Convection System Detection and Tracking Method Based on Density Clustering and Radar Echo Top Height Image

Abstract

The present invention provides a method for detecting and tracking convective systems based on density clustering and radar echo top height images. The original image data is preprocessed, and the preprocessed image data is clustered by a density clustering algorithm to obtain convective cell data. The overlapping region of the convective cell data at the current moment and the previous moment is analyzed to obtain the evolution path of the convective cell. Among them, each convective cell data in the evolution path has a number, and each convective cell data inherits the number of the monomer with the largest intersection at the previous moment. The convective cell data with the same number at each moment is corrected, and each convective cell is re-assigned a different number. The trajectory of the convective cell is drawn, and the pixel coordinates are converted into longitude and latitude to obtain a data set for locating the convective cell. The beneficial effects of the present invention are as follows: the function of automatically detecting and tracking convective cells is realized, and the continuity of the generated data is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of meteorological monitoring, and particularly relates to a method for detecting and tracking convective systems based on density clustering and radar echo top height images. Background Art

[0002] Doppler radar is one of the main detection tools for mesoscale and small-scale severe weather, and has the advantages of high spatio-temporal resolution. The identification, tracking, and early warning of convective systems and convective cells based on radar have always been an important research direction for short-term and nowcasting. A convective system or convective storm is a spatially continuous region with a certain volume and intensity reaching a certain threshold. Representative storm identification and tracking algorithms are the TITAN algorithm proposed by the National Center for Atmospheric Research in the United States and the SCIT algorithm proposed by the National Severe Storms Laboratory in the United States. Although these two algorithms have been widely applied to actual operations, their applicable ranges have certain limitations. The TITAN algorithm is applicable to the identification and tracking of storm bands, and the SCIT algorithm is applicable to the identification and tracking of storm centroids.

[0003] In actual operations, however, convective systems are constantly changing both in form and intensity. There are characteristics such as splitting and merging in form, and strengthening and weakening in intensity. Due to these complex evolutions, traditional algorithms are difficult to achieve long-term and continuous identification and tracking of convective storms. For storms embedded in large-scale mixed precipitation echoes, there are often number jumps (for example, at time t, a total of 3 convective storms are identified, and when a new cell forms at time t + 1, the storm originally numbered 1 is assigned another number). This situation results in discontinuous identification of convective storms, and the identification results lose the significance of convective system classification.

[0004] The echo top height detected by Doppler weather radar represents the maximum height that the particles in the precipitation cloud detected by the radar can reach, which is a reflection of the strength of the updraft in the precipitation cloud and is closely related to the atmospheric stability. Therefore, it helps to understand the vertical structure of weather systems. In the past, meteorological researchers used a subjective screening method for the identification and tracking of convective storm positions, which has problems such as low efficiency and large subjective interference. There is still a lack of an automatic detection and tracking technology for convective systems based on radar echo top height images. Summary of the Invention

[0005] In view of this, the present invention aims to propose a method for detecting and tracking convective systems based on density clustering and radar echo top height images, in order to solve at least one of the above-mentioned partial technical problems.

[0006] To achieve the above object, the technical solution of the present invention is realized as follows:

[0007] Obtain the radar echo top height image as the original image data, preprocess the original image data, and perform clustering processing on the preprocessed image data through a density clustering algorithm to obtain convective cell data;

[0008] Analyze the overlapping area of the convective cell data at the current moment and the previous moment to obtain the evolution path of the convective cell; among them, each convective cell data in the evolution path has a number, and each convective cell data inherits the number of the monomer with the largest intersection at the previous moment;

[0009] Perform correction processing on the convective cell data with the same number at each moment, and reassign different numbers to each convective cell;

[0010] Draw the trajectory of the convective cell, convert the pixel coordinates to longitude and latitude to obtain a data set for positioning the convective cell, and position the convective cell according to the data set for positioning the convective cell.

[0011] Further, the process of preprocessing the original image data includes performing binary processing and morphological operations on the original image data; where:

[0012] The binary processing is specifically as follows: set a critical threshold and use a binary function to process the original image data, and judge whether the value corresponding to each pixel point in the original image data is higher than the critical threshold; if it is higher, mark the corresponding pixel point as white; otherwise, mark it as black;

[0013] The morphological operation is specifically as follows: fill in the missing values in the image by performing a closing operation, and reduce the noise in the original image data by performing an opening operation.

