CN112924974B - A method, device and electronic equipment for identifying cloud clusters using DBSCAN clustering algorithm - Google Patents

Abstract

The invention discloses a method for identifying cloud clusters by using a DBSCAN clustering algorithm, comprising the following steps: obtaining the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; according to the difference between the reflectivity factors of different cloud clusters Different echo thresholds are set; according to the reflectivity factor data and the echo threshold, a DBSCAN algorithm model for identifying cloud clusters is established; cloud clusters are classified and identified according to the DBSCAN algorithm model; the present invention can identify cloud clusters Carry out classification and identification to make the classification results of cloud clusters more intuitive, and the edge, size and type of cloud clusters are obvious.

Description

A method, device and electronic equipment for identifying cloud clusters using DBSCAN clustering algorithm

technical field

The invention relates to the field of atmospheric science research, in particular to a method, device and electronic equipment for identifying cloud clusters using a DBSCAN clustering algorithm.

Background technique

Rainstorm is one of the main disastrous weather phenomena in our country, which is very harmful to the economy and society of our country. Rainstorms in my country are characterized by suddenness, persistence and frequency. Effective and real-time monitoring of precipitation is one of the powerful means to improve the ability of rainstorm warning. The rainfall system is generally cumulus mixed cloud precipitation, large stratiform cloud precipitation is inlaid with convective cloud precipitation, and the convective cloud area is the main cause of the strong convective weather process, and the stratiform cloud area controls the duration and duration of the heavy rain. precipitation. Because convective clouds and stratiform clouds have different generation mechanisms, changes in generation and dissipation, and moving speeds, and they contribute differently to changes in heat in the atmosphere, so identifying and judging convective clouds and stratiform clouds in heavy rains will help us understand more clearly. A good understanding of the mechanism of precipitation, improvement of precipitation estimation capabilities, monitoring and early warning of disastrous weather, aerospace and artificial weather modification operation command are of great help.

So far, domestic and foreign meteorologists have done a lot of research on the identification of convective clouds and stratiform clouds. Early identification of stratiform cloud precipitation was mostly done through zero-degree bright bands, but this method has certain limitations, that is, stratiform clouds must be developed to a relatively mature stage before they can be more accurately identified. There are also many methods based on rain gauge data. They will set a reflectivity factor threshold. In areas with precipitation echoes, as long as the echo reaches the set threshold, it will be judged as convective clouds. It is a stratiform cloud. This technique is called Background-Exceedence Technique (BET). BET technology can generally be used to determine the center of convective cloud precipitation, but the disadvantage is that it cannot determine the edge range of convective cloud precipitation.

Contents of the invention

(1) Purpose of the invention

The object of the present invention is to provide a method, device and electronic equipment for identifying cloud clusters using the DBSCAN clustering algorithm, which can classify and identify cloud clusters, so that the cloud cluster classification results are more intuitive, and the cloud cluster edge, size and type are obvious.

(2) Technical solution

In order to solve the above problems, on the one hand, the present invention provides a kind of method utilizing DBSCAN clustering algorithm to identify cloud clusters, comprising the following steps: obtaining the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; The difference between group reflectivity factors sets different echo thresholds; according to the reflectivity factor data and the echo threshold, set up the DBSCAN algorithm model for identifying cloud clusters; classify cloud clusters according to the DBSCAN algorithm model identify.

Optionally, the acquisition of reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle includes: for convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud cluster, outputting the second single-layer elevation angle Albedo factor data; the albedo factor data of the second single-layer elevation angle is processed to obtain the albedo factor data of the first monolayer elevation angle; convective clouds and multi-layer stratiform clouds that are vigorously developed in the cloud cluster , output the reflectivity factor data of the second multi-layer elevation angle; process the reflectivity factor data of the second multi-layer elevation angle to obtain the reflectivity factor data of the first multi-layer elevation angle; Clouds are cloud clusters that are less than 6km from the ground, and vigorous convective clouds are cloud clusters that are 6-8km from the ground.

Optionally, the reflectivity factor data of the second single-layer elevation angle is all reflectivity factor values within the first preset value; the reflectivity factor data of the second multi-layer elevation angle is within the second preset value All reflectance factor values.

