Method, device and medium for identifying and tracking high and low pressure systems in weather situation map

Abstract

The application relates to a high and low pressure system identification and tracking method, equipment and medium in a weather situation map, and belongs to the technical field of weather situation identification. Multi-source meteorological data is collected, and an isogram dataset is constructed based on the multi-source meteorological data; a closed isogram in the isogram dataset is extracted, the high and low pressure properties of each closed isogram are determined through a high and low pressure system identification model; the minimum circumscribed rectangle of the closed isogram is calculated; all the closed isograms are traversed, it is judged whether the minimum circumscribed rectangle is contained by the minimum circumscribed rectangle of other isograms, the outermost closed isogram is determined, and a set U is formed; a hierarchical high and low pressure system model is established based on an N-ary tree structure, and the coverage range and the influence area of the high and low pressure system are calculated based on the N-ary tree hierarchical structure, so as to assist in meteorological prediction. The application does not require a large amount of prior training data, has good interpretability, and solves the problems of limited identification and lack of identification ability for specific high and low pressure weather situations.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of weather situation identification, and particularly relates to a high and low pressure system identification and tracking method in a weather situation map, equipment and a medium. BACKGROUND

[0002] The identification of high and low pressure systems in a weather situation map mainly depends on the distribution of isobar lines, and is initially mainly subjective identification by experienced forecasters. With the development of computer technology, artificial intelligence and image recognition algorithms, weather situation identification based on intelligent technology has begun to appear, but is mainly applied to the automatic identification of dense fog weather situation and low visibility weather, or a classification model is used to carry out weather situation identification, and the main disadvantage is that a weather situation classification model needs to be trained in advance, which may lead to limitations of identification accuracy and insufficient identification ability for specific high and low pressure weather situations.

[0003] Therefore, how to solve the problems of limited automatic identification and insufficient identification ability for specific high and low pressure weather situations in the prior art is a subject worthy of study. SUMMARY

[0004] In view of the above analysis, the embodiments of the present application aim to provide a high and low pressure system identification and tracking method in a weather situation map, equipment and a medium, and aim to solve the problems of limited automatic identification and insufficient identification ability for specific high and low pressure weather situations in the prior art.

[0005] The first aspect of the present application provides a high and low pressure system identification and tracking method in a weather situation map, comprising:

[0006] Collecting multi-source meteorological data including meteorological observation data, satellite remote sensing data and radar data, and constructing an isobar data set based on the multi-source meteorological data;

[0007] Extracting closed isobars in the isobar data set, determining the high and low pressure properties of each closed isobar through a high and low pressure system identification model;

[0008] Calculating the minimum circumscribed rectangle of the closed isobar;

[0009] Traversing all the closed isobars to determine whether the minimum circumscribed rectangle is contained by the minimum circumscribed rectangle of other isobars, determining the outermost closed isobar, and forming a set U;

[0010] Establishing a hierarchical high and low pressure system model based on an N-ary tree structure, taking each closed isobar in the set U as a root node, traversing the isobars contained therein, and establishing an N-ary tree hierarchical structure to represent the nesting relationship of the high and low pressure system;

[0011] Based on the N-ary tree hierarchy, coverage and influence area of high and low pressure systems are calculated to assist weather prediction.

[0012] Optionally, the method further comprises:

[0013] Contour data at different time steps are obtained, and a first contour set Set0 and a second contour set Set1 are respectively constructed; the first contour set Set0 is contour data corresponding to a previous time T-1, and the second contour set Set1 is contour data corresponding to a current time T;

[0014] A distance matrix of the first contour set Set0 and the second contour set Set1 is established to determine a matching relationship and a moving direction between contours at different time steps;

[0015] A matrix operation is used to calculate the moving direction of the high and low pressure systems, and to identify contour merging, splitting, new birth and disappearance, so as to dynamically track the high and low pressure systems.

