An Adaptive Simulation Scheduling and Tornado-like Vortex Potential Identification Method and System for Disastrous Wind Type Classification

By adaptive simulation scheduling of hazardous wind types and identification of tornado-like vortex potential, the problems of difficulty in distinguishing between straight and rotating winds and inconsistent parameters of mixed winds in existing technologies have been solved, achieving accurate simulation path selection and efficient resource utilization.

CN123116032APending Publication Date: 2026-11-17HUNAN BRANCH OF CHINA METEOROLOGICAL ADMINISTRATION METEOROLOGICAL CADRE TRAINING COLLEGE (HUNAN INFORMATION ENG SCHOOL)
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611614592.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively distinguish between linear and rotating severe winds, and inconsistent parameter calculations in mixed wind types lead to unnecessary high-resolution simulation missions. Furthermore, the lack of clear parameter calculation and scoring methods prevents stable conversion to corresponding simulation paths.

Method used

By calculating the scores of straight-line and rotating gale parameters separately, and determining the type of hazardous gale based on the score results, large eddy simulation and tornado-like vortex identification are performed only when the rotating gale meets the conditions. This avoids duplicate parameter extraction and inconsistent data states, establishes clear parameter normalization and scoring methods, and ensures the accuracy of the simulation path and the relevance of resources.

Benefits of technology

It improves the accuracy of identifying hazardous wind types, reduces unnecessary high-resolution simulation computations, ensures the stability of simulated paths and efficient use of resources, and enhances the reliability of simulation identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116032A_ABST
    Figure CN123116032A_ABST
Patent Text Reader

Abstract

This invention discloses an adaptive simulation scheduling and tornado-like vortex potential identification method and system for identifying hazardous wind types. It acquires numerical model background fields, surface observation data, and Doppler weather radar data for the target area during the target time period, generates a variational assimilation analysis field, and performs convective-scale numerical simulation. From the simulation results, it extracts linear and rotating wind parameters to identify the hazardous wind types of candidate convective cells. For linear winds, it performs linear wind diagnosis based on non-large eddy simulation results. For rotating winds, it performs tornado potential triggering discrimination; when the tornado potential triggering conditions are met, it generates large eddy simulation task parameters and performs large eddy simulation, performing objective identification of tornado-like vortices based on the large eddy simulation results. For mixed-type winds, it simultaneously performs linear wind diagnosis and tornado potential triggering discrimination. This invention can reduce unnecessary large eddy simulation computational costs and improve the targeting of simulation identification for different types of hazardous winds.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to numerical simulation and objective identification techniques for severe weather, specifically to an adaptive simulation scheduling and tornado-like vortex potential identification method and system for identifying severe wind types. Background Technology

[0002] Disastrous gales include straight-line gales primarily driven by cold pool outflow, gust fronts, and momentum downflow, as well as rotating gales accompanied by mesocyclones, low-level closed circulations, or tornado-like vortices. The diagnostic parameters and numerical simulation resolutions required for these two types of processes differ: straight-line gales can typically be characterized by convective or sub-kilometer-scale simulations, while near-surface small-scale eddies in rotating gales may require simulations of large eddies at the hundred-meter scale.

[0003] Four-dimensional variational assimilation and nested large eddy simulation for detailed reproduction of historical tornado cases are known research methods. They can improve the initial field and resolve strong rotating convection structures. Publicly available studies of this type mainly illustrate the effects of assimilation and large eddy simulation in specific cases, but do not provide simulation path selection methods for different types of hazardous winds.

[0004] Existing thunderstorm and gale identification technologies can extract multi-dimensional feature parameters from data such as radar, wind profiles, radiosondes, and ground stations, and output the probability of thunderstorms and gales after normalization and fusion. High-resolution thunderstorm and gale research has also disclosed diagnostic criteria such as bow echoes, cold pools, gust fronts, and momentum transfer. These methods address the question of whether thunderstorms and gales have occurred or the analysis of individual case structures, but they do not determine whether to activate WRF-LES based on the difference between linear and rotational patterns.

[0005] However, existing studies mostly provide case-specific diagnostic indicators for straight-line or rotating winds, and cannot directly form an engineering process of "classification identification—simulated path selection—conditional triggering WRF-LES". Implementing this process presents at least the following technical challenges: First, the same candidate convective cell may simultaneously exhibit cold pool outflow, momentum transfer, and low-level rotation. These different types of characteristics are not mutually exclusive, making it difficult to distinguish between straight-line, rotating, and mixed-type winds using a single score or threshold. Second, mixed-type winds require parameter calculation results from the same candidate convective cell at the same time point as input, simultaneously executing both straight-line and rotating branches. This necessitates avoiding duplicate parameter extraction while ensuring consistency in the data used by both branches. Third, straight-line wind parameters, rotating wind parameters, and tornado-like vortex potential parameters have different dimensions and value ranges, requiring the establishment of clear parameter calculation, normalization, and evaluation mechanisms. Only through a segmentation method can the segmentation results be stably converted into the corresponding simulation path; fourth, the storm's relative helicity, low-level vertical wind shear, and lifting condensation height are environmental conditions, while the low-level maximum vertical vorticity and the overlap rate of rotating updrafts are rotating updraft structure conditions of the candidate cell. If the above parameters are simply averaged arithmetically, the higher environmental parameters may compensate for insufficient low-level rotating structure, leading to unnecessary activation of WRF-LES; fifth, after the triggering conditions are met, it is also necessary to determine the nested region prediction center based on the current position, moving speed, and predetermined simulation duration of the candidate rotating cell, and convert the triggering results into executable WRF-LES task parameters such as grid spacing, start time, and output interval. Existing technologies have not yet provided clear data transfer relationships and judgment rules between the above-mentioned stages. Therefore, the core technical problem that this invention needs to solve is not to reintroduce four-dimensional variational assimilation, large eddy simulation, or single hazardous wind index, but to establish an integrated engineering process consisting of hazardous wind dual-scoring classification, simulation path selection, independent judgment of tornado-like vortex potential parameters, and WRF-LES task generation, so that the numerical simulation path and computational resource input correspond to the hazardous wind type and low-level rotation conditions. Summary of the Invention

[0006] The technical problem to be solved by this invention is: 1. Within the same candidate convective cell, cold pool outflow, momentum transfer, and low-level rotation may coexist. These different types of characteristics are not mutually exclusive, and it is difficult to distinguish between linear, rotating, and mixed winds using a single score or threshold.

[0007] 2. Mixed-type gales require the parameter calculation results of the same candidate convective cell at the same time as input, and simultaneously execute linear and rotating branches. It is necessary to avoid duplicate parameter extraction and ensure that the data status used by the two branches is consistent. In addition, the parameters of linear gales, rotating gales and tornado-like vortex potential parameters have different dimensions and different value ranges. It is necessary to establish clear parameter calculation, normalization and scoring methods to stably convert the classification results into the corresponding simulation path.

[0008] 3. The relative helicity of the storm, the lower-level vertical wind shear, and the lifting condensation height are environmental conditions, while the lower-level maximum vertical vorticity and the overlap rate of the rotating updraft are rotating updraft structure conditions of the candidate cell. If the above parameters are simply averaged, the higher environmental parameters may compensate for the insufficient lower-level rotating structure, leading to the unnecessary activation of WRF-LES.

[0009] 4. Existing thunderstorm wind identification technologies do not determine whether to activate WRF-LES based on the difference between linear and rotational winds. Furthermore, linear winds can usually be simulated and characterized by convective or sub-kilometer scales, while small-scale near-surface eddies in rotational winds may require simulation of large eddies at the hundred-meter scale.

[0010] 5. Existing technologies, after meeting the triggering conditions, still need to determine the nested region prediction center based on the current position, moving speed, and predetermined simulation duration of the candidate rotating unit, and convert the triggering results into executable WRF-LES task parameters such as grid spacing, reporting start time, and output interval.

[0011] To address the aforementioned problems in existing technologies, this paper provides an adaptive simulation scheduling and tornado-like vortex potential identification method and system for identifying the types of severe gales. The method clarifies the calculation and normalization methods for each type parameter, and performs fine-grained reproduction and objective identification of tornado-like vortices only for the rotating branches of rotating or mixed gales when the low-level rotating structure and environmental conditions reach preset thresholds, without initiating 100-meter-level large eddy simulation for straight gales.

[0012] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides an adaptive simulation scheduling and tornado-like vortex potential identification method for identifying the types of disastrous strong winds, comprising the following steps: Acquire the numerical model background field, ground observation data, and Doppler weather radar data for the target area during the target time period, and perform quality control and format conversion on the ground observation data and Doppler weather radar data; Using the numerical model background field as the initial background, ground observation data after quality control and format conversion and Doppler weather radar data are used as observation inputs to generate a variational assimilation analysis field, and convective-scale numerical simulation is performed based on the variational assimilation analysis field. The straight-line wind parameters and rotational wind parameters of each candidate convective cell are extracted from the convective-scale numerical simulation results. The straight-line wind score and the rotating wind score of each candidate convective cell are calculated based on the straight-line wind parameters and the rotating wind parameters, respectively. The hazardous wind type of each candidate convective cell is determined based on the comparison results of the straight-line wind score, the rotating wind score and the corresponding threshold. If the straight-line wind score is not less than the corresponding threshold and the rotating wind score is less than the corresponding threshold, then the hazardous wind type of the corresponding candidate convective cell is straight-line. According to the preset straight-line fine-tuning flag, the convective-scale numerical simulation results or the sub-kilometer-scale non-large eddy simulation results are called. Based on the called simulation results, the candidate convective cell is diagnosed as having a straight-line wind and a straight-line wind risk product is output. Large eddy simulation is not initiated for the candidate convective cell. If the rotating gale score is not less than the corresponding threshold and the straight gale score is less than the corresponding threshold, then the hazardous gale type of the corresponding candidate convective cell is rotating. Tornado potential triggering is determined based on the rotating gale parameters of the candidate convective cell. When the tornado potential triggering condition is met, large eddy simulation task parameters are generated and large eddy simulation is performed on the candidate convective cell. Based on the large eddy simulation results, tornado-like vortex objective identification is performed, and tornado-like vortex potential products are output. When the tornado potential triggering condition is not met, rotating gale potential monitoring products are output. If both the straight-line gale score and the rotating gale score are not less than the corresponding threshold, then the hazardous gale type of the corresponding candidate convective cell is mixed. At the same time, straight-line gale diagnosis and tornado potential triggering discrimination are performed on the candidate convective cell, and the corresponding straight-line gale risk product and rotating gale potential monitoring product or tornado-like vortex potential product are output. If both the straight-line gale score and the rotating gale score are less than the corresponding threshold, the hazardous gale type of the corresponding candidate convective cell is not satisfied with the classification conditions. The straight-line gale diagnosis and the tornado potential triggering judgment are not performed, and the result of not satisfying the classification conditions is output.

