A method for identifying and tracking convective cloud clusters from meteorological satellites

Abstract

The present invention discloses a method for identifying and tracking the boundaries of meteorological satellite convective clouds, comprising the following steps: reading meteorological satellite grid point data and representing it according to a certain data structure; formulating grid point encoding rules and effectively encoding each grid point; identifying and tracking each grid point according to an algorithm to ensure that it belongs to a storm cloud cluster; performing boundary scanning on the storm cloud cluster to obtain its boundary point set; the final result is a cloud cluster set containing several cloud clusters, each cloud cluster having a number, a boundary, and grid points that constitute the cloud cluster, and each grid point self-contained with spatial and physical properties. The meteorological satellite convective cloud cluster identification and boundary tracking method provided by the present invention is no longer limited to image grayscale recognition, but implements a feature recognition and boundary tracking method for a regular grid field; it can obtain full information data on the spatial and physical properties of each grid point, providing data support for refined data analysis.

Description

Technical Field

[0001] The present invention belongs to the technical field of meteorological monitoring, and in particular relates to a method for identifying and tracing the boundaries of convective clouds by a meteorological satellite. Background Art

[0002] Severe convective weather, such as thunderstorms, gales, hail, short bursts of heavy rainfall, and tornadoes, often causes significant damage to people's daily lives and production, and poses a significant threat to aviation safety. Severe convective weather, characterized by its small spatial scale, rapid onset and development, and high destructiveness, has always been a challenge in meteorological forecasting services and a pressing need for monitoring and warning of disastrous weather. Meteorological satellite monitoring data can provide large-scale, all-weather, multi-channel, high-temporal and spatial resolution detection information, greatly improving natural disaster monitoring capabilities and providing a wealth of useful information for weather forecasting and climate change monitoring. Satellite data offers irreplaceable advantages, particularly in areas where conventional observation data is insufficient, such as high mountains, plateaus, deserts, oceans, and polar regions, and for monitoring small and medium-scale severe convective weather, facilitating the tracking and warning of disastrous weather.

[0003] Existing technologies typically include OpenCV-based image grayscale tracking and image-based edge detection algorithms. Existing meteorological satellite cloud identification methods are mostly based on satellite imagery and employ thresholding for grayscale tracking. While these methods can capture the extent and movement of convective clouds, they are unable to precisely characterize the fine-grained physical properties within the clouds. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a method for identifying and tracking the boundaries of convective clouds using a meteorological satellite.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] A method for identifying and tracking convective cloud clusters by a meteorological satellite comprises the following steps:

[0007] S1 data processing: read meteorological satellite grid data and represent it according to a certain data structure;

[0008] S2 Grid Coding: Formulate grid coding rules and effectively encode each grid point;

[0009] S3 Cloud Identification: Identify and track each grid point based on an algorithm to determine its belonging to a storm cloud cluster;

[0010] S4 boundary tracking: Scan the boundary of the storm cloud to obtain its boundary point set;

[0011] S5 result output: The final result is a cloud cluster set, which contains several cloud clusters. Each cloud cluster has a number, a boundary, and grid points that make up the cloud cluster. Each grid point contains its own spatial attributes and physical attributes.

[0012] Furthermore, before step S2, it also includes establishing a grid attribute mapping: reading the grid data, storing the row and column values ​​of each grid point in the row and col fields of the data structure POINT, storing the longitude and latitude values ​​in the lon and lat fields, storing the grid point value in the value field, and setting the flag field to 0 to indicate that it has not been tracked.

[0013] Furthermore, the grid point encoding rules are as follows:

[0014] Encode the grid point P and set the pid field value to row*COLs+col, where row and col are the row and column values ​​of the grid point P respectively, and COLs is the total number of columns in the grid; arrange the pid in descending order by row and column number, ensuring that the grid point with the smallest pid has the smallest row number and the smallest column number on that row; after encoding, ensure that each grid point has a unique ID value.

[0015] Furthermore, in step S3, cloud cluster identification, a threshold filter is used to screen out all grid points that meet the conditions, and then cloud clusters are grouped. The specific steps are as follows:

[0016] S301 encodes all grid points according to the grid point encoding rule so that each grid point has a unique ID;

[0017] S302 sets a threshold, collects grid points that meet the conditions to form a grid point set, and arranges them in descending order of grid point ID size;

[0018] S303 defines the subscript i and initializes the group number gid=0;

[0019] S304 sets i=0; group number gid increases by 1;

[0020] S305: Take out the P with the smallest ID value and tracking flag 0 from the grid point set. i Grid, set P i The group tag GroupID=gid, tracking tag flag=1;

[0021] S306 in P i Search in the eight neighborhood directions and find the point P adjacent to it i+1 , set P i+1 The group tag GroupID=gid, tracking tag flag=1;

[0022] S307 subscript i increases by 1;

[0023] S308 repeats step S306. If no points meeting the conditions are found, the search for this group ends.

