obtained based on multi-scale spatiotemporal units and extreme climate disaster event sequences. The support count of frequent patterns of extreme climate disaster events occurring within multi-scale spatiotemporal units is used as the basic metric, and then the support, confidence, and lift of the pattern are dynamically calculated as derived metrics.

[0015] Preferably, the support, confidence, and lift are calculated as follows:

[0016]

[0017] Where sup represents support, conf represents confidence, lift represents lift, supct represents the support count of frequent patterns of extreme climate disaster events within a multi-scale spatiotemporal unit, and d represents the support of the pattern. x As the antecedent of the rule, d y For the rule consequent, N is the total number of spatiotemporal transactions within a multi-scale spatiotemporal unit.

[0018] Preferably, the spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics as follows:

[0019] First, the basic spatial and temporal dimensions of the multi-scale spatiotemporal unit are divided into a pyramid layer and a custom query layer. Then, based on the coupling relationship between the pyramid layer and the custom query layer, a spatiotemporal coupling layer is established, and the association rule metric within the spatiotemporal unit is calculated. Geographic association dimension is established based on the categories and levels of extreme climate disaster events. Finally, a spatiotemporal association cube model is established based on the spatial dimension, temporal dimension, spatiotemporal coupling layer, geographic association dimension, and extreme climate disaster event sequence.

[0020] Preferably, the pyramid layer is composed of multi-level standard granularity. Based on the spatial pyramid layer constructed by S2 grid, the minimum non-intersection of tree nodes covering the query domain is obtained by using the quadtree intersection algorithm. Based on the spatiotemporal topological relationship determination, any spatiotemporal query domain can be adaptively described by the pyramid layer as a custom query layer with multi-level spatiotemporal granularity.

[0021] Preferably, the mining of extreme climate disaster events based on the spatiotemporal correlation cube model specifically involves extending the OLAP operations of block division, roll-up / drill-down, and rotation on the basis of the spatiotemporal correlation cube model to describe the mining functional connotations in the spatiotemporal and geographic correlation dimensions.

[0022] A real-time interactive system for mining the temporal correlations of extreme climate disasters, comprising:

[0023] Spatiotemporal partitioning module: used to construct multi-scale spatiotemporal units;

[0024] Event Extraction Module: Used to extract sequences of extreme climate disaster events based on global standard data, and obtain extreme climate disaster event sequences;

[0025] Rule calculation module: used to obtain association rule metrics based on multi-scale spatiotemporal units and extreme climate disaster event sequences;

[0026] Building module: Used to establish a spatiotemporal correlation cube model based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics;

[0027] Knowledge mining module: Mining extreme climate disaster events based on the spatiotemporal correlation cube model.

[0028] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of a real-time interactive method for mining correlations of extreme climate disaster events.

[0029] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a real-time interactive method for mining correlations of extreme climate disaster events.

[0030] Compared with existing technologies, this invention has the following advantages: This invention provides a real-time interactive method for mining associations of extreme climate disaster events. From a knowledge aggregation perspective, this invention integrates spatiotemporal data cube technology into the interactive mining process of spatiotemporal association patterns, overcoming the key bottleneck of the inability to simultaneously balance complex mining constraints and real-time interactive performance in human-computer interaction. This method breaks through the traditional limitation of orthogonal and independent cube dimensions, unifying the three complex constraints in the spatiotemporal association mining process—arbitrariness of spatiotemporal boundaries, spatiotemporal coupling, and multidimensional rules—into a multidimensional query view within the data cube. This supports analysts' ability to control views and perform online spatial analysis based on the cube's multidimensional and multi-scale capabilities, enabling the visual exploration of the spatiotemporal non-stationarity, multi-scale effects, and spatiotemporal differentiation characteristics of extreme climate disaster event association patterns.

