A method, system, and computer equipment for dynamic control of low-altitude jet streams using an interlaced grid.

By identifying target grid points of the low-level jet on the quality grid and back-mapping the wind field adjustment factor to the staggered grid, the coordination problem of low-level jet adjustment in numerical meteorological models is solved, and the stability of simulation results and automated continuous operation are achieved.

CN123117722APending Publication Date: 2026-11-17XUZHOU METEOROLOGICAL BUREAU
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611594197.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-15
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing numerical meteorological models suffer from poor coordination between the staggered grid data structure and the spatial relationship of the wind field in low-level jet regulation, resulting in unstable simulation results, easy interruption of long-term series simulations, and easy introduction of errors by manual operation.

Method used

The target grid points of the low-level jet stream are identified on the mass grid and wind field adjustment factors are generated. These factors are then back-mapped to the horizontal wind speed components on the staggered grid. The task status is monitored and restored through the SLURM job scheduling system to ensure the continuity and automation of the simulation.

Benefits of technology

It reduces additional errors in wind field processing, maintains the consistency of wind field structure, improves the continuous operation capability and automation of long-term series simulation, and reduces the risk of human error.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117722A_ABST
    Figure CN123117722A_ABST
Patent Text Reader

Abstract

This invention relates to the field of numerical meteorological simulation and high-performance computing technology, and discloses a method, system, and computer equipment for dynamic control of low-level jet streams using staggered grids. The method involves reading horizontal wind speed components from two staggered grids and mapping them to a quality grid, identifying target grid points of the low-level jet stream and generating wind field adjustment factors. These adjustment factors are then mapped back to the corresponding staggered grids to adjust the wind speed components and written back to the model data file. Numerical meteorological model integration tasks are submitted according to integration periods, and abnormal retries are performed or the next integration period is advanced using a restart file based on the operational status. This invention can reduce additional errors caused by grid transformation, help maintain the spatial relationship of the wind field, and improve the automation level and anomaly recovery capability of continuous integration tasks.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of numerical meteorological simulation and high-performance computing technology, specifically to a method, system, and computer equipment for dynamic control of low-level jet streams using staggered grids. Background Technology

[0002] Low-level jet streams are important weather systems influencing regional heavy precipitation events. They can affect the location, intensity, and duration of precipitation by continuously transporting water vapor, enhancing low-level convergence, and altering vertical wind shear. In numerical meteorology studies, to analyze the effects of low-level jet streams on specific precipitation events, sensitivity experiments are typically conducted to weaken, enhance, or locally adjust the low-level wind field in the model, and then compare the adjusted simulation results with control experiments.

[0003] Existing numerical meteorological models typically use staggered grids to store dynamic and scalar variables. Taking an Arakawa C-type staggered grid model as an example, east-west wind speed components, north-south wind speed components, and mass variables such as pressure and temperature are located at different grid positions. Identification of the low-level jet generally requires conditions such as composite wind speed, wind direction, and pressure layer, and the composite wind speed usually needs to be calculated at a unified mass grid location. Therefore, after completing the identification of the low-level jet, it is also necessary to solve the problem of how the identification results on the mass grid correspond to the horizontal wind speed components on the two staggered grids.

[0004] Some existing processing methods first interpolate the wind speed components from two staggered grids to the mass grid, directly modify the wind speed on the mass grid, and then interpolate the modified wind field back to the staggered grid. This process requires multiple wind field interpolations, which can easily introduce additional spatial smoothing errors and may alter the original spatial correspondence between the two horizontal wind speed components. In particular, when the wind speed in the target area is uniformly replaced with a fixed value, it may weaken the velocity gradient, wind direction differences, and shear structure within the original wind field, resulting in a significant discontinuity between the adjusted local wind field and the surrounding wind field, thereby affecting the stability of subsequent model integration and simulation results.

[0005] Furthermore, some low-level jet sensitivity experiments only modify the wind field once in the initial model field. As the numerical weather model continues to integrate, boundary conditions, dynamic processes, and physical processes constantly alter the model wind field, and the initial conditioning effect may gradually weaken, making it difficult to maintain the predetermined low-level jet conditioning conditions over a longer simulation period. Manually extracting and modifying restart files repeatedly between multiple integration periods is not only labor-intensive but also prone to errors due to inconsistencies in file selection, simulation time, and restart parameter settings.

[0006] Long-term series simulations typically rely on computing clusters to perform continuous calculations across multiple integration periods. Numerical meteorological models may experience task failures, computing node malfunctions, timeouts, or integration anomalies during operation. Current experimental procedures often require manual checks of task status, determination of valid model input files, adjustment and restart parameters, and resubmission of the task after these anomalies occur. This leads to frequent simulation interruptions and increases the risk of redundant calculations and errors in time-segment transitions.