Optionally, the echo threshold is set to Z≥45dBZ, 37dBZ≤Z<45dBZ, 30dBZ≤Z<37dBZ, 25dBZ≤Z<30dBZ, corresponding to strong convective clouds, weak convective clouds, strong stratiform clouds, weak Stratiform clouds; where Z is the echo threshold.

Optionally, according to the reflectivity factor data and the echo threshold, establishing a DBSCAN algorithm model for identifying clouds includes: according to the reflectivity factor data of the first single-layer elevation angle and the echo threshold, for The convective clouds and single-layer stratiform clouds that are not vigorously developed are modeled with the DBSCAN algorithm; according to the reflectivity factor data of the first multi-layer elevation angle and the echo threshold, the convective clouds that are vigorously developed and the multilayer clouds are Stratostratiform clouds were modeled with the DBSCAN algorithm.

Optionally, according to the reflectivity factor of the first single-layer elevation angle and the echo threshold, the DBSCAN algorithm modeling is performed on convective clouds and single-layer stratiform clouds that are not vigorously developed, including: DBSCAN passes the checklist The number of echo points contained in the third preset value neighborhood of each echo point in the layer elevation reflectivity factor data set is more than the fourth preset value, and the echo points exceed the set echo intensity threshold, create a cluster with one echo point as the core object.

Optionally, performing DBSCAN algorithm modeling on the vigorously developing convective clouds and multi-layer stratiform clouds according to the reflectivity factor of the first multi-layer elevation angle and the echo threshold value includes: The elevation angle is divided into a plurality of distance segments corresponding to the multi-layer elevation angle according to the number of distance libraries; according to the reflectivity factor data of the first multi-layer elevation angle, a plurality of the distance segments and the echo threshold value, the DBSCAN algorithm is constructed mold.

Optionally, performing DBSCAN algorithm modeling according to the reflectivity factor data of the first multi-layer elevation angle, multiple distance segments and the echo threshold, including: DBSCAN simultaneously checks each echo at the corresponding position in the multi-layer elevation angle The fifth preset value neighborhood of wave points is used to search clusters, if the number of echo points contained in the fifth preset value neighborhood of position G is more than the number of MinPts, and the echo points exceed the set echo points Intensity threshold, create a cluster with G as the core object; position G is the position of the same distance library at the elevation angle of each layer.

On the other hand, the present invention also provides a device for identifying clouds using the DBSCAN clustering algorithm, including: an acquisition module that acquires the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; an echo threshold setting module, Different echo thresholds are set according to the difference between different cloud group reflectivity factors; Model module, according to described reflectivity factor data and described echo threshold value, sets up the DBSCAN algorithm model of identifying cloud group; Identification module, according to instituted The above DBSCAN algorithm model is used to classify and identify cloud clusters.

Optionally, the acquisition module includes a single-layer elevation angle processing unit and a multi-layer elevation angle processing unit; the single-layer elevation angle processing unit outputs a second single-layer elevation angle to convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud cluster. The reflectivity factor data of the layer elevation angle, and the reflectivity factor data of the second single-layer elevation angle are processed to obtain the reflectivity factor data of the first single-layer elevation angle; the multi-layer elevation angle processing unit is used for cloud clusters For convective clouds and multi-layer stratiform clouds that are developing vigorously, output the reflectivity factor data of the second multi-layer elevation angle, and process the reflectivity factor data of the second multi-layer elevation angle to obtain the first multi-layer elevation angle The reflectivity factor data for .

Optionally, the model module includes a single-layer elevation model unit and a multi-layer elevation model unit; the single-layer elevation model unit, according to the reflectivity factor data of the first single-layer elevation and the echo threshold, The convective clouds and single-layer stratiform clouds that are not developing vigorously are modeled by the DBSCAN algorithm; the multi-layer elevation angle model unit, according to the reflectivity factor data of the first multi-layer elevation angle and the echo threshold, calculates the The DBSCAN algorithm is used to model the vigorously developed convective clouds and multi-layered stratiform clouds.

On the other hand, the present invention also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are realized.

On the other hand, the present invention also provides an electronic device, including a memory, a display, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the program, Steps to implement the method described above.

