Extreme climate spatial-temporal characteristic analysis system and method based on horizontal visibility graph network

Abstract

The invention discloses an extreme climate spatial-temporal characteristic analysis system and method based on a horizontal visual graph network. The system comprises a data arrangement module, a horizontal visual graph network construction module, a temporal characteristic analysis module, a similar network construction module and a spatial correlation analysis module. The data arrangement module is used for collecting and processing original meteorological data of different meteorological stations in a specific time period to obtain a time sequence of occurrence frequencies of extreme climate events of the different meteorological stations; the level visual graph network construction module constructs an extreme climate level visual graph network of different meteorological stations based on a time sequence; the time characteristic analysis module obtains time change characteristics of the extreme climate events based on the extreme climate level views; the similar network construction module is used for calculating similar coefficients among the horizontal visual graph networks of different sites, and constructing an extreme climate event similar network based on the similar coefficients; and the spatial correlation analysis module obtains spatial correlation characteristics of the extreme climate events based on the extreme climate event similar network.

Description

Technical Field

[0001] The present invention belongs to the technical field of extreme meteorological event analysis, and in particular relates to an extreme climate spatiotemporal feature analysis system and method based on a horizontal visual graph network. Background Art

[0002] In the past decade, the use of complex network theory to extract useful information from time series has become a research hotspot and frontier in many disciplines. Complex networks are a powerful tool for spatial and temporal data analysis. They have been successfully applied in many fields such as financial markets, medical health, and interpersonal relationships, but they are still in the exploratory stage in climate change research.

[0003] Visibility Graphs (VG) are an important method for converting time series into complex networks. They are intuitive and invariant to affine transformations, and can show multiple network topological characteristics, providing new possibilities for exploring the complex spatiotemporal evolution characteristics of extreme climate events. Compared with VG, Horizontal Visibility Graph (HVG) can better reflect the extreme value changes of time series and is more suitable for analyzing local extreme value changes in climate events.

[0004] In the existing extreme climate event analysis technology, linear mathematical statistics methods are difficult to meet the analysis needs of complex and evolving extreme climate events, and there is a lack of effective methods that can deeply mine the implicit relationships of climate data and accurately identify outliers. Therefore, an extreme climate analysis method using horizontal visualization is proposed. Summary of the invention

[0005] The present invention aims to solve the deficiencies of the prior art and provides the following solutions:

[0006] An extreme climate spatiotemporal characteristic analysis system based on a horizontal visual graph network, comprising: a data sorting module, a horizontal visual graph network construction module, a time characteristic analysis module, a similarity network construction module and a spatial correlation analysis module;

[0007] The data sorting module is used to collect the original meteorological data of different meteorological stations in a specific time period, and process the original meteorological data to obtain the time series of the frequency of extreme climate events at different meteorological stations;

[0008] The horizontal visible graph network construction module constructs a horizontal visible graph network of extreme climates at different meteorological sites based on the time series;

[0009] The temporal characteristic analysis module determines abnormal years and divides change stages of extreme climate events based on the horizontal visual graph network to obtain temporal change characteristics of extreme climate events;

[0010] The similarity network construction module calculates similarity coefficients between the horizontal visual graph networks of different sites using the random walk principle, and constructs a similarity network for extreme climate events based on the similarity coefficients;

[0011] The spatial correlation analysis module performs representative site analysis and regional correlation analysis on extreme climate events based on the extreme climate event similarity network to obtain spatial correlation characteristics of extreme climate events.

[0012] Preferably, the data sorting module includes: a data collection unit, a preprocessing unit and a sorting and statistics unit;

[0013] The data collection unit is used to collect original meteorological data of different meteorological stations in specific time periods;

[0014] The pre-processing unit is used to perform data cleaning and data filling on the raw data to obtain processed meteorological data;

[0015] The sorting and statistics unit is used to sort and statistics the processed meteorological data according to meteorological stations and years, so as to obtain the time series of the frequency of extreme climate events in specific time periods at different meteorological stations.