[0016] Optionally, the establishment of the distance matrix of the first contour set Set0 and the second contour set Set1, and the determination of the matching relationship and the moving direction between contours at different time steps comprise:

[0017] An m×n two-dimensional matrix D is established; Set0 has m contours, and Set1 has n contours; D[i][j] represents the matching distance between the i-th contour in Set0 and the j-th contour in Set1;

[0018] Point-to-point distances of contours in Set0 and Set1 are calculated to construct an a×b two-dimensional distance matrix; a is the number of points of a contour in Set0, and b is the number of points of a contour in Set1; the distance between each point in a and each point in b is calculated, and the two-dimensional distance matrix is filled;

[0019] A Boolean matching matrix is constructed; wherein a matrix 0 and a matrix 1 are created; matrix 0: for each row of the a×b distance matrix, find the position of the minimum value, set the position to 1, and set the remaining positions to 0; matrix 1: for each column of the a×b distance matrix, find the position of the minimum value, set the position to 1, and set the remaining positions to 0; logical AND operation of matrix 0 and matrix 1 to obtain the Boolean matching matrix; the matching relationship and the moving direction between contours at different time steps are determined through the Boolean matching matrix.

[0020] Optionally, the calculation of the moving direction of the high and low pressure systems by the matrix operation comprises:

[0021] Constructing a high-low pressure tree matching matrix of cxd; wherein, c is the number of high-low pressure trees at T-1 time; d is the number of high-low pressure trees at T time; array0[i][j] represents how many isolines in the high-low pressure tree i at T-1 time are contained by the high-low pressure tree j at T time; array1[i][j] represents how many isolines in the high-low pressure tree j at T time are contained by the high-low pressure tree i at T-1 time;

[0022] Calculating the final matching matrix array2, setting the position of the maximum value of the row in the matrix array0 as 1 and the rest as 0; setting the position of the maximum value of the column in the matrix array1 as 1 and the rest as 0; performing logical AND operation on array0 and array1 to obtain the final matching matrix array2;

[0023] Traversing array2 to find the matched high-low pressure system and calculating the moving direction vector of the matched isoline; calculating the average moving direction of all matched isolines as the motion direction of the high-low pressure system.

[0024] Optionally, the identification of the merging, splitting, new generation and disappearance of the isoline includes:

[0025] When the T-1 high-low pressure system cannot be matched, it is identified as the disappearance of the isoline;

[0026] When the T high-low pressure system cannot be matched, it is identified as the new generation of the isoline;

[0027] When multiple T-1 high-low pressure systems are matched to the same T high-low pressure system, it is identified as the merging of the isoline;

[0028] When one T-1 high-low pressure system is matched to multiple T high-low pressure systems, it is identified as the splitting of the isoline.

[0029] Optionally, it further includes:

[0030] Collecting high-low pressure system data at multiple time steps, establishing an evolution sequence of the high-low pressure system through the matching result of the high-low pressure system, and forming multiple high-low pressure system trajectories with time continuity;

[0031] For each trajectory, extracting the corresponding time sequence features to form an evolution sample set of the high-low pressure system;

[0032] Inputting the evolution sample set into a high-low pressure system evolution model based on a long-short term memory network or a Transformer structure for training to obtain a deep learning model for predicting the position, intensity and evolution trend of the high-low pressure system at a future time step

[0033] Optionally, the calculation of the coverage range and the influence area of the high-low pressure system based on the N-ary tree hierarchical structure includes:

[0034] Traverse the N-ary tree, take the union of all the minimum bounding rectangles of the isopleths, and obtain the coverage range of the high and low pressure systems;

[0035] Combine the wind field data, temperature data, and humidity data to calculate the influence range of the high and low pressure systems.

[0036] Optionally, the hierarchical high and low pressure system model is established based on the N-ary tree structure, taking each closed isopleth in the set U as a root node, traversing the isopleths contained therein, and establishing an N-ary tree hierarchical structure, including:

[0037] Traverse the isopleth set, and establish the N-ary tree hierarchical structure according to the nesting relationship between the isopleths, to represent the nesting relationship of the high and low pressure systems.

[0038] Traverse the tree structure, check whether the high and low pressure attributes of the isopleths are inconsistent with the root node; if inconsistent, trim the node to make it a new root node.

[0039] The second aspect of the present application provides a high and low pressure system identification and tracking device in a weather situation map, including a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to realize the weather situation map high and low pressure system identification and tracking method according to any one of the above.

[0040] The third aspect of the present application provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to realize the weather situation map high and low pressure system identification and tracking method according to any one of the above.