[0031] Furthermore, this invention establishes a multi-granularity spatiotemporal association pattern distribution and aggregation mechanism for real-time interactive mining. This method overcomes the theoretical problem of the non-distributable aggregation of spatiotemporal association knowledge by establishing a multi-granularity spatiotemporal association pattern distribution and aggregation mechanism that coordinates pre-aggregated knowledge query and online mining. This maximizes the reusability of existing knowledge and achieves scale transformation from local spatiotemporal association patterns to global spatiotemporal association patterns. By avoiding repeated mining of historical regions, it ensures a real-time and robust response during the exploration and interaction process. This method provides a new approach for the efficient mining of other spatiotemporal data under limited computing resources. Attached Figure Description

[0032] Figure 1 This is a flowchart of a real-time interactive method for mining correlations of extreme climate disaster events according to the present invention;

[0033] Figure 2 This is a block diagram of an extreme climate disaster event mining system for real-time interactive applications according to the present invention.

[0034] Figure 3 A schematic diagram illustrating the spatiotemporal relationships of events within a spatiotemporal neighborhood;

[0035] Figure 4 A schematic diagram of Cube modeling for spatiotemporal correlation rule mining;

[0036] Figure 5 This expands the scope of Cube's OLAP operation association mining functionality;

[0037] Figure 6 To construct multi-granularity spatiotemporal related knowledge distribution aggregation paths supported by knowledge aggregation trees, where a is the spatiotemporal index tree, b is the knowledge aggregation tree, and c is the distribution aggregation path that combines pre-aggregation query and online mining.

[0038] Figure 7 This is the system implementation technical architecture in the embodiment;

[0039] Figure 8 This is a schematic diagram for exploring multi-view linkage of geographic time and spatiotemporal correlation modes. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0041] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0042] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0043] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0044] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0045] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0046] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0047] like Figure 1 As shown, this invention provides a real-time interactive method for mining correlations of extreme climate disaster events, including:

[0048] S101 constructs multi-scale spatiotemporal units;

[0049] S102 extracts sequences of extreme climate disaster events based on global standard data to obtain extreme climate disaster event sequences;

[0050] S103 obtains association rule metrics based on multi-scale spatiotemporal units and extreme climate disaster event sequences;

[0051] S104 establishes a spatiotemporal correlation cube model based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics;

[0052] S105 uses a spatiotemporal correlation cube model to mine extreme climate disaster events.

[0053] Multi-scale spatiotemporal units are constructed using the global discrete grid Google S2 as the spatial pyramid level, and the global spatiotemporal space is divided into multiple levels using a natural time granularity hierarchical structure to form multi-scale spatiotemporal units.

[0054] Based on multi-scale spatiotemporal units and extreme climate disaster event sequences, association rule metrics are obtained. The support count of frequent patterns of extreme climate disaster events occurring within multi-scale spatiotemporal units is used as the basic metric, and then the support, confidence, and lift of the pattern are dynamically calculated as derived metrics.

[0055] The support, confidence, and lift are calculated as follows:

[0056]

[0057] Where sup represents support, conf represents confidence, lift represents lift, supct represents the support count of frequent patterns of extreme climate disaster events within a multi-scale spatiotemporal unit, and d represents the support of the pattern. x As the antecedent of the rule, d y For the rule consequent, N is the total number of spatiotemporal transactions within a multi-scale spatiotemporal unit.

[0058] The spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics as follows:

[0059] First, the basic spatial and temporal dimensions of the multi-scale spatiotemporal units are divided into a pyramid layer and a custom query layer. Then, based on the coupling relationship between the pyramid layer and the custom query layer, a spatiotemporal coupling layer is established. The association rule metric within the spatiotemporal unit is calculated. A geographic association dimension is established based on the categories and classifications of extreme climate disaster events. Finally, a spatiotemporal association cube model is constructed based on the spatial dimension, temporal dimension, spatiotemporal coupling layer, geographic association dimension, and the sequence of extreme climate disaster events.

[0060] The pyramid layer consists of multiple levels of standard granularity. It is a spatial pyramid layer constructed based on S2 grid. The quadtree intersection algorithm is used to obtain the minimum non-intersection of tree nodes covering the query domain. Based on the spatiotemporal topological relationship determination, any spatiotemporal query domain can be adaptively described by the pyramid layer as a custom query layer with multiple levels of spatiotemporal granularity.