[0016] Preferably, the workflow of the horizontal visual graph network construction module includes: traversing the time series of each weather station, for each pair of nodes and (i<j), determine whether it is visible according to the visibility criteria. If it is visible, add edges between the corresponding network nodes and construct an adjacency matrix A to represent the horizontal visibility graph network;

[0017] The visibility criteria include: if all other data points between the two data points are located below the horizontal line segment connecting the two points, then two data points are considered to be visible, that is, there is an edge between them.

[0018] Preferably, the time feature analysis module includes: an abnormal year judgment unit and a change stage division unit;

[0019] The abnormal year judgment unit calculates the first degree value of each node in each horizontal visual graph network through Gephi, and judges the abnormal year of the extreme climate event based on the size of the first degree value: the nodes whose first degree value is higher than the first preset threshold value correspond to the years when extreme climate events occur frequently and the event intensity is high, and the nodes whose first degree value is lower than the first preset threshold value correspond to the years when extreme climate events occur infrequently and the event intensity is low;

[0020] The change stage division unit calculates the communities in each horizontal visible graph network through Gephi, and renders the community discovery results in colors: the time nodes corresponding to the same color community are the same change stage, and the time nodes corresponding to different color communities are different stages of extreme climate change.

[0021] Preferably, the similar network construction module includes: a path coincidence calculation unit and a network construction unit;

[0022] The path overlap calculation unit uses the random walk principle to perform random walks on each of the horizontal visual graph networks and records the walk paths, and then calculates the path overlaps between different sites, and uses the path overlaps as the similarity coefficient;

[0023] When the similarity coefficient is greater than a second preset threshold, the network construction unit considers that the two horizontal visibility graphs are similar, and establishes an edge between the two similar horizontal visibility graphs as new nodes, traverses all the horizontal visibility graph networks, and constructs the extreme climate event similarity network.

[0024] Preferably, the spatial correlation analysis module includes: a site analysis unit and a regional correlation analysis unit;

[0025] The site analysis unit calculates the second degree value of each node in the extreme climate event similarity network through Gephi, the larger the second degree value is, the more sites similar to the extreme climate change in the region are, and the node with the second degree value greater than the third preset threshold is selected as the representative site;

[0026] The regional correlation analysis unit calculates the communities in the similarity network of each extreme climate event through Gephi, and renders the community discovery results in colors: the meteorological station groups corresponding to the same color community are areas with similar extreme climate change characteristics, and the regional correlation characteristics of extreme climate changes are obtained.

[0027] The present invention also provides a method for analyzing the spatiotemporal characteristics of extreme climate based on a horizontal visual graph network, the method being applied to any of the above-mentioned analysis systems and comprising the following steps:

[0028] Collecting raw meteorological data of specific time periods at different meteorological stations, and processing the raw meteorological data to obtain a time series of the frequency of extreme climate events at different meteorological stations;

[0029] constructing a horizontal visual graph network of extreme climates at different meteorological stations based on the time series;

[0030] Based on the horizontal visual graph network, the abnormal year judgment and change stage division of extreme climate events are carried out to obtain the time change characteristics of extreme climate events;

[0031] The similarity coefficients between the horizontal visual graph networks of different stations are calculated by using the random walk principle, and a similarity network of extreme climate events is constructed based on the similarity coefficients;

[0032] Based on the extreme climate event similarity network, representative site analysis and regional correlation analysis of extreme climate events are performed to obtain the spatial correlation characteristics of extreme climate events.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) The horizontal visual network analysis method of extreme climate events constructed by the present invention can capture the local and global characteristics of extreme climate event time series data, and can quickly, intuitively and accurately identify abnormal years of extreme climate events;

[0035] (2) The horizontal visual network analysis method of extreme climate events constructed by the present invention can well capture the stage characteristics of extreme climate and provide a new basis for the division of climate change stages;

[0036] (3) The extreme climate event similarity network analysis method constructed by the present invention on the basis of the horizontal visual graph network pays more attention to the historical evolution characteristics of events when analyzing and mining the correlation of climate change in different regions and selecting regional representative sites, and the analysis results have greater application value;

[0037] (4) The complex network analysis method formed by the present invention is usable, intuitive and efficient in analyzing and mining extreme climate data information, providing new ideas and methods for the study of climate change, and has great expansion space and application potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0039] Figure 1 A schematic diagram of the system structure of an embodiment of the present invention;