[0014] Further, the process of performing clustering processing on the preprocessed image data through a clustering algorithm includes:

[0015] According to a preset neighborhood radius, search the neighborhood of each pixel in the preprocessed image data to obtain areas with dense pixel points, divide the areas with dense pixel points into the same cluster, and exclude isolated outliers.

[0016] Further, the process of analyzing the overlapping area of the convective cell data at the current moment and the previous moment includes:

[0017] Construct a double-layer loop structure; among them, the outer loop of the double-layer loop structure traverses the convective cell data at the current moment; the inner loop of the double-layer loop structure traverses the convective cell data at the previous moment;

[0018] Obtain the intersection of the traversal results of the outer loop and the inner loop to get the intersecting pixel points, and infer the evolution path of the convective monomer data by comparing the number of intersecting pixel points.

[0019] Further, the process of inferring the evolution path of the convective monomer data by comparing the number of intersecting pixel points includes:

[0020] During the process of traversing the convective monomer data at the current moment, obtain the convective monomer data at the previous moment with the most intersecting pixel points with the convective monomer data at the current moment, and determine that the convective monomer data at the current moment is evolved from the convective monomer data at the previous moment;

[0021] The convective monomer data at the current moment inherits the number of the convective monomer data at the previous moment; wherein, the inheritance specifically means that the numbers of the convective monomer data at the previous moment and the current moment are the same.

[0022] Further, during the process that the convective monomer data at the current moment inherits the number of the convective monomer data at the previous moment, if multiple convective monomer data inherit the same number, perform the correction process on the convective monomer data with the same number.

[0023] Further, the process of performing the correction process on the convective monomer data with the same number at each moment and reassigning different numbers to each convective monomer includes;

[0024] Traverse all the convective monomer data within the same moment, merge the convective monomer data with the same number, identify and retain the number of the convective monomer data with the largest area, and modify the numbers of the remaining convective monomer data with the same number, and the modified numbers are not repeated.

[0025] Further, the process of drawing the trajectory of the convective monomer, converting the pixel coordinates to longitude and latitude, and obtaining the data set for positioning the convective monomer includes:

[0026] Draw the minimum bounding rectangle of each convective monomer data, and obtain the four corner coordinates of the minimum bounding rectangle. Convert the four corner coordinates from pixel points to longitude and latitude format to obtain the longitude and latitude coordinates of the convective monomer data. The longitude and latitude coordinates of multiple convective monomer data with the same number form a data set for positioning the convective monomer corresponding to the number.

[0027] Compared with the prior art, the convective system detection and tracking method based on density clustering and radar echo top height image of the present invention has the following beneficial effects:

[0028] 1) Take the storms in the development stage as the objects for identification and tracking. Continuously identify and track the storm areas based on the monitoring results of the echo top height, effectively excluding interfering clutter, thereby significantly improving the accuracy of convective cell detection and tracking. The use of this data not only enhances the purity of the signal but also provides clearer and more reliable information input for the tracking algorithm.

[0029] 2) Abandon the traditional algorithm based on the statistical prior experience of forecasters, and propose an objective algorithm based on DBSCAN density clustering in machine learning. Its core purpose is to accurately capture the movement path of storm cells and systematically record every detail of its path changes, which can provide more accurate and detailed geolocation data for inversion analysis, greatly enriching the understanding of the behavior patterns of convective cells. The maximum continuous time for the identification and tracking of storms can reach more than 3 hours.

[0030] 3) It has the advanced function of automatically identifying and tracking convective cells in the echo map, and can accurately correct these cells, including but not limited to: accurately identifying newly generated convective cells to ensure that newly formed cells can be monitored and recorded in a timely manner; continuously tracking existing convective cells and updating their status and position information in real time; identifying and recording the splitting and merging events of cells, providing key information for understanding complex meteorological phenomena. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0032] Figure 1 is a schematic flow chart of the convective system detection and tracking method based on density clustering and radar echo top height image according to the embodiment of the present invention;

[0033] Figure 2 is a schematic diagram of the corresponding relationship between the coordinates in the pixel space and the coordinates in the longitude and latitude space according to the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0035] The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0036] The convective system detection and tracking method based on density clustering and radar echo top height image includes the following steps:

[0037] In the field of hydrodynamics, convective cells are a phenomenon that occurs when there are density differences within a liquid or gas body. In image analysis, this phenomenon typically manifests as areas where pixel points with higher numerical values are concentrated. Specifically, in the application of the echo top height map, each pixel point in the map is associated with a specific numerical range (e.g., 0 to 21). According to the advice of meteorological experts, 9 is used as the critical value. Numerical values exceeding 9 are classified as high-value areas. The dense distribution of these high-value pixel points often represents the presence of convective cells, and the active areas of convective cells are more likely to trigger meteorological disasters. Therefore, it is necessary to obtain the radar echo top height image as the original image data and track the convective cells.