[0041] The weather situation map high and low pressure system identification and tracking method provided by the present application represents the nested isopleth structure in the high and low pressure system as an N-ary tree model, so that the center and hierarchical relationship of each system are clear. This process is determined by the physical structure of the isobaric line distribution itself and does not depend on black box model training. Using the isopleth direction and closed characteristics, the high or low pressure center is directly judged, improving the physical interpretability and accuracy of the identification, and avoiding the disadvantages of traditional classification model labeling identification. Compared with the traditional method relying on artificial identification or image classification model, the present application innovatively constructs an N-ary tree structure based on the isopleth nesting relationship, accurately representing the hierarchical relationship of the high and low pressure system. This method does not require a large amount of prior training data, has good interpretability, structure, and generalization, and is a structured modeling method for automatic identification and prediction of high and low pressure systems. Further, the present application realizes dynamic matching and system tracking of multiple time step isopleths through geometric nesting and distance matrix, further completing the structural identification and evolution trend analysis of the high and low pressure system.

[0042] In addition, the application also provides a high and low pressure system identification and tracking device and medium in a weather situation map. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0044] Figure 1 The flow chart of one specific embodiment of the high and low pressure system identification and tracking method provided by the present application in a weather situation map;

[0045] Figure 2 The implementation effect diagram of the high and low pressure system identification and tracking method provided by the present application in a weather situation map;

[0046] Figure 3 The tracking process flow chart of another specific embodiment of the high and low pressure system identification and tracking method provided by the present application;

[0047] Figure 4 The structural block diagram of the high and low pressure system identification and tracking device provided by the present application in a weather situation map. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, not all the embodiments. It should be noted that the embodiments and the features in the embodiments in the present disclosure can be combined, separated, interchanged and / or rearranged without conflict. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0049] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Furthermore, to the extent that the terms "comprising," "including," "containing," and / or "having" and variations thereof are used herein, such terms are intended to be inclusive, in an aspect, it is noted that the terms "substantial," "approximately," and other similar terms are used as terms of approximation and not as terms of degree, accordingly, they are utilized to account for inherent deviations in measurements, calculations, and / or provided values that would be recognized by those of ordinary skill in the art.

[0050] A flow chart of one embodiment of the method for identifying and tracking high and low pressure systems in a weather situation map provided by the present application is shown in FIG. 1, which specifically includes the following steps: Figure 1

[0051] S101: Collect multi-source meteorological data, including meteorological observation data, satellite remote sensing data, and radar data, and construct an isopleth data set based on the multi-source meteorological data.

[0052] The multi-source meteorological data includes meteorological observation data, satellite remote sensing data, and radar data. The meteorological observation data can be obtained from ground meteorological stations and sounding stations, and data such as air pressure, temperature, and humidity can be collected. The satellite remote sensing data is used to extract large-scale air pressure fields and temperature fields. The radar data is used to capture the edge structure changes of small and medium scale weather systems.

[0053] Based on the multi-source meteorological data, an isopleth data set is constructed. Specifically, spatial interpolation can be performed on variables such as air pressure, temperature, and humidity; continuous grid fields (such as isobaric fields and isothermal fields) are generated; isopleths corresponding to specific isopleths are extracted; and an isopleth data set that can be used for subsequent high and low pressure system identification and modeling is formed.

[0054] S102: Extract closed isopleths in the isopleth data set, and determine the high and low pressure properties of each closed isopleth by a high and low pressure system identification model.

[0055] The closed isopleths are determined according to whether the starting point and the ending point of the coordinate points are connected and whether the shape is closed.

[0056] The determination of the high and low pressure properties can be: if the isobaric line is a closed structure with gradually increasing air pressure, it is determined to be a high pressure property; if it is a closed structure with gradually decreasing air pressure, it is determined to be a low pressure property.

[0057] ​As a specific embodiment, the trained high-low pressure system recognition model can be used to determine the high-low pressure attribute.

[0058] S103: Calculate the minimum circumscribed rectangle of the closed contour.

[0059] Calculate the minimum circumscribed rectangle of each closed contour (a point set).

[0060] S104: Traverse all closed contours, determine whether the minimum circumscribed rectangle is contained by the minimum circumscribed rectangle of other contours, determine the outermost closed contour, and form a set U.