[0024] S309 repeats step S304. If all points have been tracked, the preliminary grouping is completed.

[0025] After the above steps, S310 obtains several grid point groups;

[0026] S311 checks whether there are adjacent grid point groups or groups sharing one or more identical grid points;

[0027] S312: if there are adjacent grid point groups, merge the adjacent grid point groups into one group, and the group number is the smaller value of the group numbers of the two groups;

[0028] S313 repeats step S311 until there are no more adjacent grid point groups, and the grouping process ends.

[0029] Furthermore, the specific steps of step S4 boundary tracking are as follows:

[0030] S401 finds the first valid grid point of a cloud cluster, where the grid point P0 has the minimum ID value in the cloud cluster;

[0031] S402 defines a direction variable dir, and sets dir=7, which indicates the previous moving direction from the previous boundary point to the current boundary point along the boundary; and detects the boundary according to the eight-neighborhood region;

[0032] S403 searches the eight neighborhoods of the current grid point in a counterclockwise direction to find the first point P adjacent to the current grid point. i , obtain the eight-neighborhood direction value pdir relative to the current grid point P0. When dir is an odd number, the value is (pdir+6) mod 8. When dir is an even number, the value is (pdir+7) mod 8. It is used to update dir;

[0033] S404 If P i The same position as P1, and P i-1 If the position is the same as P0, stop searching, otherwise repeat step S403;

[0034] The boundary searched in S405 consists of P0, P1, ..., P i-2 composition;

[0035] S406 Boundary scanning is completed, and a set of boundary points of each group / cloud is obtained.

[0036] Furthermore, in the result output of step S5, the final output result is a collection of data structures GROUP, where the number of GROUPs is the number of identified cloud clusters. According to the definition of the above data structure, each GROUP output result contains not only the row and column information and longitude and latitude information of the cloud cluster boundary, but also the attribute information of each grid point in the cloud cluster.

[0037] Beneficial Effects: The meteorological satellite convective cloud identification and boundary tracking method provided by this invention utilizes multi-channel data detected by geostationary meteorological satellites and employs computer-generated automated processing techniques based on appropriate algorithms to rapidly identify convective clouds and track their boundaries. This method can promptly detect the location, development intensity, and movement trajectory of severe convective clouds, a weather system capable of causing severe disasters, thereby providing early forecasts and warnings of severe convective weather. Furthermore, the method goes beyond image grayscale recognition and implements feature recognition and boundary tracking methods for regular grid fields. It can obtain comprehensive information on the spatial and physical properties of each grid point, providing data support for refined data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A method framework diagram of an embodiment of the present invention;

[0039] Figure 2 Schematic diagram of the spatial topology structure of eight neighborhoods of regular grid points in an embodiment of the present invention;

[0040] Figure 3 This is a flow chart of cloud cluster identification in an embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of boundary tracking in an embodiment of the present invention;

[0042] Figure 5 Flowchart of boundary tracking in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The present invention is described below with reference to specific examples. Those skilled in the art will appreciate that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention in any way.

[0044] like Figure 1 As shown, a method for identifying and tracking the boundaries of convective clouds by a meteorological satellite includes the following steps:

[0045] S1 data processing: read meteorological satellite grid data and represent POINT, GROUP, EDGE according to a certain data structure;

[0046] S2 Grid Coding: Formulate grid coding rules and effectively encode each grid point;

[0047] S3 Cloud Identification: Identify and track each grid point based on an algorithm to determine its belonging to a storm cloud cluster;

[0048] S4 boundary tracking: Scan the boundary of the storm cloud to obtain its boundary point set;

[0049] S5 result output: The final result is a cloud cluster set, which contains several cloud clusters. Each cloud cluster has a number, a boundary, and grid points that make up the cloud cluster. Each grid point contains its own spatial attributes and physical attributes.

[0050] Define the following data structure POINT to describe it:

[0051] typedef structPOINT

[0052] {

[0053] int row; / / row

[0054] int col; / / column

[0055] int pid; / / Point ID

[0056] int GroupID; / / Group ID

[0057] int EdgeID; / / Border ID

[0058] int flag; / / mark

[0059] float lon; / / longitude

[0060] float lat; / / latitude

[0061] float value; / / element value

[0062] }.