[0038] S1. Preprocess the original image data.

[0039] The process of step S1 includes performing binary processing and morphological operations on the original image data; where:

[0040] The binary processing is specifically as follows: set the critical threshold and use the binary function to process the original image data, and determine whether the numerical value corresponding to each pixel point in the original image data is higher than the critical threshold; if it is higher, mark the corresponding pixel point as white; otherwise, mark it as black.

[0041] The morphological operation is specifically as follows: fill the missing numerical values in the image by performing a closing operation, and reduce the noise in the original image data by performing an opening operation.

[0042] Preferably, a specific implementation process of step S1 is as follows:

[0043] S11. At the technical level, first use binary operation to filter out the part below 9:

[0044] cv2.threshold(image, 9, 22, cv2.THRESH_BINARY);

[0045] Among them, cv2.threshold is the binary function in OpenCV, image is the original picture data, 9 is the critical value, 22 is the maximum value, and cv2.THRESH_BINARY means that values higher than the threshold are black and values lower than the threshold are white.

[0046] In the process of image recognition of convective cells, 9 is used as the critical threshold, the area below this threshold is marked as 0, and the area above this threshold is marked as 255.

[0047] Subsequently, perform morphological operations such as opening and closing operations on the preprocessed image, aiming to eliminate the density differences inside the cells to improve the recognition accuracy.

[0048] S12: From the perspective of image processing, it is expected that the internal density of each convective monomer is uniform. However, in the physical detection of the real world, existing technologies often cannot automatically process and correct this problem of uneven density.

[0049] Closing operation plays an important role here. It can fill the possible numerical gaps inside the monomer, thus achieving density homogenization.

[0050] cv2.morphologyEx(image, cv2.MORPH_CLOSE, kernel);

[0051] Among them, cv2.morphologyEx represents the function for performing morphological operations, image is the original image data, cv2.MORPH_CLOSE represents closing operation, and kernel is the kernel size.

[0052] Opening operation helps to reduce the interference of noise, further improve the image quality, and ensure the accurate identification and tracking of convective monomers.

[0053] cv2.morphologyEx(image, cv2.MORPH_OPEN, kernel);

[0054] Among them, cv2.MORPH_OPEN represents opening operation.

[0055] Through this carefully designed image processing process, this technology can effectively optimize the detection algorithm of convective monomers, improve its robustness and reliability under complex meteorological conditions, and optimize the image quality.

[0056] S2. Cluster the preprocessed image data through a density clustering algorithm to obtain convective monomer data.

[0057] The process of step S2 includes searching the neighborhood of each pixel in the preprocessed image data according to a preset neighborhood radius to obtain areas with dense pixel points, dividing the areas with dense pixel points into the same cluster, and excluding isolated outliers.

[0058] Preferably, a specific implementation process of step S2 is as follows:

[0059] Adopt the DBSCAN algorithm, select appropriate parameters, and perform preliminary clustering on the image data to identify and distinguish different convective monomer regions.

[0060] S21: The DBSCAN algorithm is a density-based clustering method that can perform preliminary clustering on image data.

[0061] There are differences in the sizes of image data from different regions. According to the sizes of different image data, the corresponding parameters eps and min_samples are selected. Taking the Bohai Rim region as an example: the size of the original image data is 801×701 pixels. When applying the DBSCAN algorithm, the parameter eps (neighborhood radius) is set to 50, and min_samples (the minimum number of samples required to become a "core point") is 800: DBSCAN(eps = 50, min_samples = 800);

[0062] The selection of these parameters aims to adapt to the range of the original data, ensure that the clustering process can accurately capture the density changes in the image, and maintain the accuracy of geographical coordinates. According to the specific characteristics of the data and the analysis objectives, these parameters can be appropriately adjusted to optimize the clustering results;

[0063] After grouping the pixel points on the image using the DBSCAN density clustering algorithm, this process will retain those pixel points representing the main clusters and simultaneously identify and exclude those isolated outliers, which are usually regarded as image noise;

[0064] The algorithm process can more accurately identify different convective cell regions in the image. In this way, the effective information in the image is refined, providing a clearer and more accurate dataset for further analysis and processing.