[0013] Furthermore, the quality control includes checking for missing data, physical extent, and consistency of ground observation data, as well as velocity deblurring, ground clutter removal, and outlier removal of Doppler weather radar data; the format conversion includes converting ground observation data into the site observation format of the variational assimilation system, and converting radar polar coordinate data in Doppler weather radar data into a radar observation format that the variational assimilation system can read.

[0014] Furthermore, the convective-scale numerical simulation results include a three-dimensional wind field, temperature and humidity field, radar reflectivity factor, vertical velocity, vertical vorticity, storm relative helicity, and 0 to 1 km vertical wind shear. The storm relative helicity and 0 to 1 km vertical wind shear are calculated based on the three-dimensional wind field. When extracting the straight-line wind parameters and rotational wind parameters for each candidate convective cell from the convective-scale numerical simulation results, the following are included: Connected regions with radar reflectivity factors not less than a preset reflectivity threshold are selected as candidate convective cells. Calculate the major-minor axis ratio, leading-edge curvature, and maximum horizontal reflectivity gradient of the candidate convection cell based on the radar reflectivity factor corresponding to the candidate convection cell. Calculate the near-surface temperature difference based on the 2 m temperature in the temperature and humidity field corresponding to the candidate convective cell; Calculate the maximum divergence and descending airflow in the lower layer based on the three-dimensional wind field corresponding to the candidate convective cells; The major-minor axis ratio, leading edge curvature, maximum horizontal reflectivity gradient, near-surface temperature difference, maximum lower-level divergence, and descending air flow rate of the same candidate convective cell are used as the straight-line wind parameters corresponding to the candidate convective cell. Calculate the maximum vertical vorticity in the lower layer, the vertical wind shear from 0 to 1 km, and the lower layer circulation based on the three-dimensional wind field corresponding to the candidate convective cells. The overlap rate of the rotating updraft is calculated based on the three-dimensional wind field, vertical velocity, and vertical vorticity corresponding to the candidate convective cells. The relative helicity of the storm is calculated based on the three-dimensional wind field corresponding to the candidate convective cell and the moving speed of the candidate cell. The maximum vertical vorticity in the lower layer, the vertical wind shear from 0 to 1 km, the relative helicity of the storm, and the overlap rate of the lower-level circulation and the rotating updraft corresponding to the same candidate convective cell are used as the rotating wind parameters corresponding to the candidate convective cell.

[0015] Furthermore, the formula for calculating the overlap rate of the rotating upward airflow is as follows: U k = {(i,j)∈Ω | w(i,j,z k ≥ 5 m / s V k 0= {(i,j)∈Ω | ζ(i,j,z k ) ≥ ZL} R = max k [A(U k ∩ V k ) / A(V k )] Where R represents the overlap rate of the rotating updraft, k represents the model layer number of the candidate convective cell in the range of 0 to 3 km altitude, and U k V represents the set of grids in which the vertical velocity of the k-th mode layer within a candidate convection cell is not less than a preset vertical velocity threshold; k 0 This represents the set of grids where the vertical vorticity of the k-th model layer within a candidate convective cell is not less than the lower bound of the normalized vertical vorticity of the lower layers, for V k 0 Perform connectivity marking for edge- or corner-connected regions, and select the connected region containing the largest vertical vorticity grid in that layer and with a horizontal area not less than four times the area of ​​the pattern grid as the effective rotation region V. k If no connected region meets the conditions, then V k It is an empty set; max k This indicates that k is the maximum value, A(·) represents the horizontal area of ​​the grid set, (i,j) represents the horizontal grid coordinates within the candidate convective cell, w represents the vertical velocity, ζ represents the vertical vorticity, and z k ZL represents the vertical coordinate of the k-th model layer, and ZL represents the lower bound of the normalized vertical vorticity in the lower layer, ZL = 0.002 + 0.0015(3-Δx), where Δx is the horizontal grid spacing for convective-scale simulation; when V k It is an empty set or A(V) k When )=0, the area ratio of this layer is 0.

[0016] Furthermore, when calculating the straight-line wind score and the rotational wind score for each candidate convective cell based on the straight-line wind parameters and the rotational wind parameters, the process includes: The straight-line strong wind parameters are normalized according to their corresponding lower and upper limits, as shown in the following formula: p i = min{1, max[0, (x i -a i ) / (b i -a i )]} Where, p i Let x represent the normalized linear wind parameter of the i-th candidate convective cell. i Let a represent the linear wind parameter of the i-th candidate convective cell. i b represents the lower limit of the i-th linear strong wind parameter.i Let b represent the upper limit of the parameter for the i-th linear strong wind, and b i Greater than a i ; The linear gale score is calculated based on preset non-negative weights, using the following formula: S1= (Σ i w i p i ) / (Σ i w i ) Where S1 represents the linear wind score of the current candidate convective cell, w i The weights w represent the weights of the i-th linear strong wind parameters. i All are non-negative and Σ i w i Greater than 0; The rotational wind parameters are normalized according to their corresponding lower and upper limits, as shown in the following formula: q j = min{1, max[0, (y j -c j ) / (d j -c j )]} Where, q j y represents the j-th normalized rotating wind parameter of the current candidate convective cell. j c represents the j-th rotational wind parameter of the current candidate convective cell. j d represents the lower limit of the j-th rotating gale parameter. j Let d represent the upper limit of the j-th rotating gale parameter, and d j Greater than c j ; The rotational gale score is calculated based on preset non-negative weights, using the following formula: S2= (Σ j v j q j ) / (Σ j v j ) Where S2 represents the rotational wind score of the current candidate convective cell, v j This represents the weight of the j-th rotating gale parameter, and each v j All are non-negative and Σ j v j Greater than 0.

[0017] Furthermore, when calling simulation results based on preset linear refinement flags and performing linear high-wind diagnosis on the candidate convective cells, the process includes: The system reads a preset linear refinement flag; when the linear refinement flag is 0, it calls the convective-scale numerical simulation results; when the linear refinement flag is 1, it uses the convective-scale numerical simulation results as the initial field and boundary conditions, performs a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km and boundary layer parameterization, and calls the sub-kilometer-scale non-large eddy simulation results; it extracts 10 m wind speed, near-surface temperature difference, lower-level maximum divergence, and descending air flow from the called simulation results. A connected region consisting of grids with wind speeds of 10 m or more not less than the preset gale speed threshold and whose edges or corners are connected is defined as a straight-line gale zone. The maximum wind speed and the period of impact are determined based on the number of consecutive outputs. Output the location, maximum wind speed, and duration of impact of a straight-line gale as a risk product for straight-line gales.

[0018] Furthermore, the convective-scale numerical simulation results also include the elevation of condensation height. When determining tornado potential triggering based on the rotating wind parameters of the candidate convective cells, the following includes: The storm's relative helicity, vertical wind shear from 0 to 1 km, and maximum vertical vorticity at lower levels were normalized using a positive normalization function N. + The process involves using an inverse normalization function N to adjust the condensation height. - The processing and calculation formula is as follows: h = N + (H;100,250) s = N + (S;7.5,12.5) z = N + (Z;ZL,ZU) l = N - (L;750,1500) Where h, s, z, and l represent the normalized relative helicity of the storm, the normalized vertical wind shear from 0 to 1 km, the normalized maximum vertical vorticity in the lower layer, and the normalized lifting condensation height, respectively; H, S, Z, and L represent the relative helicity of the storm, the vertical wind shear from 0 to 1 km, the maximum vertical vorticity in the lower layer, and the lifting condensation height, respectively; ZL and ZU represent the lower and upper limits of the normalized vertical vorticity in the lower layer, respectively; ZL = 0.002 + 0.0015(3-Δx), ZU = 0.004 + 0.004(3-Δx), where Δx is the horizontal grid spacing of the convective scale simulation. The tornado potential trigger condition is determined to be met when the following conditions are met: z ≥ 0.60 ∧ R ≥ 0.50 ∧ [I(h ≥ 0.60) + I(s ≥ 0.60) + I(l ≥ 0.60)≥ 2] Where ∧ represents logical AND, R represents the unnormalized original rotating updraft overlap rate, I is 1 when the condition is met, otherwise it is 0.

[0019] Further, when generating large eddy simulation (LES) task parameters and performing LES simulation on the candidate convective cells, and objectively identifying tornado-like vortices based on the LES simulation results, the following steps are included: The candidate convective cell is denoted as the candidate rotating cell. The current center c0 of the candidate rotating cell is converted into the model projection plane coordinates. Its moving velocity vector in the same projection coordinate system is denoted as v. The nested region prediction center c1 after a predetermined simulation duration T is calculated according to c1 = c0 + vT. The candidate cell moving velocity vectors c and v used for storm relative helicity calculation represent the vectors of the moving velocity of the same candidate cell in the corresponding coordinate system. According to the prediction center c1, an outer large eddy simulation region and an inner eddy analysis region are set. The horizontal resolution of the inner eddy analysis region is higher than that of the outer large eddy simulation region. The reporting time is set to a predetermined time period before the triggering time, and the output interval is set to a predetermined time interval. Boundary layer parameterization and cumulus parameterization are turned off in the large eddy simulation region, and three-dimensional subgrid turbulence closure is adopted. Connectivity region identification is performed on the low-level wind field from 0 to 1 km in consecutive output time intervals; new objects are created for rotating structures that appear for the first time and cannot be associated with existing objects; in adjacent time intervals, rotating structures with a center distance of no more than D are associated with the same object, where D = (Vc + 10 m / s)Δt, Vc is the moving speed of the same rotating object in the previous time interval, and Δt is the output time interval. Let A denote the horizontally connected region of the lowest-level rotational structure of the pattern. c Its closed boundary is denoted as A c According to Γ c = ∮ A c V·dl = A c ζdA calculates the circulation, where ζ represents the vertical vorticity; in the Northern Hemisphere, when Γ c A value greater than 0 indicates that the rotating structure exhibits a cyclonic closed circulation. The maximum near-surface wind speed is taken as A. c The maximum horizontal wind speed at the lowest level within the model range. This applies when the same rotating structure simultaneously exhibits a cyclonic closed circulation, lasting at least 2 minutes, and a maximum vertical vorticity of at least 0.20 s. -1Furthermore, if the maximum horizontal wind speed at the lowest level of the model is not less than 29 m / s, it is determined to be a tornado-like vortex, and the simulation identification conclusion, along with the corresponding occurrence time, center location, movement path, maximum vertical vorticity, and maximum horizontal wind speed at the lowest level of the model, is output; otherwise, the evaluation conclusion and corresponding indicators for not meeting the tornado-like vortex standard are output. The above positive or negative simulation identification results together constitute the tornado-like vortex potential product.