[0040] Figure 2 Schematic diagram of converting observation values ​​as nodes into a horizontal visual graph network according to an embodiment of the present invention, wherein a is a bar graph of a time series, and b is a horizontal visual graph network graph;

[0041] Figure 3 A schematic diagram of constructing a similarity network for extreme climate events according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Embodiment 1

[0045] In this embodiment, if Figure 1 As shown, a system for analyzing the spatiotemporal characteristics of extreme climate based on a horizontal visible graph network includes: a data sorting module, a horizontal visible graph network construction module, a time characteristic analysis module, a similarity network construction module and a spatial correlation analysis module.

[0046] The data collation module is used to collect the original meteorological data of different meteorological stations in a specific time period, and process the original meteorological data to obtain the time series of the frequency of extreme climate events at different meteorological stations.

[0047] The data sorting module includes: data collection unit, preprocessing unit and sorting and statistics unit. The data collection unit is used to collect the original meteorological data of different meteorological stations in a specific time period; the preprocessing unit is used to clean and fill the original data to obtain processed meteorological data; the sorting and statistics unit is used to sort and count the processed meteorological data according to meteorological stations and years to obtain the time series of the frequency of extreme climate events in different meteorological stations in a specific time period.

[0048] In this embodiment, the data collection unit collects meteorological data of a specific time period (such as 1951-2019) of the relevant meteorological station, and the preprocessing unit cleans the original meteorological data to remove erroneous data and missing values. For a small number of missing values, interpolation methods (such as linear interpolation) are used to supplement; the data comes from the actual observation records of the meteorological observation station, and undergoes strict data quality control and preprocessing to ensure the accuracy and completeness of the data. The sorting and statistical unit sorts the data according to the meteorological station and year to form a time series of extreme climate events for each meteorological station. The 90% quantile value of the period is calculated for different meteorological stations respectively, and this is used as the threshold for the occurrence of extreme climate events. The number of extreme climate events each year is statistically analyzed to obtain a time series of the frequency of extreme climate events in specific periods of different meteorological stations.

[0049] The horizontal visibility graph network construction module constructs a horizontal visibility graph network of extreme climate at different meteorological stations based on time series.

[0050] The workflow of the horizontal visibility graph network building module includes: traversing the time series of each weather station, for each pair of nodes and (i<j), determine whether it is visible according to the visibility criteria. If it is visible, add edges between the corresponding network nodes and construct an adjacency matrix A to represent the horizontal visibility graph network; the visibility criteria include: if all other data points between two data points are located below the horizontal line segment connecting the two points, then the two data points are considered to be visible, that is, there is an edge between them.

[0051] Horizontal Visibility Graph (HVG) is a method to transform time series data into a complex network. The core idea is to map each data point in the time series into a two-dimensional coordinate system, such as Figure 2 (a), where the horizontal axis represents time and the vertical axis represents data value. In this embodiment, according to the principle of horizontal visibility graph (HVG), the time series of interannual extreme climate events in characteristic time periods of different weather stations are converted into extreme climate horizontal visibility graph network (CE-HVGN), and the calculation process is implemented by Python. That is, the horizontal visibility graph algorithm is implemented using Python programming language. The time series of each weather station is traversed, and for each pair of nodes and (i<j), according to the visibility criteria, determine whether it is visible. If it is visible, add edges between the corresponding network nodes and construct an adjacency matrix A to represent the horizontal visible graph network. If there is an edge between nodes and, then A ij =A ij =1; otherwise A ij =A ij = 0. The visual criteria include: if all other data points between two data points are located below the horizontal line segment connecting the two points, then the two data points are considered to be visible, that is, there is an edge between them. The horizontal visual graph network constructed is as follows Figure 2 (b) as shown.

[0052] The temporal characteristic analysis module determines abnormal years and divides change stages of extreme climate events based on a horizontal visual graph network, and obtains the temporal change characteristics of extreme climate events.

