A fast identification method for mesoscale vortices based on the construction of stream functions

Abstract

The present invention relates to a method for quickly identifying mesoscale vortices based on the construction of stream functions. The original data of mesoscale vortices in the identification area is input into a processing model for data cleaning, and then input into a flow velocity limitation filter and a flow velocity zero-variation filter for preliminary data screening. The vortex information after preliminary screening is input into a flow direction around the center filter for further data screening. The vortex information after further screening is input into a stream function filter, which corrects the vortex information after further screening, captures the shape, obtains the vortex shape, and calculates the centroid of the vortex shape as the vortex center. The error between the vortex center and the initial vortex center is compared to obtain the confirmed vortex information. The main body of this method is constructed in a layered structure based on the physical structure characteristics of vortices, so as to achieve the rapid fusion of the model, and realize the lightweight, easy deployment and easy migration of the model.

Description

Technical Field

[0001] The present invention relates to the technical field of marine environment monitoring, and in particular to a method for rapid identification of mesoscale vortices based on the construction of stream functions. Background Art

[0002] As a cross-scale marine dynamic phenomenon, mesoscale vortices have significant impacts on many marine-related activities such as marine fisheries and military operations. In the analysis of the marine environmental situation, the identification of mesoscale vortices has important theoretical and practical significance. Currently, existing vortex identification methods such as data statistics methods, curvature center methods, neural network methods, etc., can be classified into three categories in terms of theoretical essence, namely empirical identification and computer learning identification.

[0003] Empirical identification depends on human subjective decisions, and its dynamic adaptability in terms of identification efficiency and accuracy is poor; while machine learning and deep learning identification based on computer technology are limited in their application deployment scenarios due to high requirements for data volume, hardware facilities, long model training time, and poor reusability. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for rapid identification of mesoscale vortices based on the construction of stream functions to solve the problems encountered in the above background art.

[0005] To achieve the above purpose, the technical solution of the present invention is as follows:

[0006] A method for rapid identification of mesoscale vortices based on the construction of stream functions, the method comprising the following steps:

[0007] S1. Input the original data of mesoscale vortices in the identification area into a processing model for data cleaning;

[0008] S2. Input the mesoscale vortex data after data cleaning into a flow velocity limitation filter and a flow velocity zero change filter for preliminary data screening to obtain the vortex information after preliminary screening;

[0009] S3. Input the vortex information after preliminary screening into a flow direction around the center filter for data re-screening to obtain the vortex information after re-screening;

[0010] S4. Input the vortex information after re-screening into a stream function filter to correct the vortex information after re-screening;

[0011] S5. Capture the shape to obtain the vortex shape, and calculate the centroid of the vortex shape as the vortex center;

[0012] S6. Compare the error between the vortex center and the initial vortex center in the vortex information after primary screening. If the error is less than 2 km, directly output the vortex information; otherwise, update the coordinates of the potential vortex center to obtain the confirmed vortex information.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The method is mainly constructed in a layered structure based on the physical structure characteristics of the vortex, and mainly includes 5 process steps: cleaning, primary screening, secondary screening, correction, and confirmation. The vortex position (latitude and longitude), type, and shape are extracted from the flow field through a multi-stage filter. This method has the characteristics of being lightweight, easy to deploy, and easy to migrate. The method can be expanded as needed. By adding other vortex identification and positioning models in the vortex screening stage of this model, including but not limited to empirical, physical feature-based, and machine learning-based models, rapid fusion of the models can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The disclosure of the present invention will be described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0015] Figure 1 is a schematic diagram of the working process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] In order to make the technical means, creative features, achieved purposes, and effects of the present invention easy to understand, the present invention will be further described in detail below with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, and only illustrate the basic structure of the present invention in a schematic manner. Therefore, they only show the components related to the present invention.

[0017] According to the technical solution of the present invention, without changing the essential spirit of the present invention, those of ordinary skill in the art can propose various structural ways and implementation methods that can be mutually replaced. Therefore, the following detailed description of the embodiments and the accompanying drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or restriction on the technical solution of the present invention.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0019] As Figure 1 shown, a method for rapid identification of mesoscale vortices based on stream function construction, the method includes the following steps:

[0020] S1. Input the original data of mesoscale vortices in the identification area into the processing model for data cleaning.