[0007] Therefore, existing technologies still need to improve the coordination between the low-level jet regulation results and the original dynamic field while maintaining the original staggered grid data structure and wind field spatial relationship of the numerical meteorological model. Furthermore, they need to ensure that model data processing, time-segmented integration, operational status monitoring, anomaly recovery, and advancement of the next integration period can be seamlessly connected. This would reduce errors caused by grid conversion and manual operation, and improve the continuous operation capability and automation level of long-term low-level jet sensitivity experiments. Summary of the Invention

[0008] The technical problem to be solved by this invention is: how to continuously and dynamically adjust the low-level jet stream while maintaining the staggered grid data structure and spatial coordination of the wind field in the numerical meteorological model, and to reliably connect wind field processing, time-segmented integration, operation monitoring and anomaly recovery.

[0009] To address the aforementioned technical problems, this invention provides a method for dynamic control of interlaced grid low-level jet streams, comprising the following steps:

[0010] S1. Obtain the data file of the mode to be processed, and read the first horizontal wind speed component and the second horizontal wind speed component located on the first interlaced grid and the second interlaced grid respectively in the data file of the mode to be processed.

[0011] S2. Map the first horizontal wind speed component and the second horizontal wind speed component to the mass grid to obtain the mass grid wind field; determine the target grid point of the low-level jet from the mass grid according to the preset low-level jet identification conditions, and generate the wind field adjustment factor corresponding to each mass grid grid point.

[0012] S3. Based on the positional correspondence between the quality grid and the first staggered grid and the second staggered grid, the wind field adjustment factor is reverse-mapped to the first staggered grid and the second staggered grid, respectively; the first horizontal wind speed component and the second horizontal wind speed component are adjusted according to the wind field adjustment factor obtained by the reverse mapping, and the adjusted horizontal wind speed component is written back to the mode data file to be processed.

[0013] S4. Determine the model input for the current integration period based on the written-back model data file, execute the numerical meteorological model integration task for the current integration period, and monitor the operational status of the numerical meteorological model integration task.

[0014] S5. When the job status is completed, obtain the restart file output for the current integration period and determine whether the preset simulation end condition is met. If the preset simulation end condition is met, end the numerical meteorological model simulation. If the preset simulation end condition is not met, use the restart file as the pending mode data file for the next integration period and return to execute S1. When the job status is abnormally interrupted, resubmit the numerical meteorological model integration task for the current integration period based on the mode input already determined for the current integration period, and continue to monitor the job status after resubmission.

[0015] Furthermore, the numerical meteorological model is a WRF model, and the model data file to be processed is either a model initial field file or a restart file. The model initial field file is a met_em file, and the restart file is a wrfrst file. When the model data file to be processed is a model initial field file, after writing the adjusted horizontal wind speed component back to the model initial field file, the real.exe program is run to generate the model initial conditions and boundary conditions required for the first integration. When the model data file to be processed is a restart file, the adjusted horizontal wind speed component is written back to the restart file, and the written-back restart file is used as the restart input for the current integration period. The numerical meteorological model integration task includes running the wrf.exe program.

[0016] Further, the first horizontal wind speed component and the second horizontal wind speed component are mapped to the mass grid, including calculating the first center wind speed component on the mass grid according to the following formula. Second center wind speed component :

[0017] ;

[0018] ;

[0019] The first central wind speed component is calculated according to the following formula. Second center wind speed component Transform to Earth coordinate system:

[0020] ;

[0021] ;

[0022] Calculate the composite wind speed using the following formula. :

[0023] ;

[0024] in, and The first horizontal wind speed component adjacent along the interlacing direction on the first interlaced grid. and This refers to the second horizontal wind speed component adjacent to each other along the interlacing direction on the second staggered grid. , For grid indexing, The map projection rotation angle. This represents the east-west wind speed component in the Earth coordinate system. This represents the north-south wind speed component in the Earth coordinate system.

[0025] Furthermore, the mass grid points in each model layer are traversed; when the air pressure corresponding to the mass grid point is ≥700hPa and the composite wind speed is ≥700hPa, the mass grid points are traversed. ≥10m / s and north-south wind speed component When the value is greater than 0, the mass grid point is designated as the target grid point for the low-level jet stream, and the wind field adjustment factor corresponding to the mass grid point is set to 0.2; when the mass grid point does not simultaneously meet the above conditions, the wind field adjustment factor corresponding to the mass grid point is set to 1.0.