[0061] The specific method for mining extreme climate disaster events based on the spatiotemporal correlation cube model is to expand the OLAP operations of extended segmentation, roll-up / drill-down, and rotation on the basis of the spatiotemporal correlation cube model, thereby enhancing the mining capabilities in the spatiotemporal and geographic correlation dimensions.

[0062] like Figure 2 As shown, the present invention also provides a real-time interactive system for mining correlations of extreme climate disaster events, including:

[0063] Spatiotemporal partitioning module: used to construct multi-scale spatiotemporal units;

[0064] Event Extraction Module: Used to extract sequences of extreme climate disaster events based on global standard data, and obtain extreme climate disaster event sequences;

[0065] Rule calculation module: used to obtain association rule metrics based on multi-scale spatiotemporal units and extreme climate disaster event sequences;

[0066] Model building module: used to build a spatiotemporal correlation cube model based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics;

[0067] Knowledge mining module: Mining extreme climate disaster events based on the spatiotemporal correlation cube model.

[0068] Specifically, the present invention is implemented using the following technical solutions:

[0069] Constructing a spatiotemporal correlation cube of drought and wildfire events for real-time interactive mining includes the following steps:

[0070] 1) Multi-level spatiotemporal partitioning: Using the global discrete grid Google S2 as the spatial pyramid level, global spatiotemporal is partitioned into multiple levels using a natural time granularity hierarchy (year-month-day) to form multi-scale spatiotemporal units. If the spatial and temporal dimension levels are |SH| and |TH| respectively, then |SH|×|TH| different combinations of cubic cuboids will be established, forming a pyramid structure that aggregates layer by layer. The unique cubic spatiotemporal unit [T][S] is determined by the two dimension values ​​of the time code T and the spatial code S.

[0071] 2) Extreme Climate Disaster Event Sequence Extraction: For each spatiotemporal range, based on unified global precipitation grid data, the standardized precipitation index (SPI) is calculated, and the spatiotemporal distribution of hydrological drought is extracted and classified according to thresholds. Based on unified global temperature grid data, surface apparent temperature is calculated, and heat wave events are extracted based on a consecutive days threshold condition, and the frequency and duration of heat waves are statistically analyzed. Based on remote sensing fire point data, wildfire event sequences are extracted based on a consecutive days threshold condition.

[0072] 3) Association Rule Measurement Definition: Based on the above spatiotemporal multi-granularity partitioning, a measurement structure is established within each cuboid. The support count (supct) of frequent patterns of geographical events occurring within the cuboid is used as the basic metric. Then, the support (sup) of this pattern, along with the corresponding confidence (conf) and lift (lift) are dynamically calculated as derived metrics, as shown in Equation 1, where N is the total number of spatiotemporal transactions within the cuboid, and d... x As the antecedent of the rule, d y This is a consequent of the rule. This study only stores the results of association rule mining in the Cube. Therefore, to maintain the universality of this invention, it does not limit the geographic event modeling method or the specific mining method of spatiotemporal association rules.

[0073]

[0074] The co-occurrence pattern of events is determined by temporal overlap. Considering the spatiotemporal influence domain of event associations, a spatiotemporal buffer is introduced to enhance the ability to discover spatiotemporal dependency relationships. Furthermore, spatiotemporal overlap is used to distinguish between temporal cascades (e.g., drought-wildfire) and spatiotemporal co-locations (e.g., drought-heatwave). The three main cases shown in the figure are considered, and the spatiotemporal association rules (support) of events within a region are defined, as shown in Equation 2. Here, buf is the event spatiotemporal buffer function, M is the number of Moore neighborhood units, L is the time delay, and N is the total duration of the sequence.

[0075]

[0076] 4) Complex Mining Constraint Expression: The aforementioned spatiotemporal division forms the basic spatial and temporal dimensions of the spatiotemporal cube. The hierarchical structure of the dimensions is divided into a pyramid layer D.base and a custom query layer D.def. D.base can be composed of multiple levels of standard granularity: θ 0 <θ 1 …θ m Spatial pyramid hierarchy D constructed based on S2 grid s The .base method can be used to obtain the minimum non-intersection of tree nodes covering the query domain using the quadtree intersection algorithm.