(3) Beneficial effects

The technical solution of the present invention has the following beneficial technical effects:

Compared with the existing cloud group identification algorithm, the design of the present invention is scientific and effective. The present invention utilizes the radar data combined with the DBSCAN clustering algorithm, so that the present invention can effectively classify and identify convective clouds and stratiform clouds, and further improve the precipitation in my country's meteorological operations. Estimation ability, the present invention is more applicable to the clouds detected by the weather radar in our country, the data source used is universal, and can meet the requirements of popularization and use by the meteorological business department in our country.

The present invention also comprehensively considers the differences in the reflectivity factors of different cloud groups (strong convective clouds, weak convective clouds, strong stratiform clouds, and weak stratiform clouds), extracts relevant thresholds, and designs a method capable of classifying and identifying convective clouds and stratiform clouds. The method provides an objective tool for the effective identification of different cloud clusters.

In the process of clustering clusters of the same type of cloud, the invention does not need to specify the cluster center and the number of clusters, and can find clusters of any shape in the space with invalid echoes, and preferentially identify convection from the cloud according to the density Clouds and stratiform clouds have good application prospects in the analysis and prediction of precipitation, thunderstorms and other applications.

The present invention uses the radar reflectivity factor combined with the DBSCAN clustering algorithm in the machine learning algorithm to classify and identify the cloud clusters, the cloud cluster classification results are more intuitive, and the cloud cluster edge, size and type are obvious.

Description of drawings

Fig. 1 is a flowchart of the present invention;

Figure 2 is a schematic diagram of the DBSCAN clustering algorithm for identifying strong convective clouds at an elevation angle.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in combination with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are exemplary only, and are not intended to limit the scope of the present invention. Also, in the following description, descriptions of well-known structures and techniques are omitted to avoid unnecessarily obscuring the concept of the present invention.

In the description of the present invention, it should be noted that the terms "first", "second", and "third" are used for description purposes only, and should not be understood as indicating or implying relative importance.

Figure 2 is a schematic diagram of the DBSCAN clustering algorithm for identifying strong convective clouds at an elevation angle. In Figure 2, ①Z<45dBZ, ②Z≥45dBZ, A, B: the core point of the strong convective cloud, C: the boundary of the strong convective cloud, N: noise point.

As shown in Fig. 1 to Fig. 2, in an embodiment of the present invention, the present invention provides a kind of method utilizing DBSCAN clustering algorithm to identify cloud cluster, comprises the following steps:

Step S10: Obtain the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle;

Step S20: setting different echo thresholds according to the difference between different cloud reflectivity factors;

Step S30: Establishing a DBSCAN algorithm model for identifying cloud clusters according to the reflectivity factor data and the echo threshold;

Step S40: classify and identify the cloud cluster according to the DBSCAN algorithm model.

Compared with the existing cloud group identification algorithm, the design of the present invention is scientific and effective. The present invention utilizes the radar data combined with the DBSCAN clustering algorithm, so that the present invention can effectively classify and identify convective clouds and stratiform clouds, and further improve the precipitation in my country's meteorological operations. Estimation ability, the present invention is more applicable to the clouds detected by the weather radar in our country, the data source used is universal, and can meet the requirements of popularization and use by the meteorological business department in our country. The radar data refers to the weather radar reflectivity factor data.

The present invention is applicable to various types of weather radar data such as CINRAD/SA, CINRAD/SB, CINRAD/CC in the Chinese meteorological system observation system, utilizes the method of the present invention to realize the identification of convective cloud and stratiform cloud, belongs to the field of atmospheric science research, Used for the identification of different cloud groups.

In the embodiment of the present invention, in order to ensure that convective clouds of different heights and stratiform clouds of different layers are identified, two processings are performed on the weather radar reflectivity factor. That is to obtain the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle, including: for convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud cluster, output the reflectivity factor data of the second single-layer elevation angle; Process the reflectivity factor data of the second single-layer elevation angle to obtain the reflectivity factor data of the first single-layer elevation angle; for the vigorously developed convective clouds and multi-layer stratiform clouds in the cloud cluster, output the reflection of the second multi-layer elevation angle The reflectivity factor data of the elevation angle of the second multi-layer is processed to obtain the reflectivity factor data of the elevation angle of the first multi-layer. In this embodiment, the reflectivity factor data of the second single-layer elevation angle is all reflectivity factor values within the first preset value; the reflectivity factor data of the second multi-layer elevation angle is all reflectivity within the second preset value rate factor value.