[0061] Traverse the minimum circumscribed rectangle of each closed contour, and determine whether it is contained by other contours; if not, it means that it is located at the outermost layer on the graph and can be used as the root node of the tree.

[0062] S105: Based on the N-ary tree structure, establish a hierarchical high-low pressure system model, take each closed contour in set U as the root node, traverse the contained contours, and establish an N-ary tree hierarchical structure to represent the nesting relationship of the high-low pressure system.

[0063] Traverse the contour set, establish the N-ary tree hierarchical structure according to the nesting relationship between the contours, and traverse the tree structure to check whether the high-low pressure attribute of the contour is inconsistent with its root node; if not, clip the node to make it a new root node.

[0064] S106: Based on the N-ary tree hierarchical structure, calculate the coverage range and influence area of the high-low pressure system to assist in weather prediction.

[0065] Traverse the N-ary tree, take the union of the minimum circumscribed rectangles of all contours, and get the coverage range of the high-low pressure system.

[0066] Combine wind field data, temperature data, and humidity data to calculate the influence range of the high-low pressure system.

[0067] The N-ary tree model is a data structure in which each node can have no more than N child nodes. This structure provides greater flexibility and can more naturally represent certain complex relationships and data, and is widely used in finance, management, decision-making, and other fields. It is an important tool for handling complex hierarchical relationships and data, but there is currently no precedent for applying it to the identification of high pressure systems and low pressure systems in weather situation maps.

[0068] The high and low pressure system is represented as a multi-layer nested closed contour on the weather map, which increases the difficulty of organization and analysis on the weather map. The deep learning is used to identify whether each closed contour is high pressure or low pressure, and then the N-ary tree growth method is used to form the topological structure of the high and low pressure system, and the coverage and depth of the high and low pressure system are determined according to the topological structure.

[0069] Based on the N-ary tree model, the present application combines the traditional contour drawing algorithm, and based on the meteorological observation data, satellite remote sensing data, radar data and other multi-source meteorological fusion data, the deep learning method is used to identify and draw the high pressure system and low pressure system in the weather situation map, which is used to support the analysis of different influences of the high pressure system and low pressure system on the weather, and further support the application in the fields of meteorological and air quality forecast and analysis evaluation.

[0070] As a specific implementation, the specific construction process includes the following steps:

[0071] 1. Obtain the high and low pressure attribute of the closed contour, and store the minimum circumscribed rectangle (x_min, y_min, x_max, y_max) of each closed contour in the computer memory. Each closed contour has three attributes: number num, high or low pressure H_L, and minimum circumscribed rectangle (x_min, y_min, x_max, y_max).

[0072] 2. Traverse the minimum circumscribed rectangle (hereinafter referred to as rectangle) of each closed contour, judge whether the rectangle is contained by other rectangle, find out the rectangle which is not contained by other rectangle, form a set U corresponding to the closed contour.

[0073] 3. Traverse each closed contour in set U, find out all closed contours contained by the closed contour, form a set u, that is, there are as many different sets u as there are closed contours in set U.

[0074] 4. Traverse each closed contour in set U, each closed contour as a root node, each root node corresponds to a specific set u mentioned in step 3, and each root node is stored in the list of trees as a tree, so each tree corresponds to a specific set u mentioned in step 3.

[0075] 5. Take the growth of a tree as an example to explain the growth method of the tree, and the corresponding specific set is u t .

[0076] Sort all closed contours in u t in order of length from long to short to form a column list t .

[0077] Traverse list tIf the closed isoline is contained by a leaf node, the closed isoline is added to the tree as a child node of the leaf node. If the closed isoline is not contained by any leaf node, the parent nodes of the leaf nodes are traversed to determine whether the closed isoline is contained by any parent node. If the closed isoline is contained by a parent node, the closed isoline is added to the tree as a child node of the parent node. If the closed isoline is not contained by any parent node, the parent nodes of the parent nodes are traversed until a node is found that contains the closed isoline, and the closed isoline is added to the tree as a child node of the node.