[0063] The following EDGE is defined to describe it:

[0064] typedef struct EDGE

[0065] {

[0066] int ID

[0067] list <point>points;

[0068] }.

[0069] Define the following data structure GROUP to describe it:

[0070] typedef struct GROUP

[0071] {

[0072] int ID;

[0073] EDGE edge;

[0074] list <point>points

[0075] }.

[0076] Before step S2, it also includes establishing a grid point attribute mapping: reading the grid point data, storing the row and column values ​​of each grid point in the row and col fields of the data structure POINT, storing the longitude and latitude values ​​in the lon and lat fields, storing the grid point value in the value field, and setting the flag field to 0 to indicate that it has not been tracked.

[0077] The grid point encoding rules are as follows:

[0078] Encode the grid point P and set the pid field value to row*COLs+col, where row and col are the row and column values ​​of the grid point P respectively, and COLs is the total number of columns in the grid; arrange the pid in descending order by row and column number, ensuring that the grid point with the smallest pid has the smallest row number and the smallest column number on that row; after encoding, ensure that each grid point has a unique ID value.

[0079] like Figure 2 The topological structure of the eight-neighborhood space of a regular grid point is shown. The figure shows the grid point distribution, using numbers 0 to 7 to represent eight directions. The grid (0, 0) represents the current grid point. The value in each direction grid represents the position of the grid point relative to the current grid point, which is the relative value of the row and column values ​​of the grid point and the current grid point.

[0080] like Figure 3 As shown, in step S3, in cloud cluster identification, a threshold filter is used to screen out all grid points that meet the conditions, and then cloud clusters are grouped. The specific steps are as follows:

[0081] S301 encodes all grid points according to the grid point encoding rule so that each grid point has a unique ID;

[0082] S302 sets a threshold, collects grid points that meet the conditions to form a grid point set, and arranges them in descending order of grid point ID size;

[0083] S303 defines the subscript i and initializes the group number gid=0;

[0084] S304 sets i=0; group number gid increases by 1;

[0085] S305: Take out the P with the smallest ID value and tracking flag 0 from the grid point set. i Grid, set P i The group tag GroupID=gid, tracking tag flag=1;

[0086] S306 in P i Search in the eight neighborhood directions and find the point P adjacent to it i+1 , set P i+1 The group tag GroupID=gid, tracking tag flag=1;

[0087] S307 subscript i increases by 1;

[0088] S308 repeats step S306. If no points meeting the conditions are found, the search for this group ends.

[0089] S309 repeats step S304. If all points have been tracked, the preliminary grouping is completed.

[0090] After the above steps, S310 obtains several grid point groups;

[0091] S311 checks whether there are adjacent grid point groups or groups sharing one or more identical grid points;

[0092] S312: if there are adjacent grid point groups, merge the adjacent grid point groups into one group, and the group number is the smaller value of the group numbers of the two groups;

[0093] S313 repeats step S311 until there are no more adjacent grid point groups, and the grouping process ends.

[0094] like Figure 4 The schematic diagram of boundary tracking is shown in the figure. In the figure, the square grid represents the valid data point, and the grid filled with shaded space represents the starting tracking grid point. According to the eight-neighborhood distribution, the numbers 0 to 7 are defined to represent the eight directions of the grid point.

[0095] like Figure 5 As shown, the specific steps of step S4 boundary tracking are as follows:

[0096] S401 finds the first valid grid point of a cloud cluster, where the grid point P0 has the minimum ID value in the cloud cluster;

[0097] S402 defines a direction variable dir, and sets dir=7, which indicates the previous moving direction from the previous boundary point to the current boundary point along the boundary; and detects the boundary according to the eight-neighborhood region;

[0098] S403 searches the eight neighborhoods of the current grid point in a counterclockwise direction to find the first point P adjacent to the current grid point. i , obtain the eight-neighborhood direction value pdir relative to the current grid point P0. When dir is an odd number, the value is (pdir+6) mod 8. When dir is an even number, the value is (pdir+7) mod 8. It is used to update dir;

[0099] S404 If P i The same position as P1, and P i-1 If the position is the same as P0, stop searching, otherwise repeat step S403;

[0100] The boundary searched in S405 consists of P0, P1, ..., P i-2 composition;

[0101] S406 Boundary scanning is completed, and a set of boundary points of each group / cloud is obtained.