[0020] Furthermore, when outputting the rotating high wind potential monitoring product, the output specifically includes the normalized storm relative helicity, the normalized vertical wind shear from 0 to 1 km, the normalized lower-level maximum vertical vorticity, the normalized lifting condensation height, and the overlapping rate of the rotating updraft, and marks the parameters that do not meet the tornado potential triggering conditions.

[0021] Secondly, the present invention provides an adaptive simulation scheduling and tornado-like vortex potential identification system for identifying the types of severe winds, comprising: The data input module is used to acquire the numerical model background field, ground observation data, and Doppler weather radar data of the target area during the target time period; The quality control and format conversion module is used to perform quality control and format conversion on the ground observation data and Doppler weather radar data. The analysis field and convective-scale simulation module is used to generate a variational assimilation analysis field by taking the numerical model background field as the initial background, the ground observation data after quality control and format conversion and the Doppler weather radar data as the observation input, and to perform convective-scale numerical simulation based on the variational assimilation analysis field. The hazardous wind classification module is used to extract the straight-line wind parameters and rotational wind parameters for each candidate convective cell from the convective-scale numerical simulation results. Based on these parameters, it calculates the straight-line wind score and rotational wind score for each candidate convective cell, and determines the hazardous wind type for each candidate convective cell by comparing the scores with corresponding thresholds. If the straight-line wind score is not less than the corresponding threshold and the rotational wind score is less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is straight-line. If the rotational wind score is not less than the corresponding threshold and the straight-line wind score is less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is rotational. If both the straight-line wind score and the rotational wind score are not less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is mixed. If both the straight-line wind score and the rotational wind score are less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell does not meet the classification conditions. The simulation path selection module is used to selectively trigger the non-large eddy simulation path processing module, the WRF-LES condition triggering module, or both simultaneously, based on the hazardous wind type of the candidate convective cell. When the hazardous wind type is straight, the non-large eddy simulation path processing module is triggered; when the hazardous wind type is rotating, the WRF-LES condition triggering module is triggered; and when the hazardous wind type is mixed, both the non-large eddy simulation path processing module and the WRF-LES condition triggering module are triggered simultaneously. The non-large eddy simulation path processing module is used to call up convective-scale numerical simulation results according to a preset straight-line fine-tuning flag when the type of hazardous gale is linear or mixed. Alternatively, it can use the convective-scale numerical simulation results as the initial field and boundary conditions to perform a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km. Based on the called simulation results, it performs a straight-line gale diagnosis and outputs a straight-line gale risk product to the result output module. The non-large eddy simulation path processing module and the mixed-type straight-line branch do not activate WRF-LES. Whether the mixed-type rotating branch activates WRF-LES is determined by the WRF-LES condition triggering module. The WRF-LES condition triggering module is used to determine the tornado potential triggering based on the rotating wind parameters of the candidate convective cell when the type of severe gale is rotating or mixed. Based on the determination result, it connects to the adaptive WRF-LES execution module or the non-triggered processing module. An adaptive WRF-LES execution module is used to generate large eddy simulation task parameters and perform large eddy simulation on the candidate convective cell when the tornado potential triggering condition is met. The tornado-like vortex objective identification module is used to perform objective identification of tornado-like vortices based on the large eddy simulation results output by the adaptive WRF-LES execution module, and output the tornado-like vortex potential product to the result output module. The untriggered processing module is used to output the rotating high wind potential monitoring product to the result output module when the tornado potential triggering conditions are not met. The results output module is used to output the classification results and simulated paths of candidate convective cells, and output one or more of the following: linear high wind risk products, rotating high wind potential monitoring products, and tornado-like vortex potential products, or output results that do not meet the classification conditions.

[0022] Compared with the prior art, the advantages of the present invention are as follows: 1. Addressing the issue that a single candidate convective cell may simultaneously exhibit cold pool outflow, momentum transfer, and low-level rotation, and that different types of characteristics are not mutually exclusive, making it difficult to distinguish between linear, rotating, and mixed-type gales using a single score or threshold, this invention calculates the linear and rotating gale scores for each candidate convective cell based on linear and rotating gale parameters respectively. It then determines the hazardous gale type of each candidate convective cell based on the comparison results of the linear and rotating gale scores with the corresponding thresholds. This allows the same candidate cell to be identified as linear, rotating, a mixed type exhibiting both characteristics, or not meeting the classification criteria. The combination of the two scores directly selects the subsequent simulation path or prevents it from entering the simulation path, thus solving the technical problem that a single score cannot distinguish mixed-type convective cells and improving the accuracy of hazardous gale type identification.

[0023] 2. To address the issue of mixed-type gales requiring parameter calculations from the same candidate convective cell at the same time point as input, and simultaneously executing linear and rotating branches, it is crucial to avoid redundant parameter extraction while ensuring data consistency between the two branches. Furthermore, linear gale parameters, rotating gale parameters, and tornado-like vortex potential parameters have different dimensions and value ranges, necessitating the establishment of clear parameter calculation, normalization, and scoring methods to stably convert the classification results into corresponding simulation paths. This invention extracts the linear and rotating gale parameters for each candidate convective cell from the convective-scale numerical simulation results and simultaneously performs [further processing / analysis] when the gale is classified as mixed-type. The candidate convective cells undergo linear strong wind diagnosis and tornado potential triggering discrimination, converting the classification results into fixed simulation paths: linear types call non-large eddy simulation results, rotating types enter triggering discrimination, and mixed types execute both branches simultaneously. If the classification conditions are not met, the simulation path is not entered, thus avoiding directly starting the high-resolution simulation task with a single thunderstorm strong wind probability. This ensures that the two branches run based on the same time and the same data state, and converts physical quantities of different dimensions into dimensionless scores through scoring calculation. A stable mapping relationship from parameter extraction to simulation path selection is established, thereby avoiding repeated parameter extraction and ensuring the consistency and synchronization of data states.

[0024] 3. Since storm relative helicity, low-level vertical wind shear, and lifting condensation height are environmental conditions, while low-level maximum vertical vorticity and the overlap rate of rotating updrafts are rotating updraft structure conditions of candidate cells, simply averaging these parameters may lead to insufficient compensation for low-level rotating structure, resulting in unnecessary activation of WRF-LES. This invention uses the rotating wind parameters of candidate convective cells to determine tornado potential triggering. Large eddy simulation (LES) is activated when the rotating wind parameters meet the tornado potential triggering conditions; otherwise, it is not activated. Instead of averaging the four normalized values ​​and the overlap rate of rotating updrafts, the normalized value of low-level maximum vertical vorticity and the overlap rate of rotating updrafts are set as parameters that must reach a threshold. Furthermore, at least two of the storm relative helicity, 0-1 km vertical wind shear, and low lifting condensation height must reach the threshold. This avoids excessively high environmental parameters compensating for insufficient low-level rotating structure and thus erroneously triggering WRF-LES, reducing unnecessary LES calculations and improving the targeted allocation of computational resources.

[0025] 4. Addressing the issues that existing thunderstorm wind identification technologies do not determine whether to activate WRF-LES based on the difference between linear and rotating wind types, and that linear winds can usually be simulated at the convective or sub-kilometer scales while near-surface small-scale eddies in rotating winds may require large eddy simulations at the hundred-meter scale, this invention performs linear wind diagnosis without activating WRF-LES for the linear branch when the hazardous wind type is linear or mixed. When the hazardous wind type is rotating or mixed, tornado potential triggering is determined through the rotating branch, allocating only hundred-meter-scale computational resources to candidate cells that simultaneously meet both low-level rotating structure and environmental conditions. Therefore, a risk product can be output without activating WRF-LES for pure linear processes, and the two branches of mixed processes are not interchangeable, reducing unnecessary large eddy simulation computations while maintaining the diagnostic capabilities for the location and intensity of linear winds.

[0026] 5. Addressing the issue that existing technologies, after meeting triggering conditions, still require determining the nested region prediction center based on the current position, speed, and predetermined simulation duration of candidate rotating cells, and converting the triggering results into executable WRF-LES task parameters such as grid spacing, reporting time, and output interval, this invention generates large eddy simulation task parameters and performs large eddy simulation on the candidate convective cells when the rotating wind parameters meet the tornado potential triggering conditions. This enables the classification results to be converted into executable WRF-LES task parameters, achieving integrated scheduling from hazardous wind classification identification to high-resolution simulation execution.

[0027] 6. When this invention performs objective identification of tornado-like vortices based on large eddy simulation results and outputs tornado-like vortex potential products, it uses a combination of closed circulation, duration, maximum vertical vorticity, and maximum near-surface wind speed conditions to output tornado-like vortex identification conclusions for WRF-LES results. This facilitates the verification of tornado-like vortex reproduction results and improves the traceability of the simulation identification process and the credibility of business applications. Attached Figure Description

[0028] Figure 1 A flowchart illustrating the method for selecting simulated paths and identifying tornado-like vortex potential in the classification of severe winds provided in this embodiment of the invention.

[0029] Figure 2 The structural block diagram of the hazardous gale classification condition triggering WRF-LES and tornado-like vortex objective identification system provided in the embodiments of the present invention.

[0030] Figure 3 The results are simulations of a typical linear strong wind process at different times and resolutions.

[0031] Figure 4 The simulation results are for a typical rotating gale process at 0620 UTC, 0630 UTC, 0640 UTC, 0650 UTC, 0810 UTC, 0820 UTC, 0830 UTC and 0840 UTC.

[0032] Figure 5 This is a superimposed image of radar reflectivity factor at 1.5 km altitude, model lowest-level wind field, and 2–5 km integral ascending helicity UH for cell 1 at 0830 UTC on June 23, 2016, in region D05 at a resolution of 111.11 m. Detailed Implementation

[0033] The present invention will be further described below with reference to the accompanying drawings and specific preferred embodiments, but this does not limit the scope of protection of the present invention.

[0034] This embodiment provides a method for identifying tornado-like vortex potential based on the classification of severe gale-force winds and conditionally triggered large eddy simulation. The overall process is as follows: Figure 1 As shown, it includes the following steps: S101, acquires multi-source meteorological data.