[0077] Based on spatiotemporal topological relationship determination, any spatiotemporal query domain can be adaptively described with multi-level spatiotemporal granularity.

[0078] D.base→D.def.

[0079] To address the spatiotemporal coupling constraints, a time granularity function θ, which varies with the spatial dimension, is defined based on practical application requirements. t =f(d s This describes the coupling relationship between the time dimension and the space dimension hierarchy, and then establishes a hierarchy based on <θ. s ,θ t >This represents the spatiotemporal coupling hierarchy at the basic granularity of combination.

[0080] To address multidimensional rule constraints, a geographic association dimension D is introduced. a As a member of the rule constraint set, each geographic association dimension represents a certain state (such as wildfire duration) or attribute (such as drought level) of a geographic event (phenomenon), or the spatiotemporal relationship between geographic events (such as the distance between the fire point and the road).

[0081] Ultimately, by combining different granularities across all dimensions, the Cube is divided into multi-granularity cubic units (cuboids), allowing any complex constraint to be uniformly expressed through a combination of query domains and query granularities across three dimensions. The function is the query domain of the dimension, and θ is the query granularity of the dimension.

[0082]

[0083] 5) Definition of Association Mining Operations: Building upon this foundation, the mining capabilities of OLAP operations such as Dice, Roll-Up / Drill-Down, and Pivot are expanded to encompass spatiotemporal and geographic association dimensions. This allows analysts to execute spatiotemporal association rule mining commands in any region of the Cube using interactive OLAP queries, such as... Figure 4 As shown.

[0084] 6) Association metric distribution aggregation mechanism: For the aggregation process of spatiotemporal association rules with multiple granularities, a rule knowledge aggregation tree that is the reverse of the spatiotemporal index tree is dynamically constructed in memory according to the index path of the spatiotemporal query domain in the Cube. Each layer on the tree represents a spatiotemporal granularity in the aggregation path.

[0085] Supported by a rule-based knowledge aggregation tree, a distributed aggregation method for multi-level frequent pattern recursive decomposition is established. The basic idea of ​​the single-level frequent pattern decomposition algorithm is as follows: assuming an N-layer spatiotemporal cuboid C is divided into two cuboids, C1 and C2, at the N-1 layer, then C = C1 ∪ C2 and... Let the frequent patterns implied by the three be G = Fp(C), E = Fp(C1), and F = Fp(C2), respectively, and let there exist relations. Here, E∪F consists of the intersection E∩F and the symmetric difference E△F. In E△F, only itemsets with a support count less than the minimum support count (min_supct) require online mining. Therefore, G can be decomposed into a distributable aggregation part and a mining part, as shown in Equation 4.

[0086]

[0087] The specific steps are as follows: For each level, recursively execute the single-level frequent pattern decomposition algorithm downwards and the rule aggregation process upwards until the query granularity condition is met. The specific route is as follows: Figure 6 The process is as follows: ① Merge all frequent pattern trees in the lower layer corresponding to the cuboid at this layer to obtain the union of frequent patterns. ② Decompose frequent patterns to obtain candidate itemsets for mining, and prune the supersets of infrequent itemsets using prior knowledge. ③ Based on the candidate itemsets, perform recursive mining of lower-level rules. ④ Summarize the mining results and pre-aggregation rules, and update the support count using Equation 6. ⑤ Determine and update frequent patterns, and proceed to the next layer of mining. To reduce the volatility of online mining latency caused by spatiotemporal analysis, it is proposed to store the transaction mining table corresponding to cuboids in a relational database or a KV (Key-Value) database to optimize collaborative aggregation efficiency.

[0088] 7) Aggregated query collaboration mechanism: Based on existing mining results, select key geographic event distribution patterns for the analysis topic. <e1,e2,…,e n (e.g., Wildfire, Drought) The spatiotemporal hotspot distribution of this pattern in the Cube is detected using the Getis-Ord Gi* spatiotemporal statistical method, as shown in Equation 5.