Wherein, the convective clouds that do not develop vigorously are cloud clusters that are less than 6 km from the ground, and the convective clouds that develop vigorously are cloud clusters that are 6 to 8 km away from the ground; Group, the convective cloud that develops exuberantly is the cloud group that is apart from ground 7km; 6km or 8km clouds. In this embodiment, the first preset value and the second preset value are both set to 230km, in other embodiments, the first preset value and the second preset value can also be set to 190km, 200km, 210km , 220km, of course, the first preset value and the second preset value can be set according to measurement requirements.

In the embodiment of the present invention, the echo threshold is set to Z≥45dBZ, 37dBZ≤Z<45dBZ, 30dBZ≤Z<37dBZ, 25dBZ≤Z<30dBZ, respectively corresponding to strong convective clouds, weak convective clouds, and strong stratiform clouds , Weak stratiform clouds; where Z is the echo threshold. In this embodiment, the echo threshold is set according to the difference in echo intensity between convective clouds and stratiform clouds, so as to achieve better modeling. In an embodiment of the present invention, according to the reflectivity factor data of the elevation angle and the echo threshold value, the DBSCAN algorithm is performed on the cloud group to establish a model, including: according to the reflectivity factor data and the echo threshold value of the single layer elevation angle, The DBSCAN algorithm is used to model the underdeveloped convective clouds and single-layer stratiform clouds; according to the reflectivity factor data and echo threshold of the multi-layer elevation angle, the DBSCAN algorithm is used to model the vigorously developed convective clouds and multi-layer stratiform clouds.

In the present invention, the reflectivity factor data of the first single-layer elevation angle is illustrated as an embodiment of the present invention with the elevation angle reflectivity factor data of 0.50°; the reflectivity factor data of the first multi-layer elevation angle is 0.50°, The reflectivity factor data of elevation angles of 1.45° and 2.40° are illustrated as an embodiment of the present invention.

In an embodiment of the present invention, according to the reflectivity factor and the echo threshold value of the first single-layer elevation angle, the convective cloud and the single-layer layered cloud that are not vigorously developed are carried out to the DBSCAN algorithm modeling, including: DBSCAN by checking the single-layer elevation angle If the third preset value neighborhood of each echo point in the reflectivity factor data set contains more echo points than the fourth preset value, and the echo points exceed the set echo intensity threshold, then create A cluster with an echo point as the core object. In this embodiment, the third preset value is set to 1 km, and in other embodiments, the third preset value can also be set to 2 km, 3 km, 4 km, etc. In this embodiment, the fourth preset value is set to 5, and in other embodiments, the fourth preset value can also be set to 3, 4, 6, 7, 8, 9, etc. .

The present invention carries out the DBSCAN algorithm modeling to the convective cloud and the single-layer stratiform cloud which are not vigorously developed according to the reflectivity factor and the echo threshold value of the first single-layer elevation angle, and illustrates with examples:

(1) DBSCAN randomly selects an echo point A (as shown in Figure 2) from the processed 0.50° elevation angle reflectivity factor data, and searches for at least 5 echo points greater than or equal to 45dBZ within 1km of the neighborhood of echo point A , mark echo point A as the core point of strong convective cloud (if there are at least 5 echo points greater than or equal to 37dBZ and less than 45dBZ within 1km of the neighborhood of echo point A, mark echo point A as the core point of weak convective cloud If there are at least 5 echo points greater than or equal to 30dBZ and less than 37dBZ within 1km of the neighborhood of echo point A, mark echo point A as the core point of strong stratiform clouds. If there are at least 5 echo points within 1km of the neighborhood of echo point A 5 echo points greater than or equal to 25dBZ and less than 30dBZ, mark echo point A as weak stratiform cloud core point);

(2) Then judge each echo point in the neighborhood of echo point A (assuming to judge echo point B), if it is satisfied that there are at least 5 echo points greater than or equal to 45dBZ within 1km of the neighborhood of echo point B, then Also mark this echo point as the core point of strong convective cloud (if there are at least 5 echo points greater than or equal to 37dBZ and less than 45dBZ within 1km of the neighborhood of echo point B, then mark this echo point as weak convection Cloud core point. If it is satisfied that there are at least 5 echo points greater than or equal to 30dBZ and less than 37dBZ within 1km of the neighborhood of echo point B, the echo point is also marked as a strong stratiform cloud core point. If the echo point B is satisfied If there are at least 5 echo points greater than or equal to 25dBZ and less than 30dBZ within 1km of the neighborhood of point B, this echo point is also marked as a weak stratiform cloud core point);