[0035] The numerical model background field, ground observation data, and Doppler weather radar data for the target area during the target time period are acquired. Ground observation data includes data from ground meteorological stations or regional automatic weather stations, while Doppler weather radar data includes Doppler weather radar reflectivity factors and radial velocity data. The numerical model background field is used for variational assimilation of the initial background and subsequent model boundaries, while the ground observation data and Doppler weather radar data are used for observational constraints.

[0036] S102, Quality Control and Format Conversion.

[0037] The ground observation data and Doppler weather radar data undergo quality control and format conversion. Quality control includes checking for missing data, physical extent, and consistency in ground observations, and performing velocity deblurring, ground clutter removal, and outlier removal on Doppler weather radar data. Format conversion transforms the ground observation data and Doppler weather radar data into station observation formats and radar observation formats that the variational assimilation system can read, respectively. This includes converting ground observations to station observation formats and converting radar polar coordinate data in Doppler weather radar data to radar observation formats. This step only performs the above-mentioned quality control and format conversion.

[0038] S103 generates the variational assimilation analysis field and performs convective-scale simulation.

[0039] Using the numerical model background field as the initial background, ground observation data after quality control and format conversion, and Doppler weather radar data are used as observation inputs to generate a variational assimilation analysis field. In this embodiment, four-dimensional variational assimilation is used to generate the analysis field. Four-dimensional variational assimilation is a well-known technology in the field, and this solution does not involve any improvement or innovation to it; therefore, its specific implementation process will not be described in detail here.

[0040] Based on the variational assimilation analysis field, a convective-scale numerical simulation is performed. Specifically, a convective-scale simulation with a horizontal grid spacing of 1 km or 3 km is performed on the analysis field. The output includes convective-scale numerical simulation results of three-dimensional wind field, temperature and humidity field, radar reflectivity factor, vertical velocity, vertical vorticity, convective effective potential energy, lifting condensation height, storm relative helicity, and 0 to 1 km vertical wind shear. These results are used for classification and subsequent two simulation branches. The storm relative helicity and 0 to 1 km vertical wind shear are calculated based on the three-dimensional wind field.

[0041] S104 Extract the parameters for classifying severe winds and calculate the scores for linear and rotary winds.

[0042] First, the linear and rotational wind parameters for each candidate convective cell are extracted from the convective-scale numerical simulation results. The linear parameters include the major-to-minor axis ratio, leading-edge curvature, maximum horizontal reflectivity gradient, near-surface temperature difference, maximum lower-level divergence, and descending airflow. The rotational parameters include the maximum lower-level vertical vorticity, vertical wind shear from 0 to 1 km, storm relative helicity, lower-level circulation, and overlap rate of rotating updrafts. All parameters are calculated from the model variables of the convective-scale numerical simulation results output in step S103. The calculation process is as follows: Connected regions with radar reflectivity factors not less than a preset reflectivity threshold are selected as candidate convective cells. In this embodiment, the preset reflectivity threshold is 35 dBZ. The radar reflectivity factor at grid (i,j) is denoted as Zr(i,j). Grids with Zr(i,j) not less than 35 dBZ are denoted as 1, and the remaining grids are denoted as 0. Therefore: B(i,j) = 1, Zr(i,j) ≥ 35 dBZ; B(i,j) = 0, Zr(i,j)<35 dBZ Assign the same number to grids whose edges or corners are connected and whose B(i,j)=1. Delete connected regions whose horizontal area is smaller than the area of ​​the four pattern grids. Treat the remaining connected regions as candidate convection units, and denote their horizontal grid areas as Ω.

[0043] The major-minor axis ratio, leading-edge curvature, and maximum horizontal reflectivity gradient of the candidate convective cell are calculated based on the radar reflectivity factor corresponding to the candidate convective cell. The near-surface temperature difference is calculated based on the 2 m temperature in the temperature and humidity field corresponding to the candidate convective cell. The maximum lower-level divergence and descending airflow are calculated based on the three-dimensional wind field corresponding to the candidate convective cell. The major-minor axis ratio, leading-edge curvature, maximum horizontal reflectivity gradient, near-surface temperature difference, maximum lower-level divergence, and descending airflow corresponding to the same candidate convective cell are used as the straight-line wind parameters corresponding to the candidate convective cell. The major-minor axis ratio, leading-edge curvature, maximum horizontal reflectivity gradient, near-surface temperature difference, maximum lower-level divergence, and descending airflow are sequentially recorded as six straight-line wind parameters x1 to x6. Specifically: Linear parameters are calculated from model variables within the candidate convection cell Ω. The coordinate variances along the major and minor axes are represented by two eigenvalues ​​λ1 and λ2 of the grid coordinate covariance matrix within Ω, respectively, where λ1 is not less than λ2, and the major-to-minor axis ratio E is: E = (λ1 / λ2) 1 / 2 Arrange the leading edge mesh of the candidate unit along the moving direction according to the arc length and perform three-point smoothing. Calculate the maximum leading edge curvature K using the leading edge curve coordinates x(s) and y(s): K = max |x′y″-y′x″| / (x′ 2 + y′ 2) 3 / 2 The maximum horizontal reflectivity gradient G is: G = maxΩ [( Zr / x) 2 + ( Zr / y) 2 ] 1 / 2 The average temperature T within 2 m of the annular zone 10 km to 30 km outside the candidate monomer. e Subtracting the lowest temperature within 2 m of the candidate cell yields the near-surface temperature difference ΔT; the maximum lower-level divergence D is taken as the maximum horizontal divergence value within the range of 0 to 1 km; the descending gas flow rate M is taken as the maximum calculated result when the vertical velocity is less than 0 within the range of 0.5 km to 3 km. ΔT = T e -minΩ(T2) D = maxΩ, 0≤z≤1 km( u / x + v / y) M = maxΩ,0.5≤z≤3 km,w<0[-w(u 2 + v 2 ) 1 / 2 ] The maximum vertical vorticity, vertical wind shear from 0 to 1 km, and low-level circulation are calculated based on the three-dimensional wind field corresponding to the candidate convective cell. The overlap rate of rotating updrafts is calculated based on the three-dimensional wind field, vertical velocity, and vertical vorticity corresponding to the candidate convective cell. The relative helicity of the storm is calculated based on the three-dimensional wind field and the moving velocity of the candidate convective cell. The maximum vertical vorticity, vertical wind shear from 0 to 1 km, relative helicity of the storm, overlap rate of low-level circulation, and overlap rate of rotating updrafts corresponding to the same candidate convective cell are used as the rotating wind parameters corresponding to the candidate convective cell. The maximum vertical vorticity, vertical wind shear from 0 to 1 km, relative helicity of the storm, overlap rate of low-level circulation, and overlap rate of rotating updrafts are sequentially recorded as five rotating wind parameters y1 to y5. Specifically: Rotational parameters are also calculated from model variables within the candidate convective cell Ω. The maximum lower-level vertical vorticity Z, vertical wind shear S from 0 to 1 km, and storm relative helicity H are as follows: Z = maxΩ, 0≤z≤1 km( v / x- u / y) S = |V(1 km)-V(10 m)| H = dz Where V(z) is the horizontal wind vector at height z, c is the candidate cell movement velocity vector used for calculating the relative helicity of the storm, corresponding to the movement of the same candidate cell as the movement velocity vector v in the nested regional prediction center calculation, and k is the vertical unit vector.

[0044] Let the horizontal grid spacing for the convective-scale simulation be Δx, where Δx is in km and takes the value of 1 or 3. Then, calculate the lower limit ZL and upper limit ZU of the normalized lower-level vertical vorticity using the following formula: ZL = 0.002 + 0.0015(3-Δx) ZU = 0.004 + 0.004(3-Δx) Within the range of 0 to 1 km, the model layer containing the maximum vertical vorticity Z in the lower layer is determined; on this model layer, the connected region containing the grid of maximum vertical vorticity and with a vertical vorticity not less than ZL is denoted as C, and the lower-level circulation Γ is calculated according to the Stokes relation: Γ = C( v / x- u / y)dA In each model layer from 0 to 3 km, the set of grids with a vertical velocity of not less than 5 m / s within the candidate cell is denoted as U. k The set of grids with a vertical vorticity not less than ZL is denoted as V. k 0 ; For V k 0 Perform connectivity marking for edge- or corner-connected regions, and denote the connected region containing the grid with the maximum vertical vorticity of that layer as V. k When V k When the horizontal area is less than the area of ​​the four pattern grids, V k Set to an empty set. The overlap rate of the k-th mode layer is the rotation region V. k The area proportion that simultaneously satisfies the strong updraft condition in the rotating region, and the overlapping rate R of the rotating updraft is taken as the maximum value of each layer. The overlapping rate R of the rotating updraft is calculated according to the maximum area proportion that simultaneously satisfies the strong updraft condition in the rotating region: U k = {(i,j)∈Ω | w(i,j,z k ≥ 5 m / s V k 0 = {(i,j)∈Ω | ζ(i,j,z k ) ≥ ZL} R = max k [A(U k ∩ V k ) / A(V k )] In the above formula, R represents the overlap rate of the rotating updrafts, k represents the model layer number, and U k U represents the set of grids in the k-th model layer within the candidate convective cell where the vertical velocity is not less than a preset vertical velocity threshold. In this embodiment, the set of grids in the k-th model layer within the candidate convective cell at an altitude of 0 to 3 km with a vertical velocity not less than 5 m / s is denoted as U. k V k This represents the effective rotation region connected region, specifically V. k 0 After performing edge or corner-connected connectivity marking, take the connected region containing the grid with the maximum vertical vorticity of that layer, V. k 0 V represents the set of grids in the k-th model layer within the candidate convective cell where the vertical vorticity is not less than the lower limit of the normalized vertical vorticity of the lower layers. In this embodiment, the set of grids in the k-th model layer within the 0 to 3 km height of the candidate convective cell where the vertical vorticity is not less than ZL is denoted as V. k 0 max k This indicates that k is the maximum value, A(·) represents the horizontal area of ​​the grid set, (i,j) represents the horizontal grid coordinates within the candidate convective cell, w represents the vertical velocity, ζ represents the vertical vorticity, and z represents the vertical vorticity. k ZL represents the vertical coordinate of the k-th model layer within the candidate convective cell, and ZL represents the lower bound of the normalized vertical vorticity in the lower layer.

[0045] When V k When the horizontal area is less than the area of ​​the four pattern grids, V k Set to an empty set; in this embodiment, the rotation region V is used. k The maximum value of R is calculated for the area proportion that simultaneously satisfies the condition of strong updraft; when V k When empty, the area ratio of this layer is 0. R is a dimensionless variable between 0 and 1; R not less than 0.50 indicates that at least half of the effective rotating regions in at least one model layer simultaneously satisfy the strong updraft condition.