[0053] The time feature analysis module includes: an abnormal year judgment unit and a change stage division unit. The abnormal year judgment unit calculates the first degree value of each node in each horizontal visual graph network through Gephi, and judges the abnormal year of extreme climate events based on the size of the first degree value: nodes with a first degree value higher than the first preset threshold correspond to years with frequent extreme climate events and high event intensity, and nodes with a first degree value lower than the first preset threshold correspond to years with infrequent extreme climate events and low event intensity; the change stage division unit calculates the communities in each horizontal visual graph network through Gephi, and color-renders the community discovery results: the time nodes corresponding to the same color community are the same change stage, and the time nodes corresponding to different color communities are different stages of extreme climate change. In this embodiment, the setting of the first preset threshold needs to be determined according to the node degree values ​​of different networks.

[0054] In this embodiment, the abnormal year judgment unit uses Gephi software to calculate the degree of each node of the horizontal visual graph network, that is, the number of edges connected to each node in the network. In a complex network, the greater the node degree, the higher its importance. Based on this principle, the degree of the node is set to network rendering through Gephi software, and the final horizontal visual graph network is presented as a network of different sizes. In the analysis of extreme climate events, nodes with higher degree values ​​correspond to years when extreme climate events occur more frequently or with greater intensity, and there is a strong correlation change with events in other years. Nodes with higher degree values ​​correspond to years when extreme climate events do not occur frequently or with low intensity.

[0055] The change stage division unit uses Gephi software to calculate the network communities of each horizontal visual graph, that is, the relatively closely connected node groups in the network. Through the community discovery function of the software, the community discovery of the horizontal visual graph network is carried out (the resolution is set to the standard resolution, and the resolution can also be set according to actual needs. The higher the resolution, the more communities are divided). More than 5 division attempts are made during the division process, and the one with the highest community modularity (modularity value) is selected as the final community discovery result. Then, the different community discovery results are rendered in color through Gephi software. Finally, the time nodes corresponding to the same color community are the same change stage, and the time nodes corresponding to different color communities are different stages of extreme climate change.

[0056] By integrating the results of the abnormal year judgment unit and the change stage division unit, the temporal change characteristics of extreme climate events are obtained.

[0057] The similarity network construction module uses the random walk principle to calculate the similarity coefficients between the horizontal visual graph networks of different sites, and constructs a similarity network of extreme climate events based on the similarity coefficients.

[0058] The similarity network construction module includes: a path overlap calculation unit and a network construction unit. The path overlap calculation unit uses the random walk principle to perform random walks on each horizontal visibility graph network and record the walk paths, then calculates the path overlap between different sites and uses the path overlap as the similarity coefficient; the network construction unit considers that the two horizontal visibility graphs are similar when the similarity coefficient is greater than the second preset threshold, and establishes an edge between the two similar horizontal visibility graphs as new nodes, traverses all horizontal visibility graph networks, and constructs a similar network for extreme climate events.

[0059] In this embodiment, the path overlap calculation unit uses the random walk principle to calculate the similarity coefficient between horizontal visual graph networks of different extreme climate events. That is, by performing random walks on two horizontal visual graph networks respectively and recording the walk paths, the path overlap can be calculated. Based on the path overlap, the similarity between networks is judged, thereby judging the similarity of extreme climate change conditions at different meteorological stations. The calculation process is implemented in Python, and the basic principle of random walk is as follows: (1) In the horizontal visual graph networks G1 = (V1, E1) and G2 = (V2, E2), V1 represents the node set of network G1, E1 represents the edge set of network G1, V2 represents the node set of network G2, and E2 represents the edge set of network G2. The random walk starts from node u∈V1 and node v∈V2 (here u and v can be nodes corresponding to the horizontal visual graph principle, that is, nodes corresponding to years with similar extreme climate event characteristics). At each step, node v is added with probability Move to its neighbor node y, where d(x) is the degree of node x in the visual graph network (i.e., the number of nodes connected by visual criteria). (2) Record the random walk path P1 = {u0 = u, u1, u2, ..., u T} and P2 = {v0 = v, v1, v2, ..., v T}. Then the path overlap is calculated as a similarity indicator, for example If u i and v i Under a certain mapping, they are corresponding nodes (nodes corresponding to the same climate characteristic years in the extreme climate event visual graph network), where δ(u i , v i )=1, otherwise δ(u i , v i )=0.