[0021] Among them, the processing model includes a land mask model and an abnormal flow velocity smoothing model. In the land mask model, the land without flow field information is identified by special marks, and the grid points are sequentially moved to search and statistically calculate the flow velocity magnitude within 5 km of the grid points. In the abnormal flow velocity smoothing model, if the difference between the grid point flow velocity and the average flow velocity of the statistical area is 5 times the standard deviation of the flow velocity within the statistical area, the flow velocity of the grid point is considered an abnormal flow velocity, and the flow velocity of the grid point is replaced by the average flow velocity of the 8 neighboring grid points of the grid point.

[0022] In this step, through the data cleaning model, the input flow field data is standardized and the noise of the flow field data is filtered, which can avoid the interference of abnormal data on vortex identification.

[0023] S2. Input the mesoscale vortex data after data cleaning into the flow velocity limitation filter and the flow velocity zero change filter for preliminary data screening to obtain the vortex information after preliminary screening, and the vortex information after preliminary screening is the longitude of the vortex center and the latitude of the vortex center.

[0024] The flow velocity at the vortex center is significantly lower than that of the vortex arm. Based on this structural feature, a flow velocity local filter is designed. In the W-E direction, the flow direction of the V-direction flow is opposite on both sides of the vortex center. In the N-S direction, the flow direction of the U-direction flow is opposite on both sides of the vortex center. Based on this structural feature, a flow velocity zero change filter is designed. The flow velocity limitation filter sequentially moves the grid points to calculate the longitude and latitude positions of the local minimum flow velocity within the 24 neighboring grid points of the grid point, which is marked as the potential vortex center. The flow velocity zero change filter checks the changes in the flow direction in the N-S and W-E directions of the potential vortex center and filters out the vortices that do not conform to the vortex structural features.

[0025] In this step, the flow field data is filtered through the most basic features of the vortex structure, which can reduce the computational complexity of subsequent vortex identification.

[0026] S3. Input the vortex information after preliminary screening into the flow direction around the center filter for data re-screening to obtain the vortex information after re-screening, and the vortex information after re-screening is the vortex type, the basic contour of the vortex, the longitude of the vortex center, and the latitude of the vortex center;

[0027] The flow direction around the center filter performs a quadrant cycle according to the flow direction in the vortex, for example, a quadrant cycle change like "1→2→3→4→1". The flow direction around the center filter is used to calculate the quadrant where the flow direction of the outer edge grid points around the 3 grid points of the potential vortex center is located, and filters out the vortices that do not conform to the vortex structural features.

[0028] In this step, the flow field data is further filtered, which can quickly locate the center position and corresponding type of the vortex in the grid flow field background.

[0029] S4. Input the vortex information after re-screening into the stream function filter to correct the vortex information after re-screening;

[0030] Since the contour lines of the stream function of the flow field are the streamlines of the flow field, a stream function filter is designed based on this principle. Among them, the initial search radius R of the stream function filter is 10 km. Calculate the distribution of the stream function within the range of R of the potential vortex center, gradually increase the search radius, obtain the closed streamline in the stream function field that contains the initial vortex center, interpolate the flow velocities at the connection points of the closed streamline and take the average, arrange them according to the area size of the closed figure, and select the closed streamline with the largest area and non-decaying average flow velocity of the closed figure as the vortex shape.

[0031] This step uses the potential vortex center as the contour recognition entrance, which can quickly capture the shape of the vortex in the background of the grid flow field. And taking the average flow velocity of the connection points of the closed figure as the basis for selecting the vortex shape with such non-strict conditions can reduce the interference of minor anomalies in the local flow velocity on the capture of the vortex shape.

[0032] S5. Capture the shape to obtain the vortex shape, and calculate the centroid of the vortex shape as the vortex center.

[0033] This step is based on the centroid calculation of the closed polygon, which can realize the re-calibration of the vortex center position.

[0034] S6. Compare the error between the vortex center and the initial vortex center in the vortex information after preliminary screening. If the error is less than 2 km, directly output the vortex information; otherwise, update the coordinates of the potential vortex center to obtain the confirmed vortex information. The confirmed vortex information includes vortex type, vortex shape, longitude of the vortex center, and latitude of the vortex center.

[0035] This step can realize the repeated correction of the vortex center and contour by setting threshold filtering conditions and iterating multiple times, and obtain more accurate vortex recognition information.