[0089]

[0090] Where n represents the number of all basic cuboids, supct j w represents the support count of the frequent pattern corresponding to the j-th cuboid in the sliding spatiotemporal neighborhood. i,j This is the weight matrix. Its Z-score and p-value can be used to identify hot and cold areas as well as non-statistically significant areas. Furthermore, since recent cuboids have a higher access frequency than historical cuboids, and the accumulation of historical cuboids can lead to an infinite increase in memory size, a Gaussian time decay function is added as a weight during the hot and cold point detection process to achieve a decay process of the hot and cold point distribution in the time dimension, ensuring that historical cuboids have a greater migration probability (Equation 6), where μ is the decay start time.

[0091]

[0092] Secondly, based on the identification of hotspot areas, and using a migration threshold selection, the cuboids that need to be migrated from the database are determined. Based on the |Z| values ​​from hotspot analysis, the cuboid set is divided using an equal-quantity hierarchical method. Combined with the estimated cuboid data volume from stratified random sampling, the total cuboid data volume corresponding to different levels is then statistically analyzed. Finally, the maximum |Z| value that satisfies the predetermined memory capacity is determined as the migration threshold. As the total data volume changes, the migration threshold is dynamically updated.

[0093] Advantages of this invention:

[0094] The construction of human-computer interaction mining models is a core method and technical challenge in supporting the scientific discovery of correlation patterns in extreme climate disaster events using Earth big data. This invention, from a knowledge aggregation perspective, integrates spatiotemporal data cube technology into the interactive mining process of spatiotemporal correlation patterns, overcoming the key bottleneck of balancing complex mining constraints and real-time interactive performance in human-computer interaction. This method breaks through the traditional limitation of orthogonal and independent cube dimensions, unifying the three complex constraints in the spatiotemporal correlation mining process—arbitrariness of spatiotemporal boundaries, spatiotemporal coupling, and multidimensional rules—into a multidimensional query view within the data cube. This supports analysts' ability to control views and perform online spatial analysis based on the cube's multidimensional and multi-scale capabilities, enabling the visual exploration of the spatiotemporal non-stationarity, multi-scale effects, and spatiotemporal differentiation characteristics of correlation patterns in extreme climate disaster events.

[0095] This invention establishes a multi-granularity spatiotemporal association pattern distribution and aggregation mechanism for real-time interactive mining. This method overcomes the theoretical problem of the non-distributable aggregation of spatiotemporal association knowledge by establishing a multi-granularity spatiotemporal association pattern distribution and aggregation mechanism that coordinates pre-aggregated knowledge query and online mining. This maximizes the reusability of existing knowledge and achieves scale transformation from local spatiotemporal association patterns to global spatiotemporal association patterns. By avoiding repeated mining of historical regions, it ensures a real-time and robust response during the exploration and interaction process. This method provides a new approach for the efficient mining of other spatiotemporal data under limited computing resources.

[0096] Example

[0097] Cube instance construction:

[0098] Experimental data include global remote sensing observation products related to the application theme since 2000, as well as downscaled ground observation gridded products, as detailed in Table 1. Based on rainfall and evapotranspiration, the standardized precipitation index (SPI) was calculated, and drought events and their levels were extracted. Heatwave events were extracted and classified from time-series temperature data using the heatwaveR package, with a grid resolution of 4 km × day. In addition, classification indicators such as vegetation water shortage index, phenological period, road network density, DEM, and population density were calculated and extracted. A corresponding Cube instance structure was designed, with a temporal dimension of (month, year, 5 years) and a spatial grid dimension of (40 km, 80 km, 160 km). A spatiotemporal coupling layer based on vegetation phenology and a geographic correlation dimension (including drought level, heatwave intensity, wildfire duration, burned area, wind speed, vegetation water shortage, population density, geographic accessibility, and topographic factors) were constructed. Using a basic spatiotemporal granularity of 40km×month as the mining unit, Stars (a spatiotemporal analysis R package) is used to obtain climate events / variables associated with the spatiotemporal location of each fire event. A transaction mining table is established in MongoDB, and association rules are mined based on the ECLAT algorithm. Finally, following the technical solution and roadmap of this invention, a Cube instance is constructed.