(3) Continue to judge the echo points in the neighborhood of echo point A (assuming to judge echo point C), if there are at least 5 echo points greater than or equal to 45dBZ within 1km of the neighborhood of echo point C, mark the echo point Wave point C is a strong convective cloud boundary point (if there are at least 5 points greater than or equal to 37dBZ and less than 45dBZ within 1km of the neighborhood of echo point C, mark echo point C as a weak convective cloud boundary point. If echo point C If there are at least 5 clouds greater than or equal to 30dBZ and less than 37dBZ within 1km of the neighborhood, mark echo point C as a strong stratiform cloud boundary point. If there are at least 5 clouds greater than or equal to 25dBZ within 1km of the neighborhood of echo point C and If it is less than 30dBZ, mark the echo point C as the weak convective cloud boundary point);

(4) If there is an echo point N that neither satisfies the conditions of the core point nor the boundary point, mark the echo point P as a noise point. When no new point is marked, the process ends.

In an embodiment of the present invention, according to the reflectivity factor and the echo threshold of the first multi-layer elevation angle, the DBSCAN algorithm modeling is carried out to the convective cloud and the multi-layer layer cloud which are vigorously developed, including: Divide multiple distance segments corresponding to the multi-layer elevation angle; perform DBSCAN algorithm modeling according to the reflectivity factor data of the first multi-layer elevation angle, multiple distance segments and the echo threshold. In an embodiment of the present invention, the elevation angles of 0.50°, 1.45°, and 2.40° are divided into three distance sections of 0-120km, 120-180km, and 180-230km according to the number of distance banks.

In an embodiment of the present invention, the DBSCAN algorithm modeling is carried out according to the reflectivity factor data of the first multi-layer elevation angle, multiple distance segments and echo thresholds, including: DBSCAN simultaneously checks each echo point at the corresponding position in the multi-layer elevation angle If the number of echo points contained in the fifth preset value neighborhood of position G is more than the number of MinPts, and the echo points exceed the set echo intensity threshold , then create a cluster with G as the core object; where the position G is the position of the same distance library at the elevation angle of each layer. In this embodiment, the fifth preset value is set to 1km, and in other embodiments, the fifth preset value can also be set to 2km, 3km, 4km, etc.

The present invention carries out DBSCAN algorithm modeling according to the reflectivity factor data of the first multi-layer elevation angle, multiple distance segments and echo threshold, and illustrates with examples:

(1) Divide the three-layer elevation angles into three distance segments of 0-120km, 120-180km, and 180-230km according to the number of distance banks, and use the reflectivity factor data of the three-layer elevation angles of 0.50°, 1.45°, and 2.40°, three Each distance segment and echo threshold are modeled by DBSCAN algorithm respectively.

(2) DBSCAN randomly selects an echo point A in the reflectivity factor data of the three-layer elevation angle (the three layers correspond to the same position), and searches for the echo point A adjacent to the corresponding position in the three-layer elevation angle in the 0-120km distance segment There are at least 10 echo points greater than or equal to 45dBZ within 1km, and echo point A is marked as the core point of strong convective clouds (if there are at least 10 echo points greater than or equal to 37dBZ and less than 45dBZ within 1km of echo point A neighborhood Echo point, mark echo point A as weak convective cloud core point; if there are at least 10 echo points greater than or equal to 30dBZ and less than 37dBZ within 1km of echo point A neighborhood, mark echo point A as strong Stratiform cloud core point; if there are at least 10 echo points greater than or equal to 25dBZ and less than 30dBZ within 1km of the neighborhood of echo point A, mark echo point A as a weak stratiform cloud core point).

There are at least 8 echo points in the 1km neighborhood of the 120-180km distance search echo point, and at least 5 echo points in the 180-230km distance search 1km neighborhood.