[0036] The main body of this method is constructed in a layered structure based on the physical structure characteristics of the vortex, mainly including 5 process steps: cleaning, preliminary screening, re-screening, correction, and confirmation. The vortex position (longitude and latitude), type, and shape are extracted from the flow field through multi-level filters. This method has the characteristics of light weight, easy deployment, and easy migration. The method can be expanded according to needs. By adding other vortex recognition and positioning models in the vortex screening stage of this model, including but not limited to empirical, physical feature-based, and machine learning-based models, the rapid fusion of the models can be achieved.

[0037] Construct an identification model for multi-layer filter fusion based on the stream function, which can enhance the effect of vortex identification compared with traditional vortex identification methods. Compared with machine / deep learning models, this method has a more concise structure, requires less data, has lower hardware performance requirements, and can be applied to multiple scenarios, enabling lightweight and rapid engineering production deployment. It solves the problems of low identification efficiency, high misidentification rate, poor interpretation effect of current traditional vortex identification methods, as well as high hardware requirements, high data volume dependence, difficult adjustment of model hyperparameters, and poor reusability of vortex identification methods based on deep learning.

[0038] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for rapid identification of mesoscale vortices based on the construction of stream functions, characterized in that, The method includes the following steps: S1. Input the original data of mesoscale vortices in the recognition area into a processing model for data cleaning; S2. Input the mesoscale vortex data after data cleaning into a flow velocity limit filter and a flow velocity zero change filter for preliminary data screening to obtain the vortex information after preliminary screening; S3. Input the vortex information after preliminary screening into a flow direction around the center filter for data re-screening to obtain the vortex information after re-screening; The flow direction around the center filter performs quadrant cycling according to the flow field direction in the vortex. The flow direction around the center filter is used to calculate the quadrant where the flow direction of the outer edge grid points around 3 grid points of the potential vortex center is located, and filter the vortices that do not conform to the vortex structure characteristics; S4. Input the vortex information after re-screening into a stream function filter to correct the vortex information after re-screening; S5. Capture the shape to obtain the vortex shape, and calculate the centroid of the vortex shape as the vortex center; S6. Compare the error between the vortex center and the initial vortex center in the vortex information after preliminary screening. If the error is less than 2 km, directly output the vortex information. Otherwise, update the coordinates of the potential vortex center to obtain the confirmed vortex information.

2. The rapid identification method of mesoscale vortices based on stream function construction according to claim 1, wherein: The vortex information after preliminary screening is the longitude and latitude of the vortex center. The vortex information after re-screening is the vortex type, the basic contour of the vortex, the longitude and latitude of the vortex center. The confirmed vortex information is the vortex type, the vortex shape, the longitude and latitude of the vortex center.

3. A method for quickly identifying mesoscale vortices based on stream function construction according to claim 1, characterized in that: In step S1, the processing model includes a land mask model and an abnormal flow velocity smoothing model. In the land mask model, the land without flow field information is marked by special marks, and the flow velocity magnitude within 5 km of the grid points is searched and statistically counted by moving the grid points in turn; in the abnormal flow velocity smoothing model, if the difference between the grid point flow velocity and the average flow velocity of the statistical area is 5 times the standard deviation of the flow velocity within the statistical area, the flow velocity of this grid point is considered as an abnormal flow velocity, and the flow velocity of this grid point is replaced by the average flow velocity of the 8 neighboring grid points of this grid point.

4. A method for rapid identification of mesoscale vortices constructed based on stream functions according to claim 1, characterized in that: In step S2, the flow velocity limit filter moves the grid points in turn to calculate the longitude and latitude positions of the local minimum flow velocity within the 24-neighboring range of the grid points, and marks them as potential vortex centers. The flow velocity zero change filter checks the changes in the flow direction in the N-S and W-E directions of the potential vortex center, and filters the vortices that do not conform to the vortex structure characteristics.

5. A method for rapidly identifying mesoscale vortices based on stream function construction according to claim 1, characterized in that: In step S4, the initial search radius R of the stream function filter is 10 km. Calculate the stream function distribution within the range of R of the potential vortex center, obtain the streamline that is closed and contains the initial vortex center in the stream function field, interpolate and average the flow velocities of the connection points of the closed streamline, arrange them according to the area size of the closed figure, and select the closed streamline with the largest area and non-decaying average flow velocity of the closed figure as the vortex shape.

Images