[0078] 6. Traversing each tree in the tree list by layer, if the high-low pressure attribute of a node is different from that of the root node of the tree, the node is pruned as a root node and stored in the tree list. Step 6 is repeated until the length of the tree list no longer increases. The effect is as shown in Figure 2

[0079] In addition, the application can further use a point-to-point distance matrix in combination with a matching Boolean matrix to complete the correspondence between isolines; and based on matching of the high-low pressure system tree, dynamic tracking of the high-low pressure system level, calculation of a moving direction, and identification of merging / splitting are realized.

[0080] As Figure 3 Another specific implementation of the high-low pressure system recognition and tracking method provided in the application is shown in a tracking process flowchart, and the process specifically includes the following steps:

[0081] S301: Isoline data at different time steps is acquired, and a first isoline set Set0 and a second isoline set Set1 are respectively constructed; wherein the first isoline set Set0 corresponds to isoline data at a previous moment T-1, and the second isoline set Set1 corresponds to isoline data at a current moment T.

[0082] S302: A distance matrix of the first isoline set Set0 and the second isoline set Set1 is established, and a matching relationship and a moving direction between isolines at different time steps are determined.

[0083] An m*n two-dimensional matrix D is established; wherein Set0 has m isolines, and Set1 has n isolines; D[i][j] represents a matching distance between the ith isoline in Set0 and the jth isoline in Set1.

[0084] ​The point-to-point distances of the contour lines in Set0 and Set1 are calculated to construct a two-dimensional distance matrix of a x b; where a is the number of points of a contour line in Set0, and b is the number of points of a contour line in Set1, the distance between each point in a and each point in b is calculated, and the two-dimensional distance matrix is filled.

[0085] A Boolean matching matrix is constructed; where matrix 0 and matrix 1 are created; matrix 0: for each row of the a x b distance matrix, the position of the minimum value is found, which is set to 1, and the remaining positions are set to 0; matrix 1: for each column of the a x b distance matrix, the position of the minimum value is found, which is set to 1, and the remaining positions are set to 0; the logical AND operation of matrix 0 and matrix 1 obtains the Boolean matching matrix; through the Boolean matching matrix, the matching relationship and the moving direction between the contour lines at different time steps are determined.

[0086] S303: The moving direction of the high-low pressure system is calculated by matrix operation, and the merging, splitting, new birth and disappearance of the contour lines are identified to dynamically track the high-low pressure system.

[0087] A high-low pressure tree matching matrix of c x d is constructed; where c is the number of high-low pressure trees at T-1 time; d is the number of high-low pressure trees at T time; array0[i][j] represents how many contour lines in high-low pressure tree i at T-1 time are contained by high-low pressure tree j at T time; array1[i][j] represents how many contour lines in high-low pressure tree j at T time are contained by high-low pressure tree i at T-1 time.

[0088] The final matching matrix array2 is calculated, the position of the maximum value in the row of matrix array0 is set to 1, and the remaining positions are set to 0; the position of the maximum value in the column of matrix array1 is set to 1, and the remaining positions are set to 0; the logical AND operation of array0 and array1 obtains the final matching matrix array2.

[0089] array2 is traversed to find the matching high-low pressure system, and the moving direction vector of the matching contour line is calculated; the average moving direction of all matching contour lines is calculated as the motion direction of the high-low pressure system.

[0090] The process of identifying the merging, splitting, new birth and disappearance of the contour lines includes:

[0091] When the T-1 high-low pressure system cannot be matched, it is identified as the disappearance of the contour line.

[0092] When the T high-low pressure system cannot be matched, it is identified as the new birth of the contour line.

[0093] When multiple T-1 high-low pressure systems are matched to the same T high-low pressure system, it is identified as the merging of the contour line.

[0094] In a case where a T-1 high-low pressure system is matched to multiple T high-low pressure systems, the isopleth is identified as being split.

[0095] As a specific implementation, the tracking module is implemented as follows:

[0096] (1) Extract all isopleths of the previous time T-1, and set the ordered set as Set0. Extract all isopleths of the next time T, and set the ordered set as Set1.

[0097] (2) Map the isopleths in Set0 and Set1 to an image with a resolution of w*h, so that the coordinates of the isopleths in Set0 and Set1 are all picture coordinates.

[0098] (3) Clean all isopleths in Set0 and Set1, specifically: if the front and back points of an isopleth are the same, only one is kept.