(3) Then judge each echo point in the neighborhood of echo point A at the corresponding position in the elevation angle of the third layer (assuming that echo point B is judged). If there are at least 10 echo points greater than or equal to 45dBZ, the echo point will also be marked as the core point of the strong convective cloud (if there are at least 10 echo points greater than or equal to 37dBZ and less than 45dBZ within 1km of the neighborhood of echo point B). If there are at least 10 echo points greater than or equal to 30dBZ and less than 37dBZ within 1km of the neighborhood of echo point B, the echo point will also be marked as the core point of weak convective clouds. The point is the core point of strong stratiform cloud; if there are at least 10 echo points greater than or equal to 25dBZ and less than 30dBZ within 1km of the neighborhood of echo point B, this echo point is also marked as the core point of weak stratiform cloud ); In the 120-180km distance section, there are at least 8 echo points in the 1km neighborhood, and in the 180-230km distance section, there are at least 5 echo points in the 1km neighborhood.

(4) Continue to judge the echo point in the neighborhood of echo point A at the corresponding position in the elevation angle of the third layer (assuming that echo point C is judged). In the distance range of 0-120km, if the neighborhood of echo point C within 1km does not meet the requirement of at least If there are 10 echo points greater than or equal to 45dBZ, mark echo point C as a strong convective cloud boundary point (if there are at least 10 echo points greater than or equal to 37dBZ and less than 45dBZ within 1km of echo point C, mark the echo point C is the boundary point of weak convective cloud; if there are at least 10 points greater than or equal to 30dBZ and less than 37dBZ within 1km of the neighborhood of echo point C, mark echo point C as the boundary point of strong stratiform cloud; if echo point C is adjacent to If there are at least 10 points greater than or equal to 25dBZ and less than 30dBZ within 1km, mark the echo point C as the weak convective cloud boundary point).

In the 120-180km distance segment, there are at least 8 echo points in the 1km neighborhood, and in the 180-230km distance segment, there are at least 5 echo points in the 1km neighborhood.

(5) If there is an echo point N that neither satisfies the conditions of the core point nor the boundary point, mark the echo point P as a noise point. When no new point is marked, the process ends.

In the embodiment of the present invention, the cloud group is classified and identified according to the model, specifically, according to the built model, the cloud group can be identified as a strong convective cloud, a weak convective cloud, a strong stratiform cloud, and a weak stratiform cloud. Class, the model can identify the core area and edge area of each cloud class.

In an embodiment of the present invention, the present invention also provides a device for identifying cloud clusters utilizing the DBSCAN clustering algorithm, including an acquisition module that acquires the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; echo threshold setting The determination module sets different echo thresholds according to the difference between the reflectivity factors of different cloud clusters; the model module establishes the DBSCAN algorithm model for identifying cloud clusters according to the reflectivity factor data and echo threshold values; the identification module uses the DBSCAN algorithm The model classifies and identifies cloud clusters.

In the embodiment of the present invention, the acquisition module includes a single-layer elevation angle processing unit and a multi-layer elevation angle processing unit. Wherein the single-layer elevation angle unit, to the convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud cluster, output the reflectivity factor data of the second single-layer elevation angle, and carry out the reflectivity factor data of the second single-layer elevation angle processing to obtain the reflectivity factor data of the elevation angle of the first single layer. Wherein the multilayer elevation angle unit, to the convective cloud and multilayer stratiform cloud that develops exuberantly in the cloud group, output the reflectivity factor data of the second multilayer elevation angle, and process the reflectivity factor data of the second multilayer elevation angle , to obtain reflectivity factor data of the first multi-layer elevation angle.

In the embodiment of the present invention, the echo threshold of the echo threshold setting module is set to Z≥45dBZ, 37dBZ≤Z<45dBZ, 30dBZ≤Z<37dBZ, 25dBZ≤Z<30dBZ, corresponding to strong convective clouds, weak Convective clouds, strong stratiform clouds, weak stratiform clouds; where Z is the echo threshold.

In an embodiment of the present invention, the single-layer elevation angle model unit performs DBSCAN algorithm modeling on convective clouds and single-layer stratiform clouds that are not vigorously developed according to the reflectivity factor data of the first single-layer elevation angle and the echo threshold .

In the embodiment of the present invention, the modeling process of the single-layer elevation model unit: DBSCAN checks that the third preset value neighborhood of each echo point in the single-layer elevation angle reflectivity factor data set contains more echo points than the fourth preset value, and the echo point exceeds the set echo intensity threshold, a cluster with one echo point as a core object is created.