[0026] Furthermore, the wind field adjustment factor on the mass grid is interpolated to the first staggered grid and the second staggered grid, respectively, to obtain the adjustment factor of the first staggered grid. Second staggered grid adjustment factor The first staggered grid represents the first horizontal wind speed component. Where The grid, the second staggered grid represents the second horizontal wind speed component. Where Grid; using four-point equal-weight cross interpolation, for the first horizontal wind speed component Where Grid position, based on the four surrounding it. The second horizontal wind speed component at the grid points constructs the cross-interpolation component. ; second horizontal wind speed component Where Grid position, based on the four surrounding it. The first horizontal wind speed component at the grid point constructs the cross-interpolation component. And calculate according to the following formula:

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] In the four-point equal-weight cross interpolation, the weights of each adjacent grid point are all equal. According to the first horizontal wind speed component With cross-interpolation components Calculate the composite wind speed of the first staggered grid. Based on cross-interpolation components With the second horizontal wind speed component Calculate the composite wind speed of the second staggered grid. Adjust the first and second horizontal wind speed components according to the following formula:

[0032] ;

[0033] ;

[0034] in, and These represent the first and second horizontal wind speed components before adjustment, respectively. for The second horizontal wind speed component on the grid is Four equally weighted cross-interpolation components at grid locations. for The first horizontal wind speed component on the grid is Four equally weighted cross-interpolation components at grid locations. and These are the composite wind speeds from the first staggered grid and the second staggered grid, respectively. and These represent the adjusted first and second horizontal wind speed components, respectively, with 10.0 representing the preset target wind speed in m / s. By converting the identification results on the mass grid into adjustment factors and then mapping these adjustment factors back to the original staggered grid where the horizontal wind speed components reside, and simultaneously calculating the composite wind speed at the corresponding staggered grid location using the four-point equal-weight cross-interpolation results of another horizontal wind speed component, the additional errors introduced by the repeated conversion between the mass grid and the staggered grid for the complete wind field can be reduced.

[0035] Furthermore, a namelist.input file corresponding to the current integration period is generated based on the current simulation time. During the first integration period, the restart parameter in the namelist.input file is set to .false; during subsequent integration periods, the restart parameter is set to .true.

[0036] The numerical weather model integration task is submitted via the sbatch interface of the SLURM job scheduling system, and the corresponding job identifier is obtained. The job status of the numerical weather model integration task is queried based on the job identifier at a preset polling interval via the sacct interface of the SLURM job scheduling system. The job status includes running status, completed status, task failure status, node fault status, and timeout status.

[0037] When the job status is running, continue querying the job status; when the job status is completed, obtain the restart file output for the current integration period; when the job status is a task failure, node failure, or timeout, resubmit the numerical meteorological model integration task for the current integration period after a preset waiting time. Specifically, if the current integration period is the first integration period, the restart parameter is kept at .false, and the task is resubmitted based on the model initialization conditions and boundary conditions; if the current integration period is not the first integration period, the restart parameter is set to .true, and the task is resubmitted based on the restart file used for the current integration period.

[0038] Furthermore, when the data file to be processed is a restart file, the horizontal wind speed component read includes the first horizontal wind speed component of the previous time layer. Second horizontal wind speed component The first horizontal wind speed component of the current time layer Second horizontal wind speed component and the first horizontal wind speed component located on the mass grid. Second horizontal wind speed component .

[0039] A wind field processing script is generated based on the current simulation time and corresponds to the current integration period. This script reads and adjusts the horizontal wind speed components, writes the adjusted components back to the restart file, and then deletes the script after the write-back is complete. For and The mass grid wind field adjustment factor is used. To make adjustments, first calculate the composite wind speed using the mass grid. :

[0040] ;

[0041] when When the value is 0, it will be used for denominator calculation. The value is set to the preset positive wind speed protection value. Then adjust according to the following formula:

[0042] ;

[0043] ;

[0044] The present invention also provides a dynamic control system for interlaced grid low-altitude jet streams, including a wind field processing module, a task management module, an anomaly recovery module, and a time period advancement module.

[0045] The wind field processing module is used to acquire the data file of the mode to be processed, extract two horizontal wind speed components located on the first staggered grid and the second staggered grid respectively, map the two horizontal wind speed components to the quality grid to identify the target grid point of the low-level jet and generate the wind field adjustment factor, reverse map the wind field adjustment factor to the first staggered grid and the second staggered grid to adjust the two horizontal wind speed components, and write the adjusted horizontal wind speed components back to the data file of the mode to be processed.