[0046] Then, the straight-line wind score S1 and the rotary wind score S2 are calculated separately. The calculation process is as follows: For each linear strong wind parameter x i And each rotating wind parameter y j In this embodiment, each parameter x that is positively correlated with the degree of danger is positively normalized according to a preset lower limit a and an upper limit b. N +(x;a,b) = min{1, max[0, (xa) / (ba)]} Therefore, the normalized values ​​of the six linear parameters are denoted as p. i The normalized values ​​of the five rotational parameters are denoted as q. j The calculation formula is as follows: p i = min{1, max[0, (x i -a i ) / (b i -a i )]} q j = min{1, max[0, (y j -c j ) / (d j -c j )]} Where, p i Let x represent the normalized linear wind parameter of the i-th candidate convective cell. i Let a represent the linear wind parameter of the i-th candidate convective cell. i b represents the lower limit of the i-th linear strong wind parameter. i b represents the upper limit of the i-th linear strong wind parameter. i Greater than a i i = 1 to 6; q j y represents the j-th normalized rotating wind parameter of the current candidate convective cell. j c represents the j-th rotational wind parameter of the current candidate convective cell. j d represents the lower limit of the j-th rotating gale parameter. j d represents the upper limit of the j-th rotating gale parameter. j Greater than c j j = 1 to 5.

[0047] The linear and rotary wind scores are calculated using preset non-negative weights, and the calculation formulas are as follows: S1= (Σ i w i p i ) / (Σ i w i ) S2= (Σ j v j q j ) / (Σ j v j ) Among them, w i and v jThe non-negative weights are fixed before execution and satisfy Σ i w i Greater than 0, Σ j v j Greater than 0. This embodiment uses equal weights for parameters within the same score, i.e., six weights. i All values ​​are 1, five v j All are set to 1; the lower limit a for each parameter is... i c j and upper limit b i d j Determined from historical samples of the confirmed type, the 25th and 75th percentiles of the parameter sample values ​​are taken respectively, ensuring b i Greater than a i d j Greater than c j Then, the configuration file is written before the model runs and will not be changed during the run.

[0048] Finally, the type of severe wind is determined based on the comparison results of S1 and S2 with their respective thresholds, including: S105, Straight-line wind: If the straight-line wind score is not less than the corresponding threshold and the rotating wind score is less than the corresponding threshold, then the hazardous wind type of the corresponding candidate convective cell is straight-line.

[0049] S106, Mixed-type gale: If both the straight-line gale score and the rotating gale score are not less than the corresponding threshold, then the hazardous gale type of the corresponding candidate convective cell is mixed-type.

[0050] S107, Rotary wind: If the score of the rotary wind is not less than the corresponding threshold and the score of the straight wind is less than the corresponding threshold, then the hazardous wind type of the corresponding candidate convective cell is rotary.

[0051] S108, Classification conditions not met: If both the straight-line wind score and the rotating wind score are less than the corresponding threshold, then the hazardous wind type of the corresponding candidate convective cell does not meet the classification conditions.

[0052] In this embodiment, the first scoring threshold for linear gale rating and the second scoring threshold for rotary gale rating are both 0.55. When S1 is not less than 0.55 and S2 is less than 0.55, the corresponding candidate convective cell is classified as linear gale; when S2 is not less than 0.55 and S1 is less than 0.55, the corresponding candidate convective cell is classified as rotary gale; when both S1 and S2 are not less than 0.55, the corresponding candidate convective cell is classified as mixed gale; when both scores are less than 0.55, the output does not meet the classification conditions and is not entered into the gale simulation path.

[0053] After obtaining the classification results of the hazardous wind types of candidate convective cells in step S104, a simulation path is selected based on the classification results. For example... Figure 1 As shown, path scheduling is performed only based on the comparison result of S1 and S2, including: When the hazardous wind type of the candidate convective cell is linear, in this embodiment, the non-large eddy simulation results are used to output the linear wind location, maximum wind speed, and duration of impact, without activating WRF-LES. The following steps are performed: S109, non-large eddy simulation path.

[0054] Based on the preset linear fine-grained flag, the system calls up convective-scale numerical simulation results or sub-kilometer-scale non-large eddy simulation results. Based on the called simulation results, the system performs linear high wind diagnosis on the candidate convective cell and outputs a linear high wind risk product. Large eddy simulation is not initiated for the candidate convective cell.

[0055] Specifically, the linear refinement flag written to the configuration file before execution is read; when the flag is 0, the linear branch directly uses the convective-scale simulation completed in step S103; when the flag is 1, the convective-scale results are used as the initial field and boundary conditions to execute a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km and boundary layer parameterization. This sub-kilometer-scale simulation does not disable boundary layer parameterization, nor does it use the three-dimensional subgrid turbulence settings of the large eddy simulation. When performing linear-scale high-wind diagnosis on the candidate convective cell based on the invoked simulation results, the following is included: The linear branch calls the corresponding simulation results according to the linear refinement flag, without generating WRF-LES nested regions and large eddy simulation physical parameters; Read the 10 m wind speed, near-surface temperature difference, maximum lower-level divergence, and downdraft flow rate of candidate convective cells from convective-scale numerical simulation results or sub-kilometer-scale non-large eddy simulation results. A connected region consisting of grids with wind speeds of 10 m or more not less than a preset strong wind speed threshold and whose edges or corners are connected is defined as a straight-line strong wind zone. In this embodiment, the preset strong wind speed threshold is 17.2 m / s, and a connected region consisting of grids with wind speeds of 10 m or more not less than 17.2 m / s and whose edges or corners are connected is defined as a strong wind zone.

[0056] The maximum wind speed and the period of impact are determined based on the continuous output time. Specifically, the maximum 10 m wind speed in the strong wind area is calculated in the continuous output time, and the period of impact is determined according to the first and last occurrence times of the strong wind area. Output the location, maximum wind speed, and duration of impact of a straight-line gale as a risk product for straight-line gales.

[0057] When the hazardous wind type of the candidate convective cell is rotating, in this embodiment, the rotating branch determines whether the potential parameter of the tornado-like vortex reaches a preset threshold, and decides whether to activate WRF-LES based on the determination result. The following steps are executed: First, tornado potential triggering is determined based on the rotational wind parameters of the candidate convective cells.

[0058] Specifically, the rotating branch is used for identifying tornado-like vortex potential and conditionally triggering WRF-LES. Candidate convective cells entering this branch are denoted as candidate rotating cells, and their horizontal grid regions are still denoted as Ω.

[0059] Read the storm relative helicity H, vertical wind shear S from 0 to 1 km, maximum vertical vorticity Z at the lower level, and overlap rate R of rotating updrafts calculated in step S104, and read the lifting condensation height L, without extracting the same parameters repeatedly.

[0060] The storm's relative helicity H, vertical wind shear S from 0 to 1 km, and maximum lower-level vertical vorticity Z are normalized using a positive normalization function N. + The process involves applying an inverse normalization function N to raise the condensation height L. - Processing; The overlap rate R of the rotating updraft corresponds to the normalized parameter q5 in the rotating type score. The original R before normalization is used in the tornado potential triggering judgment; R is already in the range of 0 to 1, so no secondary normalization is performed: N + (x;a,b) = min{1, max[0, (xa) / (ba)]} N - (x;a,b) = min{1, max[0, (bx) / (ba)]} In this embodiment, h, s, z, and l are obtained using the following formula: h = N + (H;100,250) s = N + (S;7.5,12.5) z = N + (Z;ZL,ZU) l = N - (L;750,1500) Where h, s, z, and l represent the normalized storm relative helicity, normalized vertical wind shear from 0 to 1 km, normalized maximum vertical vorticity in the lower tier, and normalized lifting condensation height, respectively; H, S, Z, and L represent the storm relative helicity, vertical wind shear from 0 to 1 km, maximum vertical vorticity in the lower tier, and lifting condensation height, respectively; ZL and ZU represent the normalized lower and upper limits of the lower vertical vorticity calculated in step S104, respectively; and the unit of H is meters. 2 / s 2 The unit of S is m / s, and the units of Z, ZL, and ZU are s. -1 R is dimensionless, and L is in meters (m). The environmental parameters triggered by the above settings are H not less than 190 m. 2 / s 2 S is not less than 10.5 m / s and L is not greater than 1050 m; the triggering physical value of Z varies with the grid spacing, and is 0.0032 s on a 3 km grid. -1 The value is 0.0092 s on a 1 km grid. -1 .

[0061] Since the overlap rate of the maximum vertical vorticity at the lower level and the rotating updraft directly characterizes the rotating updraft structure in the candidate cell, this embodiment does not calculate the arithmetic mean of the four normalized values ​​and R. Instead, it determines that the tornado potential triggering condition is met when the following conditions are met: z ≥ 0.60 ∧ R ≥ 0.50 ∧ [I(h ≥ 0.60) + I(s ≥ 0.60) + I(l ≥ 0.60)≥ 2] Here, ∧ represents logical AND, and I takes the value 1 when the condition is true, otherwise it takes the value 0.

[0062] When the above conditions are met, the tornado potential triggering condition is satisfied. When the tornado potential triggering condition is satisfied, WRF-LES is started; otherwise, the tornado potential triggering condition is not satisfied, WRF-LES is not started, and four normalized values, R, and parameters that have not reached the threshold are output.

[0063] like Figure 1 As shown, when the normalized value of the maximum vertical vorticity in the lower layer is not less than 0.60, the overlap rate of the rotating updraft is not less than 0.50, and at least two of the other three normalized values ​​are not less than 0.60, WRF-LES is activated and its results are objectively identified as tornado-like vortices, including the following steps: S110, start adaptive WRF-LES.

[0064] Upon triggering, large eddy simulation (LES) task parameters are generated, and LES simulation is performed on the candidate convective cell, specifically including: The candidate convective cell is denoted as the candidate rotating cell, and the predicted center is calculated based on the current center c0, the moving speed v, and the predetermined simulation duration T of the candidate rotating cell: c1 = c0 + vT Based on the prediction center c1, an outer large eddy simulation region and an inner eddy analysis region are set. In this embodiment, a 333.33 m outer large eddy simulation region and a 111.11 m inner eddy analysis region are set around c1. The reporting time is set to 30 to 60 minutes before triggering, and the output interval is 60 to 120 seconds. Boundary layer parameterization and cumulus parameterization are turned off in the large eddy simulation region, and three-dimensional subgrid turbulence closure is adopted.