[0060] like Figure 3As shown in the figure, when the similarity coefficient of the network construction unit is ≥ 0.5 (the threshold of the similarity coefficient can be adjusted according to research needs), the two are considered similar, and an edge is established between the two horizontal visible graph networks as new nodes, that is, the edge between the two nodes is 1, otherwise it is 0. Traverse all horizontal visible graph networks to build a similar network of extreme climate events.

[0061] The spatial correlation analysis module conducts representative site analysis and regional correlation analysis on extreme climate events based on the extreme climate event similarity network to obtain the spatial correlation characteristics of extreme climate events.

[0062] The spatial correlation analysis module includes: a site analysis unit and a regional correlation analysis unit. The site analysis unit calculates the second degree value of each node in the extreme climate event similarity network through Gephi. The larger the second degree value, the more sites with similar extreme climate change in the region. The nodes with second degree values ​​greater than the third preset threshold are selected as representative sites; the regional correlation analysis unit calculates the communities in the extreme climate event similarity network through Gephi, and renders the community discovery results in color: the meteorological station groups corresponding to the same color community are areas with similar extreme climate change characteristics, and the regional characteristics of extreme climate change are obtained.

[0063] In this embodiment, the site analysis unit uses Gephi software to calculate the degree value of each node of the similar network, and sets the network rendering mode by the degree value of the node through Gephi software. Finally, the extreme climate similar network is presented as a network of different sizes. The larger the middle and large nodes of the network, the more nodes are connected to it, which means that there are more areas with similar extreme climate changes to this area. The nodes with the second degree value greater than the third preset threshold are selected as representative sites.

[0064] The regional correlation analysis unit uses Gephi software to calculate the network communities of each horizontal visual graph, that is, the node groups with relatively close connections in the network. Through the community discovery function of the software, the community discovery is performed on the horizontal visual graph network (the resolution is set to standard resolution, and the resolution can also be set according to actual needs. The higher the resolution, the more communities are divided). More than 5 division attempts are made during the division process, and the one with the highest community modularity (modularity value) is selected as the final community discovery result. Then, the different community discovery results are rendered in color through Gephi software. Finally, the meteorological station groups corresponding to the same color community are areas with similar extreme climate change characteristics, and the regional characteristics of extreme climate change are obtained.

[0065] By integrating the results of the abnormal year site analysis unit and the regional correlation analysis unit, the spatial variation characteristics of extreme climate events were obtained.

[0066] Based on the obtained temporal and spatial variation characteristics of extreme climate events, the analysis of temporal and spatial characteristics of extreme climate is completed.

[0067] Embodiment 2

[0068] In this embodiment, a method for analyzing spatiotemporal characteristics of extreme climate based on a horizontal visual graph network includes the following steps:

[0069] The original meteorological data of specific time periods at different meteorological stations are collected and processed to obtain the time series of the frequency of extreme climate events at different meteorological stations; a horizontal visual graph network of extreme climate at different meteorological stations is constructed based on the time series; abnormal years of extreme climate events are judged and the change stages are divided based on the horizontal visual graph of extreme climate to obtain the temporal change characteristics of extreme climate events; the similarity coefficients between the horizontal visual graph networks of different stations are calculated using the random walk principle, and a similarity network of extreme climate events is constructed based on the similarity coefficients; representative station analysis and regional correlation analysis of extreme climate events are carried out based on the similarity network of extreme climate events to obtain the spatial correlation characteristics of extreme climate events.

[0070] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network, characterized in that: include: Data sorting module, horizontal visual graph network construction module, time feature analysis module, similarity network construction module and spatial correlation analysis module; The data sorting module is used to collect the original meteorological data of different meteorological stations in a specific time period, and process the original meteorological data to obtain the time series of the frequency of extreme climate events at different meteorological stations; The horizontal visible graph network construction module constructs a horizontal visible graph network of extreme climates at different meteorological sites based on the time series; The temporal characteristic analysis module determines abnormal years and divides change stages of extreme climate events based on the horizontal visual graph network to obtain temporal change characteristics of extreme climate events; The similarity network construction module calculates similarity coefficients between the horizontal visual graph networks of different sites using the random walk principle, and constructs a similarity network for extreme climate events based on the similarity coefficients; The spatial correlation analysis module performs representative site analysis and regional correlation analysis on extreme climate events based on the extreme climate event similarity network to obtain spatial correlation characteristics of extreme climate events.