[0046] The task management module is connected to the wind field processing module and the computing cluster respectively. It is used to determine the model input for the current integration period based on the model data file to be processed after writing back, submit the numerical meteorological model integration task for the current integration period to the computing cluster based on the model input, and obtain the operation status of the integration task and the restart file output when the integration task is completed.

[0047] The anomaly recovery module is connected to the task management module and is used to trigger the task management module to resubmit the numerical meteorological model integration task for the current integration period based on the mode input that has been determined for the current integration period when the operation status is in an abnormal interruption state.

[0048] The time period advancement module is connected to the task management module and the wind farm processing module respectively. When the operation status is completed and the preset simulation end conditions are not met, the module obtains the restart file of the integral task output from the task management module, uses the restart file as the pending mode data file for the next integral time period, and sends it to the wind farm processing module.

[0049] Furthermore, the wind farm processing module includes an initial wind farm processing unit and a restart wind farm processing unit; the task management module includes a configuration generation unit, a task submission unit, and a status acquisition unit connected in sequence; the computing cluster is configured with a SLURM job scheduling system, the task submission unit communicates with the SLURM job scheduling system through the sbatch interface, and the status acquisition unit communicates with the SLURM job scheduling system through the sacct interface; the restart wind farm processing unit includes a script generation subunit, a script execution subunit, and a script deletion subunit connected in sequence.

[0050] The present invention also provides a computer device, including a processor, a memory, and a communication interface; the memory is connected to the processor and stores a computer program; the communication interface is connected to the processor and is used to communicate with a computing cluster running a numerical meteorological model; when the processor executes the computer program, it causes the computer device to execute the above-mentioned staggered grid low-level jet dynamic control method.

[0051] First, instead of directly modifying the two central wind speed components on the mass grid and then back-inserting the complete wind field onto the staggered grid, this invention completes the identification of the low-level jet stream and the generation of adjustment factors on the mass grid, and then maps the adjustment factors in reverse to the staggered grid where the two horizontal wind speed components originally reside. This reduces the additional smoothing error introduced by multiple grid transformations of the complete wind field and maintains the correspondence between the wind speed components and the corresponding staggered grid positions.

[0052] Secondly, this invention coordinates and adjusts the two horizontal wind speed components separately based on the adjustment factor after reverse mapping, the wind speed synthesized by the staggered grid, and the preset target wind speed, instead of uniformly replacing the wind speed in the target area with a single fixed value. Therefore, it is beneficial to reduce the disturbance of wind field adjustment on the original wind field direction information, velocity gradient, and spatial variation relationship, and reduce the sudden changes in wind field between the target area and the surrounding area.

[0053] Furthermore, this invention treats both the initial field file of the model and the restart file output during each integration period as processable objects. After each integration period is completed, the low-level jet stream identification and wind field adjustment are re-executed on the restart file used in the next integration period, so that the predetermined low-level jet stream control conditions can be continuously implemented along the simulation time axis, avoiding the gradual weakening of the adjustment effect as the integration process occurs when only the initial field is modified.

[0054] Furthermore, this invention seamlessly integrates the generation of mode configuration files, submission of integration tasks, query of job status, resubmission of abnormal tasks, and advancement of the next integration period. After a task fails, a node malfunctions, or a timeout occurs, the effective mode input of the current integration period can be used to re-execute the task, reducing operational errors caused by manual querying, file selection, and restarting parameter settings, and improving the automation level and continuous operation capability of long-term low-altitude jet stream sensitivity tests. Attached Figure Description

[0055] Figure 1 A flowchart of a method for dynamic control of low-altitude jet streams using an interlaced grid, provided by the present invention;

[0056] Figure 2 The structural block diagram of the staggered grid low-altitude jet stream dynamic control system provided by the present invention;

[0057] Figure 3This is a map showing the cumulative rainfall distribution obtained from the controlled experiment simulation from 02:00 on June 17, 2020 to 14:00 on June 18, 2020. Different colors represent different levels of cumulative rainfall, and the unit is mm.

[0058] Figure 4 The cumulative rainfall distribution map is a simulation obtained from 02:00 on June 17, 2020 to 14:00 on June 18, 2020 after the low-level jet stream was regulated using the scheme of this invention. Different colors represent different levels of cumulative rainfall, and the unit is mm.