[0065] S3. Analyze the overlapping region of the convective cell data at the current moment and the previous moment to obtain the evolution path of the convective cell.

[0066] The specific process of step S3 includes: constructing a double-layer loop structure; among them, the outer loop of the double-layer loop structure traverses the convective cell data at the current moment; the inner loop of the double-layer loop structure traverses the convective cell data at the previous moment;

[0067] Obtain the intersection of the results traversed by the outer loop and the results traversed by the inner loop to get the intersecting pixel points, and infer the evolution path of the convective cell data by comparing the number of intersecting pixel points.

[0068] During the process of traversing the convective cell data at the current moment, obtain the convective cell data at the previous moment with the most intersecting pixel points with the convective cell data at the current moment, and determine that the convective cell data at the current moment is evolved from the convective cell data at the previous moment;

[0069] The convective cell data at the current moment inherits the number of the convective cell data at the previous moment; among them, the inheritance specifically means that the numbers of the convective cell data at the previous moment and the current moment are the same.

[0070] Preferably, a specific implementation process of the step S3 is as follows:

[0071] By analyzing the convective cell data at the previous moment and comparing the size of the overlapping area between the convective cell at the current moment and that at the previous moment, it can be inferred whether the convective cell in the current frame is inherited from the movement, splitting or merging of the cell at the previous moment, or appears as a new cell object.

[0072] This comparison involves not only the spatial positions of the convective cells, but also the changes in their shapes and scales, thus providing key information for understanding the dynamic evolution of the convective cells.

[0073] To efficiently manage and retrieve the information of convective cells, the present technology adopts a dictionary data structure to organize and store this data. In this structure, each cell is assigned a unique number, which serves as the key of the dictionary, and the corresponding value is a set containing the coordinates of all pixel points of the cell:

[0074] self.tracked_objects_cur = defaultdict(set);

[0075] Among them, self.tracked_objects_cur is a dictionary structure for storing the information of convective cells, and defaultdict(set) defines the type of the dictionary value.

[0076] This storage method of key-value pairs not only makes the access to the cell information direct and fast, but also facilitates subsequent processing and analysis, because all the relevant information of each cell is organized in a structured and easy-to-process format.

[0077] S31: To accurately track the evolution of convective cells, a double-loop structure is adopted: the outer loop traverses all the identified convective cells at the current moment, and the inner loop traverses all the identified convective cells at the previous moment. Then, the system calculates the intersection between each cell at the current moment and each cell at the previous moment one by one.

[0078] prev_rect.intersection(rect);

[0079] Among them, prev_rect represents a convective cell identified by the inner loop traversing the previous moment, rect represents a convective cell identified by the outer loop traversing the current moment, intersection is used to calculate the intersection of two sets, and the return value is all the elements within the intersection, corresponding to the intersecting pixel points here.

[0080] S32: In the analysis of the evolution of convective cells, it is assumed that the cell traversed at the current moment evolved from the cell at the previous moment with the largest intersection. This assumption applies to the cases of cell movement, splitting, and merging, where the spatial continuity of the cell is reflected by the size of the intersection. Specifically, by comparing the intersection areas of the current cell with each cell at the previous moment, the most likely evolution path can be inferred.

[0081] S33: When traversing cells at the current moment, if it is found that the cell does not overlap with any convective cells at the previous moment, this cell can be identified as newly generated.

[0082] S34: In the process of analyzing the evolution of convective cells, all cells except newly generated cells inherit the cell numbers of the cells with the largest intersection at the previous moment. At this stage, due to the possibility of cell splitting, there may be a situation where two different cells share the same number;

[0083] However, in the current step, these convective cells with duplicate numbers are not immediately renamed because these convective cells with duplicate numbers will have further effects in the subsequent step S4. Therefore, in step S4, these convective cells with duplicate numbers are renamed;

[0084] In step S4, how to handle these convective cells with duplicate numbers will be elaborated in detail, including the role in correcting the convective cell regions and how to rename them. After calculation using the convective cells, they are renumbered so that each cell has a unique identifier, thus providing clearer and more accurate data for subsequent analysis and tracking.

[0085] S35: When dealing with the complex situation of convective cell merging, the convective cell at the current moment will inherit the number of the cell with the largest area during the merging process at the previous moment. This strategy ensures that in the event of cell merging, the characteristics of the largest cell are retained, thus maintaining the coherence and traceability of the convective cell evolution history.