[0099] (4) Group all isopleths in Set0 and Set1 according to the pressure values of the isopleths, with the same pressure value as a group.

[0100] (5) Traverse all pressure values and process them according to the following steps:

[0101] ① Assume that the number of isopleths in Set0 is m and the number of isopleths in Set1 is n, and establish a two-dimensional matrix of m*n.

[0102] ② Calculate the distance between the isopleths in Set0 and Set1, specifically: assume that the length of an isopleth in Set0 is a and the length of an isopleth in Set1 is b, establish a two-dimensional distance matrix of a*b, calculate the distance between each point in a and each point in b, and fill the two-dimensional distance matrix of a*b. Establish two empty a*b Boolean matrices, denoted as Matrix0 and Matrix1. Traverse each row in the two-dimensional distance matrix of a*b, find the position of the minimum value in the row, set the corresponding position in Matrix0 to 1 and the rest to 0. Traverse each column in the two-dimensional distance matrix of a*b, find the position of the minimum value in the column, set the corresponding position in Matrix1 to 1 and the rest to 0. Perform a logical AND operation on Matrix0 and Matrix1 to obtain Matrix2. Find the elements with a value of 1 in Matrix2 and their corresponding positions, find the corresponding distance according to the corresponding positions, and calculate the average of all distances as the distance between the two isopleths. Find the corresponding two points according to the corresponding positions, calculate the average of all direction vectors as the direction of movement of the isopleth according to the direction vectors of the two points.

[0103] (3) According to the method in step (5), the distances of the isopleths in Set0 and Set1 are obtained, and a two-dimensional matrix of m*n in step (1) is filled. According to the method of processing the distance matrix in step (2), matrix 2 similar to step ② is obtained. According to the matrix, the matching relationship of the isopleths and the moving direction vector of the isopleths are obtained.

[0104] (6) According to the above 5 steps, the matching relationship of each isopleth is obtained, and of course the matching relationship of each high-low pressure tree as a subset of all isopleths is also obtained. Assuming that the number of high-low pressure trees at the previous moment is c, and the number of high-low pressure trees at the next moment is d, two two-dimensional matrices of c*d are established, denoted as array0 and array1. Each high-low pressure tree in c is traversed, and each high-low pressure tree in d is traversed. The number of isopleths in each high-low pressure tree in c that is contained by each high-low pressure tree in d is judged, and the matrix array0 is filled. The number of isopleths in each high-low pressure tree in d that is contained by each high-low pressure tree in c is judged, and the matrix array1 is filled. The position of the maximum value in the row of the matrix array0 is set to 1, and the rest is set to 0. The position of the maximum value in the column of the matrix array1 is set to 1, and the rest is set to 0. The logical AND operation of the matrix array0 and the matrix array1 is performed to obtain the matrix array2. The position of 1 in the matrix array2 is found, which corresponds to the matched high-low pressure trees at the previous and next moments. The average of the moving direction vectors of the isopleths that can be matched in the two high-low pressure trees is taken as the moving direction of the high-low pressure system at the previous and next moments. When the T-1 high-low pressure system cannot be matched, it is identified as the disappearance of the isopleth; when the T high-low pressure system cannot be matched, it is identified as the birth of the isopleth; when multiple T-1 high-low pressure systems match the same T high-low pressure system, it is identified as the merging of the isopleth; and when one T-1 high-low pressure system matches multiple T high-low pressure systems, it is identified as the splitting of the isopleth.

[0105] In addition, on the basis of any of the above embodiments, the application can further include: collecting high-low pressure system data at multiple time steps, establishing an evolution sequence of the high-low pressure system through the high-low pressure system matching result, and forming multiple high-low pressure system trajectories with time continuity; for each trajectory, extracting corresponding time sequence features to form an evolution sample set of the high-low pressure system; and inputting the evolution sample set into a high-low pressure system evolution model based on a long-short term memory network or a Transformer structure for training to obtain a deep learning model for predicting the position, intensity and evolution trend of the high-low pressure system at a future time step.