[0099] Table 1 Experimental Data List

[0100]

[0101] System Development and Architecture:

[0102] A memory-based DataCube engine (including query and aggregation modules) was built using Python, and SOLAP functionality was encapsulated as a Web Service within the Django framework. A visual, interactive front-end for DataCube was developed based on the open-source 3D GIS system Cesium, with local map and data services provided by the GIS service engine GeoServer. Communication between the front-end and back-end uses JSON / GeoJSON data format, and a database for storing aggregated DataCube information is built on MongoDB. Based on the empirical needs of this project, NetCDF was adopted as the storage and analysis format for the original remote sensing raster dataset. The DataCube engine accesses MongoDB via PyMongo, and a spatiotemporal association rule mining function was designed using the scientific computing package Xarray. The system architecture is as follows. Figure 7 As shown.

[0103] Interactive query mode

[0104] Interactive Query of Spatiotemporal Association Patterns: The selection of query domain and query granularity is achieved by combining the Map view, parallel coordinate plot, and timeline. Analysts can draw the spatial query domain in the Map view, select the time domain through the time slider, and preset the time and space granularity through radio buttons. The spatiotemporal coupling constraints are set through layers or functions in the drop-down menu. The parallel coordinate plot is designed as a visual query component for association rules, where each axis is defined as a geographical association dimension, and the optional scales are defined as dimension members. Therefore, the rule constraint conditions can be set through the axes, scales, and rule antecedent and consequent options.

[0105] Multi-view Linkage Exploration of Spatiotemporal Association Patterns: After the mining results are returned, all associated or frequent patterns within the spatiotemporal region are real-time fed back to the parallel coordinate plot in the form of connection lines. The chunking (Dice) operation on existing knowledge can be achieved through dimension filtering on the Map view, timeline, and parallel coordinates. The rotation operation (Pivot) can be switched based on the rule antecedent and consequent options on the parallel axes, while the roll-up (Roll-Up) and drill-down (Drill-Down) operations are triggered by selecting the analysis dimension and scale operation buttons of interest in the panel query analysis component. After selecting the connection line of the association pattern to be explored in the parallel coordinate plot, the Map view will be synchronously refreshed to display the three-dimensional discrete cubic lattice distribution of the selected support or confidence in space and time. In the analysis component, the spatiotemporal hotspots and trend distributions of the association patterns in this view are obtained by integrating the Mann-Kendall trend analysis and Getis-Ord Gi spatiotemporal statistical methods, and the evolution types and spatial distributions of the association patterns are generated by integrating the time series clustering method. The distribution of association patterns in the spatial view is obtained through time dimension aggregation, and the hotspot distributions at different spatial scales of this pattern can be compared based on the roll-up and drill-down operations; the distribution of association patterns in the time view is obtained through spatial dimension aggregation, and the sequence trends and periodic characteristics at different time scales can be compared based on the roll-up and drill-down; the environmental factor is added to the rule through the drill-down operation to explore the differentiation characteristics of the association pattern distribution in different environmental factor partitions. A linkage update mechanism is planned to be established between multiple views to assist in the connection of multi-scale patterns and multi-perspective comparison during the interactive analysis process. As Figure 8 shown.

[0106] The terminal device provided by an embodiment of the present invention. The terminal device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned device embodiments are implemented.

[0107] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention.

[0108] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0109] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0110] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0111] If the modules / units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0112] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art, guided by the specification, can make many other modifications without departing from the scope of the claims of the present invention, and all of these modifications are within the scope of protection of the present invention.