[0086] S4: Correct the convective cell data with the same number at each moment and reassign different numbers to each convective cell.

[0087] The specific steps of step S4 include: in the process of the convective cell data at the current moment inheriting the numbers of the convective cell data at the previous moment, if multiple convective cell data inherit the same number, the correction process is performed on the convective cell data with the same number;

[0088] The correction process is as follows: traverse all convective cell data at the same moment, merge the convective cell data with the same number, identify and retain the number of the convective cell data with the largest area, modify the numbers of the remaining convective cell data with the same number, and the modified numbers are not repeated.

[0089] Preferably, a specific implementation process of step S4 is as follows:

[0090] S41: In step S3, the present invention has obtained the information of all convective cells at the current moment and stored it in self.tracked_objects_cur. Using the same method, obtain the information of all convective cells at the next moment and store it in the variable self.tracked_objects_after. At the same time, in the previous calculation, the convective cell information at the previous moment was recorded in self.tracked_objects_before.

[0091] S42: Compare and correct the convective cells with the same number at the previous and current moments. For example, when there is only 1 convective cell with number k at the previous moment, 2 convective cells with number k at the current moment, and 1 convective cell with number k at the next moment, merge the 2 convective cells with number k at the current moment.

[0092] After self.tracked_objects_cur is corrected, further process the convective cells with duplicate numbers. Among all the convective cells with duplicate numbers, identify the monomer with the largest area and decide to retain its original number to maintain the uniqueness and continuity of its identity. For the remaining monomers with duplicate numbers, the present invention will assign new, non-repeating numbers to them. This process ensures that the numbers of all monomers are unique, thus avoiding any potential confusion or errors.

[0093] S5. Draw the trajectory of the convective cell, convert the pixel coordinates to longitude and latitude, and obtain the data set for positioning the convective cell.

[0094] The specific steps of step S5 include: drawing the minimum bounding rectangle of each convective cell data, obtaining the four corner coordinates of the minimum bounding rectangle, converting the four corner coordinates from pixel points to longitude and latitude format, and obtaining the longitude and latitude coordinates of the convective cell data. The longitude and latitude coordinates are used to frame the range of the convective cell;

[0095] Among them, the longitude and latitude coordinates of the convective cell data with the same number form the data set for positioning the convective cell corresponding to the number.

[0096] The process of converting the four corner coordinates from pixel points to longitude and latitude format includes:

[0097] Taking the transformation of a certain point among the four corners as an example, the corresponding relationship between the coordinates in the pixel space and the coordinates in the longitude and latitude space is as follows Figure 2 shown. The image corresponds to a coordinate (a, b) in the pixel space. Assuming that the starting point of the entire longitude and latitude space corresponds to the coordinate (start x , starty), and the unit 1 in the pixel point space corresponds to δ degrees in the longitude and latitude space. Therefore, the coordinates (a, b) in the pixel point space are converted into the coordinates (a′, b′) in the longitude and latitude space through the following formula;

[0098] a' = start x + a * δ

[0099] b′ = start y + b * δ;

[0100] After the above coordinate transformation, the longitude and latitude coordinates of the convective cell data are obtained.

[0101] Preferably, a specific implementation process applied in the step S5 is as follows:

[0102] The specific steps of S5 are as follows:

[0103] S51: Traverse the convective cells at the current moment one by one. Each cell is associated with a unique key value, and this key value represents the set of pixel points of the cell. For the set of pixel points of each cell, calculate its minimum bounding rectangle to accurately determine its boundary and obtain the coordinates of the four corners of this rectangle.

[0104] S52: Use the minimum bounding rectangle information to draw the minimum bounding rectangle of the cell on the original image.

[0105] S53: To achieve the docking with the geographical location in the real world, convert the above coordinates from pixel points to the longitude and latitude format, ensuring the geographical positioning accuracy of the convective cells.

[0106] Finally, the present invention records in detail the key value of each convective cell and the longitude and latitude coordinates of its four corners, and locates the convective cell according to the dataset for locating the convective cell.

[0107] The above technical solution uses improved density clustering for identification, uses the information of the front and back frames to correct and track the convective cell information, uses improved density clustering for identification, performs morphological operations on the original data to improve the problems of uneven internal density and noise points in the convective cells in the image, and then selects appropriate parameters for density clustering.