[0106] In this embodiment, the above method is first used to process meteorological data within multiple consecutive time steps (e.g., hourly) to obtain the N-ary tree structure of high and low pressure systems at each time step. Then, the correspondence between high and low pressure systems across consecutive time steps is established using a distance matching matrix and a tree matching matrix. Based on the matching results, the evolution trajectory of the high and low pressure systems is formed, i.e., the evolution path of each high and low pressure system at different time points.

[0107] Based on this trajectory, time-series features of the high- and low-pressure system are extracted to construct an evolutionary sample sequence of the high- and low-pressure system at multiple time steps. By constructing a large number of such trajectory samples, an evolutionary sample set of the high- and low-pressure system can be formed.

[0108] Furthermore, the sample set is input into a deep neural network model based on a Long Short-Term Memory (LSTM) network or a Transformer structure for training. The training objective is to minimize the prediction error of the location, intensity, shape, and other features of high and low pressure systems within several future time steps. The model can be trained using supervised learning, with training labels derived from the identification results of real high and low pressure systems in subsequent time steps from actual meteorological data.

[0109] Once trained, the evolutionary model can be used for predicting future high and low pressure systems. Specifically, for a high or low pressure system at a certain point in time and its historical trajectory, the model can be input into the trained model to output predictions of its movement path, intensity change trend, and structural evolution (such as splitting or merging) for the next few hours, thus assisting in short- and medium-term weather forecasting.

[0110] The high and low pressure system evolution modeling method in this embodiment enables structured and temporal modeling of complex weather systems, improves the predictability of high and low pressure system behavior changes, and provides more accurate data support for meteorological monitoring and disaster early warning.

[0111] In addition, this application also provides a device for identifying and tracking high and low pressure systems in weather charts, such as... Figure 4 As shown, the device specifically includes a memory 41 and a processor 42. The memory 41 stores a computer program, which, when executed by the processor 42, implements the method for identifying and tracking high and low pressure systems in the weather chart described above.

[0112] It is understood that the high and low pressure system identification and tracking device provided in this application corresponds to each step in the above method, and its specific implementation can be referred to the corresponding description above, which will not be repeated here.

[0113] In addition, the application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the weather situation map high and low pressure system identification and tracking method according to any one of the above.

[0114] Computer readable storage media includes permanent and non-permanent, removable and non-removable media, which can be realized by any method or technology to store information. The information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. According to the definition in this paper, computer readable media does not include transitory computer readable media, such as modulated data signals and carriers.

[0115] The skilled person should also further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in the above description in general terms. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0116] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, software modules executed by a processor, or a combination of both. The software modules can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs or any other form of storage medium known in the art.

[0117] The above detailed description of the specific implementation is further detailed for the purpose of the application, technical solutions and beneficial effects, and it should be understood that the above description is only for the specific implementation of the application and is not used to limit the protection scope of the application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A method for identifying and tracking high and low pressure systems in weather maps, characterized in that, include: Collect multi-source meteorological data, including meteorological observation data, satellite remote sensing data, and radar data, and construct a contour dataset based on the multi-source meteorological data; Closed isolines are extracted from the contour dataset, and the high and low pressure attributes of each closed isoline are determined using the high and low pressure system identification model. Calculate the minimum bounding rectangle of a closed contour line; Traverse all closed contour lines, determine whether the smallest bounding rectangle is contained within the smallest bounding rectangle of other contour lines, determine the outermost closed contour lines, and form a set U; A hierarchical high- and low-pressure system model is established based on an N-ary tree structure. Each closed contour line in set U is taken as the root node. The model traverses the contour lines contained within each closed contour line to establish an N-ary tree hierarchy, representing the nesting relationship between the high- and low-pressure systems. This includes: traversing the contour line set and establishing an N-ary tree hierarchy based on the nesting relationship between the contour lines; traversing the tree structure and checking if any contour line's high- or low-pressure attributes are inconsistent with its root node; if inconsistent, pruning the node to become the new root node. Based on the aforementioned N-ary tree hierarchical structure, the coverage area and affected region of the high and low pressure system are calculated to assist in meteorological forecasting. Obtain contour data at different time steps, and construct a first contour set Set0 and a second contour set Set1 respectively; wherein, the first contour set Set0 is the contour data corresponding to the previous time T-1, and the second contour set Set1 is the contour data corresponding to the current time T; Establish a distance matrix between the first set of contour lines Set0 and the second set of contour lines Set1 to determine the matching relationship and movement direction between contour lines at different time steps, including: establishing an m × n two-dimensional matrix D; where Set0 has m contour lines and Set1 has n contour lines; D[i][j] represents the matching distance between the i-th contour line in Set0 and the j-th contour line in Set1; calculating the point-to-point distance between contour lines in Set0 and Set1, and constructing an a × b two-dimensional distance matrix; where a is the number of points on a contour line in Set0 and b is the number of points on a contour line in Set1, calculating the distance between each point in a and each point in b, and filling the two-dimensional distance matrix; constructing a Boolean matching matrix; where matrix 0 and matrix 1 are created; matrix 0: for each row of the a × b distance matrix, find the position of the minimum value, set that position to 1, and set the rest to 0; matrix 1: for each column of the a × b distance matrix, find the position of the minimum value, set that position to 1, and set the rest to 0; matrix 0 By performing a logical AND operation on matrix 1, a Boolean matching matrix is ​​obtained; using the Boolean matching matrix, the matching relationship and movement direction between contour lines at different time steps are determined; Matrix operations are used to calculate the movement direction of the high and low pressure system and identify the merging, splitting, new formation and disappearance of contour lines in order to dynamically track the high and low pressure system.