[0065] S112, Objective recognition of tornado-like vortexes.

[0066] Objective identification of tornado-like vortices is performed based on large eddy simulation results. This objective identification is only performed on the WRF-LES results generated after triggering. After WRF-LES is completed, objective identification of tornado-like vortices is performed on the low-level wind field, outputting whether a tornado-like vortex has formed and its corresponding location, time, path, and intensity parameters. Specifically, this includes: Within the 0 to 1 km layer of each output time, the vertical vorticity will be no less than 0.15 s. -1 Furthermore, adjacent meshes intersecting with regions having a vertical velocity of not less than 5 m / s are connected to form a rotational structure. For rotational structures that appear for the first time and cannot be associated with existing objects, a new object is created; in adjacent time intervals, if the center distance between two rotational structures is not greater than the allowable movement distance D, they are associated as the same object. D = (Vc + 10)Δt Where Vc is the moving speed of the rotating structure in the previous time interval, in m / s; Δt is the output time interval, in s; 10 m / s is the speed margin used for association determination; and D is the allowable moving distance, in m.

[0067] Let A denote the horizontally connected region of the lowest-level rotational structure of the pattern. c Its closed boundary is denoted as A c According to Γ c = ∮ A c V·dl = A c ζdA calculates the circulation; this embodiment is located in the Northern Hemisphere, when Γ c A value greater than 0 indicates that the rotating structure exhibits a cyclonic closed circulation. The maximum near-surface wind speed is taken as A. cThe maximum horizontal wind speed at the lowest level of the model within the specified range; in this embodiment, the lowest level of the model is approximately 33 m high. This is achieved when the same rotating structure simultaneously exhibits a cyclonic closed circulation, lasting for at least 2 minutes, and a maximum vertical vorticity of at least 0.20 s. -1 Furthermore, if the maximum horizontal wind speed at the lowest level of the model is not less than 29 m / s, it is determined to be a tornado-like vortex, and the simulation identification conclusion, along with the corresponding occurrence time, center location, movement path, maximum vertical vorticity, and maximum horizontal wind speed at the lowest level of the model, is output. If all conditions are not met, the assessment conclusion that the tornado-like vortex standard is not met and the corresponding indicators are output. The above positive or negative simulation identification results together constitute the tornado-like vortex potential product.

[0068] like Figure 1 As shown, when the tornado potential triggering conditions are not met, the following steps are executed: S111, WRF-LES not started.

[0069] The output of the rotating high wind potential monitoring product specifically includes the normalized storm relative helicity h, the normalized vertical wind shear s from 0 to 1 km, the normalized lower-level maximum vertical vorticity z, the normalized lifting condensation height l, and the overlapping rate R of the rotating updraft. It also marks parameters that do not meet the tornado potential triggering conditions, namely, parameters that do not meet the conditions that the normalized value of the lower-level maximum vertical vorticity is not less than 0.60, the overlapping rate of the rotating updraft is not less than 0.50, and at least two of the other three normalized values ​​are not less than 0.60.

[0070] like Figure 1 As shown, when the hazardous wind type of the candidate convective cell is mixed, the candidate convective cell is simultaneously subjected to straight-line wind diagnosis and tornado potential triggering discrimination. Specifically, the mixed type runs both straight-line branch and rotating branch; the straight-line branch still does not start WRF-LES, and the rotating branch still needs to meet the same triggering conditions as the pure rotating type.

[0071] like Figure 1 As shown, through the aforementioned steps, a straight-line wind risk product, a rotating wind potential monitoring product, or a tornado-like vortex potential product corresponding to the candidate convective cell is obtained. Alternatively, if the candidate convective cell does not meet the classification criteria, the following steps are performed: S113, unified output.

[0072] In this embodiment, the output includes: The classification results are: linear, rotational, mixed, or the classification conditions are not met. Simulation path, namely non-large eddy simulation path, rotational triggering path, both executed simultaneously, or not entering the simulation path; The location, maximum wind speed, and duration of impact of linear branches; h, s, z, l, R, and parameters that have not reached the threshold when WRF-LES is not triggered; h, s, z, l, R, nested region center, grid distance, reporting start time, and output interval when WRF-LES is triggered; The identification conclusions of tornado-like vortices, their occurrence time, center location, movement path, maximum vertical vorticity, and maximum near-surface wind speed.

[0073] This embodiment also proposes an adaptive simulation scheduling and tornado-like vortex potential identification system for the classification of disastrous gales. This system is used to implement the method proposed in this embodiment.

[0074] like Figure 2 As shown, the system includes a data input module, a quality control and format conversion module, an analysis field and convection scale simulation module, a severe wind classification module, a simulation path selection module, a non-large eddy simulation path processing module, a WRF-LES condition triggering module, an adaptive WRF-LES execution module, a tornado-like vortex objective identification module, a non-triggered processing module, and a result output module. Among them: The data input module is used to acquire the numerical model background field, ground observation data and Doppler weather radar data of the target area during the target time period according to step S101; The quality control and format conversion module is used to perform quality control and format conversion on the ground observation data and Doppler weather radar data according to step S102; The analysis field and convective-scale simulation module is used to generate a variational assimilation analysis field according to step S103, using the numerical model background field as the initial background, and taking the ground observation data and Doppler weather radar data after quality control and format conversion as the observation input, and performing convective-scale numerical simulation based on the variational assimilation analysis field. The hazardous wind classification module is used to extract the straight-line wind parameters and rotational wind parameters of each candidate convective cell from the convective-scale numerical simulation results according to step S104. Based on the straight-line wind parameters and rotational wind parameters, it calculates the straight-line wind score S1 and the rotational wind score S2 for each candidate convective cell. Then, according to steps S105 to S108, it determines the hazardous wind type of each candidate convective cell based on the comparison results of the straight-line wind score, the rotational wind score, and the corresponding threshold: if the straight-line wind score is not less than the corresponding threshold... If the rotating wind score is less than the corresponding threshold, the hazardous wind type of the corresponding candidate convective cell is linear. If the rotating wind score is not less than the corresponding threshold and the linear wind score is less than the corresponding threshold, the hazardous wind type of the corresponding candidate convective cell is rotating. If both the linear wind score and the rotating wind score are not less than the corresponding threshold, the hazardous wind type of the corresponding candidate convective cell is mixed. If both the linear wind score and the rotating wind score are less than the corresponding threshold, the hazardous wind type of the corresponding candidate convective cell does not meet the classification conditions. The simulation path selection module is used to selectively trigger the non-large eddy simulation path processing module, the WRF-LES condition triggering module, or both simultaneously, based on the hazardous wind type of the candidate convective cell. When the hazardous wind type is straight, the non-large eddy simulation path processing module is triggered; when the hazardous wind type is rotating, the WRF-LES condition triggering module is triggered; and when the hazardous wind type is mixed, both the non-large eddy simulation path processing module and the WRF-LES condition triggering module are triggered simultaneously. The non-large eddy simulation path processing module is used to, when the type of severe gale is straight or mixed, call the convective-scale numerical simulation results according to the preset straight-type refinement flag in step S109, or perform a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km, perform a straight-type gale diagnosis on the candidate convective cells based on the called simulation results, and output a straight-type gale risk product including the landing area, maximum wind speed and impact period to the result output module; this module and the straight-type branch of the mixed type do not start WRF-LES, and whether the rotating branch of the mixed type starts WRF-LES is determined by the WRF-LES condition triggering module; The WRF-LES condition triggering module is used to determine tornado potential triggering when the hazardous wind type is rotating or mixed. This is done based on the rotating wind parameters of the candidate convective cells. Specifically, it calculates the normalized storm relative helicity h, the normalized 0-1 km vertical wind shear s, the normalized lower-level maximum vertical vorticity z, the normalized lifting condensation height l, and the overlapping rate R of the rotating updraft. It then determines whether z and R reach thresholds and the number of parameters in h, s, and l that reach these thresholds. When the tornado potential triggering conditions are met, the adaptive WRF-LES execution module generates large eddy simulation (LES) task parameters and performs LES simulation on the candidate convective cells. The tornado-like vortex objective identification module then performs objective identification of tornado-like vortices based on the LES simulation results and outputs tornado-like vortex potential products to the result output module. When the tornado potential triggering conditions are not met, the non-triggering processing module outputs rotating wind potential monitoring products to the result output module. The adaptive WRF-LES execution module generates nested regions and model parameters and performs large eddy simulation only when the triggering conditions are met, according to step S110. The tornado-like vortex objective recognition module is used to perform rotational structure tracking and combined threshold determination on the continuous output of the adaptive WRF-LES execution module according to step S112; The untriggered processing module is used to output the rotating high wind potential monitoring product according to step S111. The results output module is used to output the classification results, simulation path, linear results, four normalized values, R and parameters that did not reach the threshold, WRF-LES task parameters and tornado-like vortex recognition results according to step S113.

[0075] The data flow relationships between the modules are as follows: the data input module connects to the quality control and format conversion module; the quality control and format conversion module connects to the analysis field and convective scale simulation module; the analysis field and convective scale simulation module connects to the severe wind classification module; the severe wind classification module connects to the non-large eddy simulation path processing module and the WRF-LES condition triggering module through the simulation path selection module; when neither score reaches the threshold, the simulation path selection module directly connects to the result output module; the WRF-LES condition triggering module connects to the adaptive WRF-LES execution module and the tornado-like vortex objective identification module when the conditions are met; and each branch finally connects to the result output module.

[0076] The following specific examples will demonstrate the effectiveness of the method proposed in this embodiment.

[0077] like Figure 3 As shown, a typical straight-line strong wind event on May 15, 2021, is used as an example. Figure 3The image shows high-resolution combined reflectance (in dBZ) and ground wind field (in m / s) at two times (1 km and 333.33 m) at 1240 UTC and 1310 UTC. Figure 3 (a) and Figure 3 (b) shows the results for 1240 UTC, where Figure 3 (a) Resolution is 1km. Figure 3 (b) The resolution is 333.33 m; Figure 3 (c) and Figure 3 (d) shows the results for 1310 UTC, where Figure 3 (c) Resolution is 1km. Figure 3 (d) Resolution of 333.33 m. The red area in the figure represents the target area. The bow-shaped echo structure and the wind field convergence and enhancement characteristics near the gust front can be clearly seen from the figure. The 333.33 m resolution result more finely depicts the strong gradient region and surface strong wind region of the echo front than the 1 km resolution, verifying the effectiveness of sub-kilometer scale non-LES simulation for diagnosing straight-line strong winds.