[0108] The above technical solution uses the front and back frame information for fine correction and tracking. It not only synthesizes the convective cell data at the current moment and the previous moment, but also further incorporates the convective cell information at the next moment. Through the comprehensive application of this three-phase information, it can accurately determine the transiently split and merged cells and make necessary corrections, ensuring not only the continuity and integrity of the convective cell path, but also providing a solid foundation for subsequent in-depth analysis.

[0109] The data used in the above technical solution is the real-time updated echo top height map.

[0110] Those of ordinary skill in the art can realize that the units and method steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0111] In several embodiments provided in the present application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the above division of units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The above units may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present invention.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered by the scope of the claims and the description of the present invention.

[0113] The above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A convective system detection and tracking method based on density clustering and radar echo top height images, characterized in that, It includes the following steps: Obtain the radar echo top height image as the original image data, preprocess the original image data, and perform clustering processing on the preprocessed image data through the density clustering algorithm to obtain convective cell data; Analyze the overlapping area of the convective cell data at the current moment and the previous moment to obtain the evolution path of the convective cell; wherein, each convective cell data in the evolution path has a number, and each convective cell data inherits the number of the monomer with the largest intersection at the previous moment; Perform correction processing on the convective cell data with the same number at each moment, and reassign different numbers to each convective cell; Draw the trajectory of the convective cell, convert the pixel coordinates to longitude and latitude to obtain a data set for positioning the convective cell, and position the convective cell according to the data set for positioning the convective cell.

2. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 1, wherein The process of preprocessing the original image data includes performing binary processing and morphological operations on the original image data; wherein: The binary processing is specifically to set a critical threshold and use a binary function to process the original image data, and judge whether the value corresponding to each pixel point in the original image data is higher than the critical threshold; if it is higher, mark the corresponding pixel point as white; otherwise, mark it as black; The morphological operation is specifically to fill in the missing values in the image by performing a closing operation, and reduce the noise in the original image data by performing an opening operation.

3. The convective system detection and tracking method based on density clustering and radar echo top height images according to claim 1, characterized in that, The process of performing clustering processing on the preprocessed image data through the density clustering algorithm includes: According to the preset neighborhood radius, search the neighborhood of each pixel in the preprocessed image data to obtain areas with dense pixel points, divide the areas with dense pixel points into the same cluster, and exclude isolated outliers.

4. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 1, characterized in that The process of analyzing the overlapping area of the convective cell data at the current moment and the previous moment includes: Construct a double-layer loop structure; wherein, the outer loop of the double-layer loop structure traverses the convective cell data at the current moment; the inner loop of the double-layer loop structure traverses the convective cell data at the previous moment; Obtain the intersection of the outer loop traversal result and the inner loop traversal result to obtain the intersecting pixel points, and infer the evolution path of the convective cell data by comparing the number of intersecting pixel points.

5. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 4, wherein The process of inferring the evolution path of the convective cell data by comparing the number of intersecting pixel points includes: During the process of traversing the convective cell data at the current moment, obtain the convective cell data at the previous moment with the most intersecting pixel points with the convective cell data at the current moment, and determine that the convective cell data at the current moment is evolved from the convective cell data at the previous moment; The convective cell data at the current moment inherits the number of the convective cell data at the previous moment; wherein, the inheritance is specifically that the numbers of the convective cell data at the previous moment and the current moment are the same.

6. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 5, characterized in that: In the process that the convective cell data at the current moment inherits the numbers of the convective cell data at the previous moment, if multiple convective cell data inherit the same number, the correction process is performed on the convective cell data with the same number.

7. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 1, characterized in that The process of performing the correction process on the convective cell data with the same number at each moment and reassigning different numbers to each convective cell includes: Traverse all the convective cell data within the same moment, merge the convective cell data with the same number, identify and retain the number of the convective cell data with the largest area, modify the numbers of the remaining convective cell data with the same number, and the modified numbers are not repeated.

8. The convective system detection and tracking method based on density clustering and radar echo top height image according to claim 1, characterized in that The process of drawing the trajectory of the convective cell, converting the pixel coordinates into longitude and latitude, and obtaining the data set for positioning the convective cell includes: Draw the minimum bounding rectangle of each convective cell data, obtain the four corner coordinates of the minimum bounding rectangle, convert the four corner coordinates from pixel points into the longitude and latitude format, and obtain the longitude and latitude coordinates of the convective cell data, where the longitude and latitude coordinates are used to frame the range of the convective cell; Among them, the longitude and latitude coordinates of the convective cell data with the same number form a data set for positioning the convective cell corresponding to the number.

Images