2. The identification and tracking method for high and low voltage systems according to claim 1, characterized in that, The calculation of the movement direction of the high and low voltage systems using matrix operations includes: Construct a c × d high-low pressure tree matching matrix; where c is the number of high-low pressure trees at time T-1; d is the number of high-low pressure trees at time T; array0[i][j] represents how many contour lines in high-low pressure tree i at time T-1 are contained in high-low pressure tree j at time T; array1[i][j] represents how many contour lines in high-low pressure tree j at time T are contained in high-low pressure tree i at time T-1; Calculate the final matching matrix array2 by setting the position of the maximum row value in matrix array0 to 1 and the rest to 0; set the position of the maximum column value in matrix array1 to 1 and the rest to 0; perform a logical AND operation between array0 and array1 to obtain the final matching matrix array2. Traverse array2 to find matching high and low pressure systems, and calculate the movement direction vector of the matching contour lines; calculate the average movement direction of all matching contour lines as the movement direction of the high and low pressure systems.

3. The identification and tracking method for high and low voltage systems according to claim 1, characterized in that, The identification of contour line merging, splitting, new formation, and disappearance includes: When a T-1 high and low pressure system cannot be matched, it is identified as a situation where contour lines disappear. When a high- or low-pressure system of T cannot be matched, it is identified as a newly formed contour line. When multiple T-1 high and low voltage systems are matched to the same T high and low voltage system, it is identified as a situation where contour lines are merged; When a T-1 high-low pressure system is matched with multiple T high-low pressure systems, it is identified as a split in the contour lines.

4. The method for identifying and tracking high and low voltage systems according to any one of claims 1 to 3, characterized in that, Also includes: Data from high and low voltage systems at multiple time steps are collected. Based on the matching results of high and low voltage systems, an evolution sequence of high and low voltage systems is established, forming multiple high and low voltage system trajectories with temporal continuity. For each trajectory, the corresponding time series features are extracted to form an evolution sample set of the high and low pressure system; The evolutionary sample set is input into a high-low pressure system evolution model based on a long short-term memory network or a Transformer structure for training, resulting in a deep learning model that predicts the location, intensity, and evolution trend of the high-low pressure system at future time steps.

5. The method for identifying and tracking high and low voltage systems according to any one of claims 1 to 3, characterized in that, The calculation of the coverage area and affected region of the high and low voltage system based on the N-ary tree hierarchical structure includes: Traverse the N-ary tree and take the union of the smallest bounding rectangles of all contour lines to obtain the coverage area of ​​the high and low voltage system. By combining wind field data, temperature data, and humidity data, the influence range of the high and low pressure system is calculated.

6. A device for identifying and tracking high and low pressure systems in a weather map, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method for identifying and tracking high and low pressure systems in a weather chart according to any one of claims 1-5.

7. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the method for identifying and tracking high and low pressure systems in a weather chart according to any one of claims 1-5.

Images