In an embodiment of the present invention, the multi-layer elevation angle model unit performs DBSCAN algorithm modeling according to the reflectivity factor data of the first multi-layer elevation angle, the plurality of distance segments and the echo threshold. The concrete multi-layer elevation angle model unit is to first divide the multi-layer elevation angle into a plurality of distance segments corresponding to the multi-layer elevation angle according to the distance library number; then according to the reflectivity factor data of the first multi-layer elevation angle, a plurality of distance segments and and The echo threshold is modeled by the DBSCAN algorithm.

The modeling process of the multi-layer elevation model unit: DBSCAN simultaneously checks the fourth preset value neighborhood of each echo point corresponding to the multi-layer elevation angle to search for clusters, if the fifth preset value neighborhood of the position G contains The number of echo points in is greater than the number of MinPts, and the echo points exceed the set echo intensity threshold, then a cluster with G as the core object is created.

In an embodiment of the present invention, the present invention also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented.

In an embodiment of the present invention, the present invention also provides an electronic device, including a memory, a display, a processor, and a computer program stored in the memory and operable on the processor, and the processor executes the The steps of the above-mentioned method are realized when the above-mentioned program is described.

The present invention comprehensively considers the differences in the reflectivity factors of different cloud groups (strong convective clouds, weak convective clouds, strong stratiform clouds, and weak stratiform clouds), extracts relevant thresholds, and designs a system capable of classifying and identifying convective clouds and stratiform clouds. This method provides an objective tool for effective identification of different cloud clusters.

In the process of clustering clusters of the same type of cloud, the invention does not need to specify the cluster center and the number of clusters, and can find clusters of any shape in the space with invalid echoes, and preferentially identify convection from the cloud according to the density Clouds and stratiform clouds have good application prospects in the analysis and prediction of precipitation, thunderstorms and other applications.

The present invention uses the radar reflectivity factor combined with the DBSCAN clustering algorithm in the machine learning algorithm to classify and identify the cloud clusters, the cloud cluster classification results are more intuitive, and the cloud cluster edge, size and type are obvious.

It should be understood that the above specific embodiments of the present invention are only used to illustrate or explain the principle of the present invention, and not to limit the present invention. Therefore, any modification, equivalent replacement, improvement, etc. made without departing from the spirit and scope of the present invention shall fall within the protection scope of the present invention. Furthermore, it is intended that the appended claims of the present invention embrace all changes and modifications that come within the scope and metesques of the appended claims, or equivalents of such scope and metes and bounds.

Claims (10)