[0102] In the result output of step S5, the final output result is a collection of data structures GROUP, where the number of GROUPs is the number of identified cloud clusters. According to the definition of the above data structure, each GROUP output result not only contains the row and column information and longitude and latitude information of the cloud cluster boundary, but also includes the attribute information of each grid point in the cloud cluster, such as row and column values, longitude and latitude, element values, etc., which can provide complete data support for refined data analysis.< / point> < / point>

Claims

1. A method for identifying and tracking convective cloud clusters using a meteorological satellite, characterized in that: The following steps are involved: S1 data processing: read meteorological satellite grid data and represent it according to a certain data structure; S2 Grid Coding: Formulate grid coding rules and effectively encode each grid point; S3 Cloud Identification: Identify and track each grid point based on an algorithm to determine its belonging to a storm cloud cluster; S4 boundary tracking: Scan the boundary of the storm cloud to obtain its boundary point set; S5 result output: The final result is a cloud cluster set, which contains several cloud clusters. Each cloud cluster has a number, a boundary, and grid points that make up the cloud cluster. Each grid point contains its own spatial and physical properties. Before step S2, the process also includes establishing a grid point attribute mapping: reading grid point data, storing the row and column values ​​of each grid point in the row and col fields of the data structure POINT, storing the latitude and longitude values ​​in the lon and lat fields, storing the grid point value in the value field, and setting the flag field to 0 to indicate that it has not been tracked; The grid point encoding rule is as follows: Encode the grid point P, set the pid field value to row*COLs+col, where row and col are the row and column values ​​of the grid point P respectively, and COLs is the total number of columns in the grid; Arrange pids in descending order of row and column numbers, ensuring that the grid point with the smallest pid has the smallest row number and the grid point above it has the smallest column number; After encoding, each grid point is guaranteed to have a unique ID value.

2. The method for identifying and tracking convective clouds by a meteorological satellite according to claim 1, wherein: In step S3, cloud cluster identification, a threshold filter is used to screen out all grid points that meet the conditions, and then cloud clusters are grouped. The specific steps are as follows: S301 encodes all grid points according to the grid point encoding rule so that each grid point has a unique ID; S302 sets a threshold, collects grid points that meet the conditions to form a grid point set, and arranges them in descending order of grid point ID size; S303 defines the subscript i and initializes the group number gid=0; S304 sets i=0; group number gid increases by 1; S305: Take out the P with the smallest ID value and tracking flag 0 from the grid point set. i Grid, set P i The group tag GroupID=gid, tracking tag flag=1; S306 in P i Search in the eight neighborhood directions and find the point P adjacent to it i+1 , set P i+1 The group tag GroupID=gid, tracking tag flag=1; S307 subscript i increases by 1; S308 repeats step S306. If no points meeting the conditions are found, the search for this group ends. S309 repeats step S304. If all points have been tracked, the preliminary grouping is completed. After the above steps, S310 obtains several grid point groups; S311 checks whether there are adjacent grid point groups or groups sharing one or more identical grid points; S312: if there are adjacent grid point groups, merge the adjacent grid point groups into one group, and the group number is the smaller value of the group numbers of the two groups; S313 repeats step S311 until there are no more adjacent grid point groups, and the grouping process ends.

3. The method for identifying and tracking convective clouds by a meteorological satellite according to claim 1, wherein: The specific steps of step S4 boundary tracking are as follows: S401 finds the first valid grid point of a cloud cluster, where the grid point P0 has the minimum ID value in the cloud cluster; S402 defines a direction variable dir, and sets dir=7, which indicates the previous moving direction from the previous boundary point to the current boundary point along the boundary; and detects the boundary according to the eight-neighborhood region; S403 searches the eight neighborhoods of the current grid point in a counterclockwise direction to find the first point P adjacent to the current grid point. i , obtain the eight-neighborhood direction value pdir relative to the current grid point P0. When dir is an odd number, the value is (pdir+6) mod 8. When dir is an even number, the value is (pdir+7) mod 8. It is used to update dir; S404 If P i The same position as P1, and P i-1 If the position is the same as P0, stop searching, otherwise repeat step S403; The boundary searched in S405 consists of P0, P1, ..., P i-2 S406: Boundary scanning is completed, and a set of boundary points of each group / cloud is obtained.

4. The method for identifying and tracking convective clouds by a meteorological satellite according to claim 1, wherein: In the result output of step S5, the final output result is a collection of data structures GROUP, where the number of GROUPs is the number of identified cloud clusters. According to the definition of the above data structure, each GROUP output result contains not only the row and column information and longitude and latitude information of the cloud cluster boundary, but also the attribute information of each grid point that makes up the cloud cluster.

Images