Claims

1. A real-time interactive method for mining correlations of extreme climate disaster events, characterized in that, include: Constructing multi-scale spatiotemporal units; Based on global standard data, sequence extraction is performed on extreme climate disaster events to obtain extreme climate disaster event sequences; Correlation rule metrics are obtained based on multi-scale spatiotemporal units and extreme climate disaster event sequences; A spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics. Extreme climate disaster events are mined using a spatiotemporal correlation cube model; Based on multi-scale spatiotemporal units and extreme climate disaster event sequences, association rule metrics are obtained. The support count of frequent patterns of extreme climate disaster events occurring within multi-scale spatiotemporal units is used as the basic metric, and then the support, confidence, and lift of the pattern are dynamically calculated as derived metrics. The support, confidence, and lift are calculated as follows: in For support, For confidence level, To increase the degree, Support counts for frequent patterns of extreme climate disaster events within multi-scale spatiotemporal units. For the antecedent of the rule, For the rule consequent, where N is the total number of spatiotemporal transactions within a multi-scale spatiotemporal unit; The spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics as follows: First, the basic spatial and temporal dimensions of the multi-scale spatiotemporal unit are divided into a pyramid layer and a custom query layer. Then, based on the coupling relationship between the pyramid layer and the custom query layer, a spatiotemporal coupling layer is established, and the association rule metric within the spatiotemporal unit is calculated. Geographic association dimension is established based on the categories and levels of extreme climate disaster events. Finally, a spatiotemporal association cube model is established based on the spatial dimension, temporal dimension, spatiotemporal coupling layer, geographic association dimension, and extreme climate disaster event sequence.

2. The real-time interactive method for mining correlations of extreme climate disaster events according to claim 1, characterized in that, Multi-scale spatiotemporal units are constructed using the global discrete grid Google S2 as the spatial pyramid level, and the global spatiotemporal space is divided into multiple levels using a natural time granularity hierarchical structure to form multi-scale spatiotemporal units.

3. The method for real-time interactive correlation mining of extreme climate disaster events according to claim 1, characterized in that, The pyramid layer consists of multiple levels of standard granularity. Based on the S2 grid, the spatial pyramid layer is constructed. The quadtree intersection algorithm is used to obtain the minimum non-intersection of tree nodes covering the query domain. Based on the spatiotemporal topological relationship determination, any spatiotemporal query domain can be adaptively described by the pyramid layer as a custom query layer with multiple levels of spatiotemporal granularity.

4. The real-time interactive method for mining correlations of extreme climate disaster events according to claim 1, characterized in that, Specifically, the mining of extreme climate disaster events based on the spatiotemporal correlation cube model involves extending the OLAP operations of block segmentation, roll-up / drill-down, and rotation to describe the mining functionalities in the spatiotemporal and geographic correlation dimensions.

5. A system for mining extreme climate disaster events in real-time interactive mode, characterized in that, include: Spatiotemporal partitioning module: used to construct multi-scale spatiotemporal units; Event Extraction Module: Used to extract sequences of extreme climate disaster events based on global standard data, and obtain extreme climate disaster event sequences; Rule calculation module: used to obtain association rule metrics based on multi-scale spatiotemporal units and extreme climate disaster event sequences; Building module: Used to establish a spatiotemporal correlation cube model based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics; Knowledge mining module: Mining extreme climate disaster events based on the spatiotemporal correlation cube model; Based on multi-scale spatiotemporal units and extreme climate disaster event sequences, association rule metrics are obtained. The support count of frequent patterns of extreme climate disaster events occurring within multi-scale spatiotemporal units is used as the basic metric, and then the support, confidence, and lift of the pattern are dynamically calculated as derived metrics. The support, confidence, and lift are calculated as follows: in For support, For confidence level, To increase the degree, Support counts for frequent patterns of extreme climate disaster events within multi-scale spatiotemporal units. For the antecedent of the rule, For the rule consequent, where N is the total number of spatiotemporal transactions within a multi-scale spatiotemporal unit; The spatiotemporal correlation cube model is established based on multi-scale spatiotemporal units, extreme climate disaster event sequences, and correlation rule metrics as follows: First, the basic spatial and temporal dimensions of the multi-scale spatiotemporal unit are divided into a pyramid layer and a custom query layer. Then, based on the coupling relationship between the pyramid layer and the custom query layer, a spatiotemporal coupling layer is established, and the association rule metric within the spatiotemporal unit is calculated. Geographic association dimension is established based on the categories and levels of extreme climate disaster events. Finally, a spatiotemporal association cube model is established based on the spatial dimension, temporal dimension, spatiotemporal coupling layer, geographic association dimension, and extreme climate disaster event sequence.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the real-time interactive extreme climate disaster event association mining method as described in any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the real-time interactive extreme climate disaster event association mining method as described in any one of claims 1 to 4.

Images