[0078] like Figure 4 As shown, taking a typical rotating gale event on June 23, 2016 as an example, the system calculates the maximum vertical vorticity at the lower level, the relative helicity of the storm, the vertical wind shear from 0 to 1 km, and the overlap rate of the rotating updraft in the convective-scale results. When the rotating gale score reaches 0.55, the system normalizes H, S, Z, and L and directly calculates R; only when z is not less than 0.60, R is not less than 0.50, and at least two of h, s, and l are not less than 0.60, does the system generate WRF-LES mission parameters based on the position and speed of the candidate rotating cell and start the refined simulation. Figure 4 This demonstrates the evolution of radar reflectivity and the lowest-level wind field in the model at a height of 1.5 km over a region with 3 km resolution. Figure 4 (a) is 0620 UTC. Figure 4 (b) is 0630 UTC. Figure 4 (c) is 0640 UTC. Figure 4 (d) is 0650 UTC. Figure 4 (e) is 0810 UTC. Figure 4 (f) is 0820 UTC. Figure 4 (g) is 0830 UTC. Figure 4 (h) is 0840 UTC. This convection scale result is used for classification and triggering determination, and is not directly used as a conclusion on the formation of tornado-like vortices.

[0079] like Figure 5As shown, in region D05 at a resolution of 111.11 m, the radar reflectivity factor (contours), the lowest-level wind field (approximately 33 m high, arrow), and the integral ascending helicity UH (filled in, not less than 800 m) of cell 1 at 0830 UTC on June 23, 2016, at an altitude of 1.5 km are displayed. 2 s -2 The figure shows a hook-shaped echo structure, a low-level rotating wind field, and a high UH value region corresponding to the center of the low-level cyclonic circulation, indicating that WRF-LES can resolve strong rotating structures with a resolution of hundreds of meters. Whether it is determined to be a tornado-like vortex is still determined according to the combination of cyclonic closed circulation, duration, maximum vertical vorticity, and maximum horizontal wind speed at the lowest level of the model.

[0080] In summary, this invention uses a linear wind score (S1) and a rotating wind score (S2) to classify hazardous winds, converting the classification results into non-large eddy simulation paths, rotating trigger paths, simultaneous execution of both paths, or no simulation path. In rotating trigger paths, the arithmetic mean of the four normalized values ​​and R is not calculated; instead, WRF-LES is activated only when the normalized value of the maximum vertical vorticity at the lower level, the overlap rate of the rotating updraft, and at least two environmental parameters reach thresholds. The innovations of this invention focus on classification discrimination, simulation path selection, condition-triggered WRF-LES, and the integration of its system modules. The innovative points are summarized as follows: 1. The S1 and S2 dual-scoring models, which are respectively oriented towards linear and rotational mechanisms, are adopted, and the combination relationship between the two scores and the threshold is used to determine whether the linear, rotational, mixed, or unmet classification conditions are met.

[0081] 2. Establish a fixed correspondence between the fractal results and the simulation path: linear fractals only call non-large eddy simulations, rotational fractals enter the trigger judgment, and hybrid fractals execute both branches simultaneously. If the fractal conditions are not met, the simulation path is not entered.

[0082] 3. The triggering condition is that the normalized value of the minimum vertical vorticity z in the lower layer is not less than 0.60, the overlap rate of the rotating updraft R is not less than 0.50, and at least two of the normalized values ​​of the three environmental parameters are not less than 0.60. WRF-LES is only activated for rotating candidate cells that simultaneously meet the lower-level rotating structure condition and the environmental condition.

[0083] 4. Calculate the predicted center of the nested region based on the center and moving speed of the candidate rotating unit, and generate two-level large eddy simulation tasks of outer and inner layers, so that the triggering conclusions can be directly converted into executable mode configurations.

[0084] 5. The tornado-like vortex identification conclusion is output based on the combination of closed circulation, duration, maximum vertical vorticity and maximum near-surface wind speed of the triggered WRF-LES results.

[0085] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0086] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive simulation scheduling and tornado-like vortex potential identification method for identifying hazardous gale types, characterized in that, Includes the following steps: Acquire the numerical model background field, ground observation data, and Doppler weather radar data for the target area during the target time period, and perform quality control and format conversion on the ground observation data and Doppler weather radar data; Using the numerical model background field as the initial background, ground observation data after quality control and format conversion and Doppler weather radar data are used as observation inputs to generate a variational assimilation analysis field, and convective-scale numerical simulation is performed based on the variational assimilation analysis field. The straight-line wind parameters and rotational wind parameters of each candidate convective cell are extracted from the convective-scale numerical simulation results. The straight-line wind score and the rotating wind score of each candidate convective cell are calculated based on the straight-line wind parameters and the rotating wind parameters, respectively. The hazardous wind type of each candidate convective cell is determined based on the comparison results of the straight-line wind score, the rotating wind score and the corresponding threshold. If the straight-line wind score is not less than the corresponding threshold and the rotating wind score is less than the corresponding threshold, then the hazardous wind type of the corresponding candidate convective cell is straight-line. According to the preset straight-line fine-tuning flag, the convective-scale numerical simulation results or the sub-kilometer-scale non-large eddy simulation results are called. Based on the called simulation results, the candidate convective cell is diagnosed as having a straight-line wind and a straight-line wind risk product is output. Large eddy simulation is not initiated for the candidate convective cell. If the rotating gale score is not less than the corresponding threshold and the straight gale score is less than the corresponding threshold, then the hazardous gale type of the corresponding candidate convective cell is rotating. Tornado potential triggering is determined based on the rotating gale parameters of the candidate convective cell. When the tornado potential triggering condition is met, large eddy simulation task parameters are generated and large eddy simulation is performed on the candidate convective cell. Based on the large eddy simulation results, tornado-like vortex objective identification is performed, and tornado-like vortex potential products are output. When the tornado potential triggering condition is not met, rotating gale potential monitoring products are output. If both the straight-line gale score and the rotating gale score are not less than the corresponding threshold, then the hazardous gale type of the corresponding candidate convective cell is mixed. At the same time, straight-line gale diagnosis and tornado potential triggering discrimination are performed on the candidate convective cell, and the corresponding straight-line gale risk product and rotating gale potential monitoring product or tornado-like vortex potential product are output. If both the straight-line gale score and the rotating gale score are less than the corresponding threshold, the hazardous gale type of the corresponding candidate convective cell is not satisfied with the classification conditions. The straight-line gale diagnosis and the tornado potential triggering judgment are not performed, and the result of not satisfying the classification conditions is output.

2. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification as described in claim 1, characterized in that, The quality control includes checking for missing data, physical extent, and consistency of ground observation data, as well as velocity de-blurring, ground clutter removal, and outlier removal of Doppler weather radar data; the format conversion includes converting ground observation data into the site observation format of the variational assimilation system, and converting radar polar coordinate data in Doppler weather radar data into a radar observation format that the variational assimilation system can read.

3. The adaptive simulation scheduling and tornado-like vortex potential identification method for identifying hazardous gale types according to claim 1, characterized in that, The convective-scale numerical simulation results include a three-dimensional wind field, temperature and humidity field, radar reflectivity factor, vertical velocity, vertical vorticity, storm relative helicity, and 0 to 1 km vertical wind shear. The storm relative helicity and 0 to 1 km vertical wind shear are calculated based on the three-dimensional wind field. Extracting the straight-line and rotational wind parameters for each candidate convective cell from the convective-scale numerical simulation results includes: Connected regions with radar reflectivity factors not less than a preset reflectivity threshold are selected as candidate convective cells. Calculate the major-minor axis ratio, leading-edge curvature, and maximum horizontal reflectivity gradient of the candidate convection cell based on the radar reflectivity factor corresponding to the candidate convection cell. Calculate the near-surface temperature difference based on the 2 m temperature in the temperature and humidity field corresponding to the candidate convective cell; Calculate the maximum divergence and descending airflow in the lower layer based on the three-dimensional wind field corresponding to the candidate convective cells; The major-minor axis ratio, leading edge curvature, maximum horizontal reflectivity gradient, near-surface temperature difference, maximum lower-level divergence, and descending air flow rate of the same candidate convective cell are used as the straight-line wind parameters corresponding to the candidate convective cell. Calculate the maximum vertical vorticity in the lower layer, the vertical wind shear from 0 to 1 km, and the lower layer circulation based on the three-dimensional wind field corresponding to the candidate convective cells. The overlap rate of the rotating updraft is calculated based on the three-dimensional wind field, vertical velocity, and vertical vorticity corresponding to the candidate convective cells. The relative helicity of the storm is calculated based on the three-dimensional wind field corresponding to the candidate convective cell and the moving speed of the candidate cell. The maximum vertical vorticity in the lower layer, the vertical wind shear from 0 to 1 km, the relative helicity of the storm, and the overlap rate of the lower-level circulation and the rotating updraft corresponding to the same candidate convective cell are used as the rotating wind parameters corresponding to the candidate convective cell.

4. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification according to claim 3, characterized in that, The formula for calculating the overlap rate of rotating updrafts is as follows: U k = {(i,j)∈Ω | w(i,j,z k ) ≥ 5 m / s} V k 0 = {(i,j)∈Ω | ζ(i,j,z k ) ≥ ZL} R = max k [A(U k ∩ V k ) / A(V k )] Where R represents the overlap rate of the rotating updraft, k represents the model layer number of the candidate convective cell in the range of 0 to 3 km altitude, and U k V represents the set of grids in which the vertical velocity of the k-th mode layer within a candidate convection cell is not less than a preset vertical velocity threshold; k 0 This represents the set of grids where the vertical vorticity of the k-th model layer within a candidate convective cell is not less than the lower bound of the normalized vertical vorticity of the lower layers, for V k 0 Perform connectivity marking for edge- or corner-connected regions, and select the connected region containing the largest vertical vorticity grid in that layer and with a horizontal area not less than four times the area of ​​the pattern grid as the effective rotation region V. k If no connected region meets the conditions, then V k It is an empty set; max k This indicates that k is the maximum value, A(·) represents the horizontal area of ​​the grid set, (i,j) represents the horizontal grid coordinates within the candidate convective cell, w represents the vertical velocity, ζ represents the vertical vorticity, and z k ZL represents the vertical coordinate of the k-th model layer, and ZL represents the lower bound of the normalized vertical vorticity in the lower layer, ZL = 0.002 + 0.0015(3-Δx), where Δx is the horizontal grid spacing for convective-scale simulation; when V k It is an empty set or A(V) k When )=0, the area ratio of this layer is 0.

5. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification according to claim 1, characterized in that, When calculating the straight-line wind score and the rotational wind score for each candidate convective cell based on the aforementioned straight-line wind parameters and rotational wind parameters, the following steps are included: The straight-line strong wind parameters are normalized according to their corresponding lower and upper limits, as shown in the following formula: p i = min{1, max[0, (x i -a i ) / (b i -a i )]} Where, p i Let x represent the normalized linear wind parameter of the i-th candidate convective cell. i Let a represent the linear wind parameter of the i-th candidate convective cell. i b represents the lower limit of the i-th linear strong wind parameter. i Let b represent the upper limit of the parameter for the i-th linear strong wind, and b i Greater than a i ; The linear gale score is calculated based on preset non-negative weights, using the following formula: S1= (S i w i p i ) / (S i w i ) Where S1 represents the linear wind score of the current candidate convective cell, w i The weights w represent the weights of the i-th linear strong wind parameters. i All are non-negative and Σ i w i Greater than 0; The rotational wind parameters are normalized according to their corresponding lower and upper limits, as shown in the following formula: q j = min{1, max[0, (y j -c j ) / (d j -c j )]} Where, q j y represents the j-th normalized rotating wind parameter of the current candidate convective cell. j c represents the j-th rotational wind parameter of the current candidate convective cell. j d represents the lower limit of the j-th rotating gale parameter. j Let d represent the upper limit of the j-th rotating gale parameter, and d j Greater than c j ; The rotational gale score is calculated based on preset non-negative weights, using the following formula: S2= (Σ j v j q j ) / (Σ j v j ) Where S2 represents the rotational wind score of the current candidate convective cell, v j This represents the weight of the j-th rotating gale parameter, and each v j All are non-negative and Σ j v j Greater than 0.

6. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification according to claim 1, characterized in that, When calling simulation results based on preset linear refinement flags and performing linear high wind diagnosis on the candidate convective cells, the following is included: The system reads a preset linear refinement flag; when the linear refinement flag is 0, it calls the convective-scale numerical simulation results; when the linear refinement flag is 1, it uses the convective-scale numerical simulation results as the initial field and boundary conditions, performs a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km and boundary layer parameterization, and calls the sub-kilometer-scale non-large eddy simulation results; it extracts 10 m wind speed, near-surface temperature difference, lower-level maximum divergence, and descending air flow from the called simulation results. A connected region consisting of grids with wind speeds of 10 m or more not less than the preset gale speed threshold and whose edges or corners are connected is defined as a straight-line gale zone. The maximum wind speed and the period of impact are determined based on the number of consecutive outputs. Output the location, maximum wind speed, and duration of impact of a straight-line gale as a risk product for straight-line gales.

7. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification according to claim 3, characterized in that, The convective-scale numerical simulation results also include the elevation of condensation height. When determining tornado potential triggering based on the rotating wind parameters of the candidate convective cells, the following is included: The storm's relative helicity, vertical wind shear from 0 to 1 km, and maximum vertical vorticity at lower levels were normalized using a positive normalization function N. + The process involves using an inverse normalization function N to adjust the condensation height. - The processing and calculation formula is as follows: h = N + (H;100,250) s = N + (S;7.5,12.5) with = N + (Z;ZL,ZU) l = N - (L;750,1500) Where h, s, z, and l represent the normalized relative helicity of the storm, the normalized vertical wind shear from 0 to 1 km, the normalized maximum vertical vorticity in the lower layer, and the normalized lifting condensation height, respectively; H, S, Z, and L represent the relative helicity of the storm, the vertical wind shear from 0 to 1 km, the maximum vertical vorticity in the lower layer, and the lifting condensation height, respectively; ZL and ZU represent the lower and upper limits of the normalized vertical vorticity in the lower layer, respectively; ZL = 0.002 + 0.0015(3-Δx), ZU = 0.004 + 0.004(3-Δx), where Δx is the horizontal grid spacing of the convective scale simulation. The tornado potential trigger condition is determined to be met when the following conditions are met: z ≥ 0.60 ∧ R ≥ 0.50 ∧ [I(h ≥ 0.60) + I(s ≥ 0.60) + I(l ≥ 0.60) ≥2] Where ∧ represents logical AND, R represents the unnormalized original rotating updraft overlap rate, I is 1 when the condition is met, otherwise it is 0.

8. The adaptive simulation scheduling and tornado-like vortex potential identification method for hazardous gale classification identification according to claim 7, characterized in that, When generating large eddy simulation (LES) task parameters and performing LES simulation on the candidate convective cells, and objectively identifying tornado-like vortices based on the LES simulation results, the following steps are included: The candidate convective cell is denoted as the candidate rotating cell. The current center c0 of the candidate rotating cell is converted into the model projection plane coordinates. Its moving velocity vector in the same projection coordinate system is denoted as v. The nested region prediction center c1 after a predetermined simulation duration T is calculated according to c1 = c0 + vT. The candidate cell moving velocity vectors c and v used for storm relative helicity calculation represent the vectors of the moving velocity of the same candidate cell in the corresponding coordinate system. According to the prediction center c1, an outer large eddy simulation region and an inner eddy analysis region are set. The horizontal resolution of the inner eddy analysis region is higher than that of the outer large eddy simulation region. The reporting time is set to a predetermined time period before the triggering time, and the output interval is set to a predetermined time interval. Boundary layer parameterization and cumulus parameterization are turned off in the large eddy simulation region, and three-dimensional subgrid turbulence closure is adopted. Connectivity region identification is performed on the low-level wind field from 0 to 1 km in consecutive output time intervals; new objects are created for rotating structures that appear for the first time and cannot be associated with existing objects; in adjacent time intervals, rotating structures with a center distance of no more than D are associated with the same object, where D = (Vc + 10 m / s)Δt, Vc is the moving speed of the same rotating object in the previous time interval, and Δt is the output time interval. Let A denote the horizontally connected region of the lowest-level rotational structure of the pattern. c Its closed boundary is denoted as A c According to Γ c = ∮ A c V·dl = A c ζdA calculates the circulation, where ζ represents the vertical vorticity; in the Northern Hemisphere, when Γ c A value greater than 0 indicates that the rotating structure has a cyclonic closed circulation; the maximum near-surface wind speed is taken as A. c The maximum horizontal wind speed at the lowest level within the model range; when the same rotating structure simultaneously exhibits a cyclonic closed circulation, lasting for at least 2 minutes, and with a maximum vertical vorticity of at least 0.20 s. -1 If the maximum horizontal wind speed at the lowest level of the model is not less than 29 m / s, it is determined to be a tornado-like vortex, and the simulation identification conclusion and the corresponding occurrence time, center location, movement path, maximum vertical vorticity and maximum horizontal wind speed at the lowest level of the model are output; otherwise, the evaluation conclusion and corresponding indicators for not meeting the tornado-like vortex standard are output; the above positive or negative simulation identification results together constitute the tornado-like vortex potential product.

9. The adaptive simulation scheduling and tornado-like vortex potential identification method for identifying hazardous gale types according to claim 7, characterized in that, When outputting rotating high wind potential monitoring products, the output includes normalized storm relative helicity, normalized vertical wind shear from 0 to 1 km, normalized minimum vertical vorticity, normalized uplift condensation height, and overlapping rate of rotating updrafts, and marks parameters that do not meet the tornado potential triggering conditions.

10. An adaptive simulation scheduling and tornado-like vortex potential identification system for identifying hazardous gale types, characterized in that, include: The data input module is used to acquire the numerical model background field, ground observation data, and Doppler weather radar data of the target area during the target time period; The quality control and format conversion module is used to perform quality control and format conversion on the ground observation data and Doppler weather radar data. The analysis field and convective-scale simulation module is used to generate a variational assimilation analysis field by taking the numerical model background field as the initial background, the ground observation data after quality control and format conversion and the Doppler weather radar data as the observation input, and to perform convective-scale numerical simulation based on the variational assimilation analysis field. The hazardous wind classification module is used to extract the straight-line wind parameters and rotational wind parameters for each candidate convective cell from the convective-scale numerical simulation results. Based on these parameters, it calculates the straight-line wind score and rotational wind score for each candidate convective cell, and determines the hazardous wind type for each candidate convective cell by comparing the scores with corresponding thresholds. If the straight-line wind score is not less than the corresponding threshold and the rotational wind score is less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is straight-line. If the rotational wind score is not less than the corresponding threshold and the straight-line wind score is less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is rotational. If both the straight-line wind score and the rotational wind score are not less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell is mixed. If both the straight-line wind score and the rotational wind score are less than the corresponding threshold, the hazardous wind type for the corresponding candidate convective cell does not meet the classification conditions. The simulation path selection module is used to selectively trigger the non-large eddy simulation path processing module, the WRF-LES condition triggering module, or both simultaneously, based on the hazardous wind type of the candidate convective cell. When the hazardous wind type is straight, the non-large eddy simulation path processing module is triggered; when the hazardous wind type is rotating, the WRF-LES condition triggering module is triggered; and when the hazardous wind type is mixed, both the non-large eddy simulation path processing module and the WRF-LES condition triggering module are triggered simultaneously. The non-large eddy simulation path processing module is used to call up convective-scale numerical simulation results according to a preset straight-line fine-tuning flag when the type of hazardous gale is linear or mixed. Alternatively, it can use the convective-scale numerical simulation results as the initial field and boundary conditions to perform a sub-kilometer-scale non-large eddy simulation with a preset horizontal grid spacing of less than 1 km. Based on the called simulation results, it performs a straight-line gale diagnosis and outputs a straight-line gale risk product to the result output module. The non-large eddy simulation path processing module and the mixed-type straight-line branch do not activate WRF-LES. Whether the mixed-type rotating branch activates WRF-LES is determined by the WRF-LES condition triggering module. The WRF-LES condition triggering module is used to determine the tornado potential triggering based on the rotating wind parameters of the candidate convective cell when the type of severe gale is rotating or mixed. Based on the determination result, it connects to the adaptive WRF-LES execution module or the non-triggered processing module. An adaptive WRF-LES execution module is used to generate large eddy simulation task parameters and perform large eddy simulation on the candidate convective cell when the tornado potential triggering condition is met. The tornado-like vortex objective identification module is used to perform objective identification of tornado-like vortices based on the large eddy simulation results output by the adaptive WRF-LES execution module, and output the tornado-like vortex potential product to the result output module. The untriggered processing module is used to output the rotating high wind potential monitoring product to the result output module when the tornado potential triggering conditions are not met. The results output module is used to output the classification results and simulated paths of candidate convective cells, and output one or more of the following: linear high wind risk products, rotating high wind potential monitoring products, and tornado-like vortex potential products, or output results that do not meet the classification conditions.