1. a method utilizing DBSCAN clustering algorithm to identify cloud clusters, is characterized in that, comprises the following steps: Obtain reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; Set different echo thresholds according to the difference between different cloud reflectivity factors; According to the reflectivity factor data and the echo threshold, a DBSCAN algorithm model for identifying cloud clusters is established; According to the DBSCAN algorithm model, cloud clusters are classified and identified; The acquisition of reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle includes: For convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud cluster, output the reflectivity factor data of the second single-layer elevation angle; Processing the reflectivity factor data of the second monolayer elevation angle to obtain the reflectivity factor data of the first monolayer elevation angle; For convective clouds and multi-layer stratiform clouds that develop vigorously in the cloud cluster, output the reflectivity factor data of the second multi-layer elevation angle; Processing the reflectivity factor data of the second multi-layer elevation angle to obtain the reflectivity factor data of the first multi-layer elevation angle; According to described albedo factor data and described echo threshold value, set up the DBSCAN algorithm model of identifying cloud cluster, comprising: According to the reflectivity factor data of the first single-layer elevation angle and the echo threshold, the DBSCAN algorithm modeling is performed on the convective clouds and single-layer stratiform clouds that are not vigorously developed; According to the reflectivity factor data of the first multi-layer elevation angle and the echo threshold, the DBSCAN algorithm modeling is performed on the vigorously developing convective clouds and multi-layer stratiform clouds. 2. The method of claim 1, wherein, The convective clouds that are not vigorously developed are cloud clusters that are less than 6km from the ground, and the convective clouds that are vigorously developed are cloud clusters that are 6-8km from the ground. 3. The method of claim 2, wherein, The reflectivity factor data of the second monolayer elevation angle is all reflectivity factor values within the first preset value; The reflectivity factor data of the second multi-layer elevation angle is all reflectivity factor values within the second preset value. 4. The method of claim 1, wherein, The echo threshold is set to Z≥45dBZ, 37dBZ≤Z<45dBZ, 30dBZ≤Z<37dBZ, 25dBZ≤Z<30dBZ, respectively corresponding to strong convective clouds, weak convective clouds, strong stratiform clouds, and weak stratiform clouds; Where Z is the echo threshold. 5. The method of claim 1, wherein, According to the reflectivity factor of the first single-layer elevation angle and the echo threshold, the DBSCAN algorithm modeling is carried out to convective clouds and single-layer stratiform clouds that are not vigorously developed, including: DBSCAN checks the number of echo points contained in the third preset value neighborhood of each echo point in the single-layer elevation reflectivity factor data set is more than the number of the fourth preset value, and the echo points exceed the set value. If the echo threshold value is specified, create a cluster with one echo point as the core object. 6. The method of claim 1, wherein, According to the reflectivity factor of the first multi-layer elevation angle and the echo threshold, the DBSCAN algorithm modeling is carried out to the vigorously developed convective clouds and multi-layer stratiform clouds, including: Dividing the multi-layer elevation angle into a plurality of distance segments corresponding to the multi-layer elevation angle according to the number of distance bins; The DBSCAN algorithm modeling is performed according to the reflectivity factor data of the first multi-layer elevation angle, the plurality of distance segments and the echo threshold. 7. The method of claim 6, wherein, Performing DBSCAN algorithm modeling according to the reflectivity factor data of the first multi-layer elevation angle, a plurality of the distance segments and the echo threshold, including: DBSCAN simultaneously checks the fifth preset value neighborhood of each echo point corresponding to the multi-layer elevation angle to search for clusters, if the number of echo points contained in the fifth preset value neighborhood of position G is greater than the number of MinPts, And the echo point exceeds the set echo threshold, then create a cluster with G as the core object; Position G is the position of the same range library for each layer elevation angle. 8. A cloud cluster device utilizing DBSCAN clustering algorithm, is characterized in that, comprising: An acquisition module that acquires the reflectivity factor data of the first single-layer elevation angle and the first multi-layer elevation angle; The echo threshold setting module sets different echo thresholds according to the difference between different cloud reflectivity factors; A model module, based on the reflectivity factor data and the echo threshold, establishes a DBSCAN algorithm model for identifying cloud clusters; Identification module, according to the DBSCAN algorithm model, cloud group is classified and identified; The acquisition module includes a single-layer elevation processing unit and a multi-layer elevation processing unit; The single-layer elevation angle processing unit outputs the reflectivity factor data of the second single-layer elevation angle for convective clouds and single-layer stratiform clouds that are not vigorously developed in the cloud group, and converts the reflectivity factor data of the second single-layer elevation angle Data is processed to obtain the reflectivity factor data of the first single-layer elevation angle; The multi-layer elevation angle processing unit outputs the reflectivity factor data of the second multi-layer elevation angle to the convective clouds and multi-layer stratiform clouds that are vigorously developing in the cloud cluster, and outputs the reflectivity factor data of the second multi-layer elevation angle Performing processing to obtain the reflectivity factor data of the first multi-layer elevation angle; The model module includes a single-layer elevation model unit and a multi-layer elevation model unit; The single-layer elevation model unit performs DBSCAN algorithm modeling on the underdeveloped convective clouds and single-layer stratiform clouds according to the reflectivity factor data of the first single-layer elevation angle and the echo threshold; The multi-layer elevation angle modeling unit performs DBSCAN algorithm modeling on the vigorously developing convective cloud and multi-layer stratiform cloud according to the reflectivity factor data of the first multi-layer elevation angle and the echo threshold. 9. A storage medium, wherein a computer program is stored on the storage medium, and when the program is executed by a processor, the steps of any one of the methods according to claims 1-7 are realized. 10. An electronic device, characterized in that it comprises a memory, a display, a processor, and a computer program stored on the memory and operable on the processor, and the processor implements the claims when executing the program The step of any one of the methods described in 1-7.

Images