2. The extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network according to claim 1, characterized in that: The data sorting module includes: a data collection unit, a preprocessing unit and a sorting and statistics unit; The data collection unit is used to collect original meteorological data of different meteorological stations in specific time periods; The pre-processing unit is used to perform data cleaning and data filling on the raw data to obtain processed meteorological data; The sorting and statistics unit is used to sort and statistics the processed meteorological data according to meteorological stations and years, so as to obtain the time series of the frequency of extreme climate events in specific time periods at different meteorological stations.

3. The extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network according to claim 1 is characterized in that: The workflow of the horizontal visual graph network construction module includes: traversing the time series of each weather station, for each pair of nodes and (i<j), determine whether it is visible according to the visibility criteria. If it is visible, add edges between the corresponding network nodes and construct an adjacency matrix A to represent the horizontal visibility graph network; The visibility criteria include: if all other data points between the two data points are located below the horizontal line segment connecting the two points, then two data points are considered to be visible, that is, there is an edge between them.

4. The extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network according to claim 1 is characterized in that: The time feature analysis module includes: an abnormal year judgment unit and a change stage division unit; The abnormal year judgment unit calculates the first degree value of each node in each horizontal visual graph network through Gephi, and judges the abnormal year of the extreme climate event based on the size of the first degree value: the nodes whose first degree value is higher than the first preset threshold value correspond to the years when extreme climate events occur frequently and the event intensity is high, and the nodes whose first degree value is lower than the first preset threshold value correspond to the years when extreme climate events occur infrequently and the event intensity is low; The change stage division unit calculates the communities in each horizontal visible graph network through Gephi, and renders the community discovery results in colors: the time nodes corresponding to the same color community are the same change stage, and the time nodes corresponding to different color communities are different stages of extreme climate change.

5. The extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network according to claim 1 is characterized in that: The similar network construction module includes: a path coincidence calculation unit and a network construction unit; The path overlap calculation unit uses the random walk principle to perform random walks on each of the horizontal visual graph networks and records the walk paths, and then calculates the path overlaps between different sites, and uses the path overlaps as the similarity coefficient; When the similarity coefficient is greater than a second preset threshold, the network construction unit considers that the two horizontal visibility graphs are similar, and establishes an edge between the two similar horizontal visibility graphs as new nodes, traverses all the horizontal visibility graph networks, and constructs the extreme climate event similarity network.

6. The extreme climate spatiotemporal characteristics analysis system based on horizontal visual graph network according to claim 1, characterized in that: The spatial correlation analysis module includes: a site analysis unit and a regional correlation analysis unit; The site analysis unit calculates the second degree value of each node in the extreme climate event similarity network through Gephi, the larger the second degree value is, the more sites similar to the extreme climate change in the region are, and the node with the second degree value greater than the third preset threshold is selected as the representative site; The regional correlation analysis unit calculates the communities in the similarity network of each extreme climate event through Gephi, and renders the community discovery results in color: the meteorological station groups corresponding to the same color community are areas with similar extreme climate change characteristics, and the regional correlation characteristics of extreme climate changes are obtained.

7. A method for analyzing spatiotemporal characteristics of extreme climate based on a horizontal visual graph network, the method being applied to the analysis system according to any one of claims 1 to 6, characterized in that: The following steps are involved: Collecting raw meteorological data of specific time periods at different meteorological stations, and processing the raw meteorological data to obtain a time series of the frequency of extreme climate events at different meteorological stations; constructing a horizontal visual graph network of extreme climates at different meteorological stations based on the time series; Based on the horizontal visual graph network, the abnormal year judgment and change stage division of extreme climate events are carried out to obtain the time change characteristics of extreme climate events; The similarity coefficients between the horizontal visual graph networks of different stations are calculated by using the random walk principle, and a similarity network of extreme climate events is constructed based on the similarity coefficients; Based on the extreme climate event similarity network, representative site analysis and regional correlation analysis of extreme climate events are performed to obtain the spatial correlation characteristics of extreme climate events.

Images