[0059] Figure 5 This is a simulated cumulative rainfall distribution map obtained from 02:00 on June 17, 2020 to 14:00 on June 18, 2020, after adjusting the low-level jet stream using a fixed wind speed method. Different colors represent different levels of cumulative rainfall, and the unit is mm. Detailed Implementation

[0060] Example 1

[0061] This embodiment provides a method for dynamic control of low-level jet streams using staggered grids. The numerical meteorological model used is the Weather Research and Forecasting Model (WRF model). The model's dynamic framework uses staggered grids to store horizontal wind speed components and a mass grid to store pressure, temperature, and other scalar variables. The first horizontal wind speed component is located in the first staggered grid, and the second horizontal wind speed component is located in the second staggered grid. The specific positional relationships between the first staggered grid, the second staggered grid, and the mass grid can be achieved using the existing Arakawa C-type staggered grid structure of the WRF model.

[0062] For ease of explanation in the following formulas, the first staggered grid will be denoted as... The grid, denoted as the second staggered grid. Grid; the above notation is only used to characterize the different staggered grid locations where the two horizontal wind speed components are located.

[0063] like Figure 1 As shown, before starting the numerical weather model simulation, the simulation start time, simulation end time, and integration period length are first determined. In this embodiment, the integration period length is set to 60 minutes to allow for the re-identification of the low-level jet stream and wind field adjustment using the restart file for the next integration period after each integration period ends. The integration period length can also be set according to the simulation area, grid resolution, and experimental requirements.

[0064] During the initial integration, the model data file to be processed is the model initial field file, which is a met_em file in NetCDF format. After obtaining the met_em file, the first and second horizontal wind speed components located on the first and second staggered grids, respectively, are read. The first horizontal wind speed component corresponds to the east-west model wind speed variables set in an alternating manner, and the second horizontal wind speed component corresponds to the north-south model wind speed variables set in an alternating manner.

[0065] Since the first and second horizontal wind speed components are located at different grid positions, the composite wind speed cannot be directly calculated at the same grid point. Therefore, the two horizontal wind speed components are first mapped to the mass grid. The first central wind speed component on the mass grid. Determined according to the following formula:

[0066] (1)

[0067] Second center wind speed component on the mass grid Determined according to the following formula:

[0068] (2)

[0069] In the formula, The first central wind speed component on the mass grid; The second central wind speed component on the mass grid; and These are two adjacent first horizontal wind speed components along the interlacing direction on the first staggered grid. and These are two adjacent second horizontal wind speed components along the interlacing direction on the second staggered grid. , These are the grid indices in two horizontal directions. The specific index directions of the first and second interlaced grids are determined according to the dimension definitions in the pattern data file.

[0070] After completing the center position mapping, the wind speed components in the pattern grid coordinate system are transformed to the Earth coordinate system according to the map projection rotation angle. The east-west wind speed components in the Earth coordinate system... Determined according to the following formula:

[0071] (3)

[0072] North-South Wind Speed ​​Components in Earth Coordinate System Determined according to the following formula:

[0073] (4)

[0074] In the formula, This represents the east-west wind speed component in the Earth coordinate system. This represents the north-south wind speed component in the Earth coordinate system. The map projection rotation angle is the direction of the pattern grid relative to the Earth's latitude and longitude directions. and These are the cosine and sine values ​​of the map projection rotation angle, respectively. The map projection rotation angle or its cosine and sine values ​​can be read directly from the model data file, or they can be obtained according to the model projection parameters using existing coordinate transformation methods.

[0075] Composite wind speed at mass grid points Determined according to the following formula:

[0076] (5)

[0077] In the formula, The summative wind speed at each grid point of the mass grid is expressed in m / s.

[0078] Then, the mass grid points in each model layer are traversed. For any mass grid point, when the corresponding air pressure is ≥700hPa and the composite wind speed is... ≥10m / s and north-south wind speed component When the value is greater than 0, the grid point is designated as a target grid point for the low-level jet stream. If a mass grid point does not simultaneously meet the above conditions, it is designated as a non-target grid point.

[0079] Based on the grid point identification results, wind field adjustment factors corresponding to each mass grid point are generated. The wind field adjustment factors on the mass grid are denoted as... For low-altitude jet stream target grid points, Set to 0.2; for non-target grid points, Set to 1.0. This creates a wind field adjustment factor field that corresponds one-to-one with the spatial location of the wind field in the mass grid.

[0080] The main processing parameters used in this embodiment are shown in Table 1 below.

[0081] Table 1. Low-level jet stream identification and adjustment parameters

[0082]

[0083] After obtaining the wind field adjustment factor on the mass grid, the wind field adjustment factor is interpolated to the first staggered grid according to the positional correspondence between the mass grid and the first staggered grid to obtain the adjustment factor of the first staggered grid. Based on the positional correspondence between the mass grid and the second staggered grid, the wind field adjustment factor is interpolated to the second staggered grid to obtain the second staggered grid adjustment factor. .

[0084] The inverse mapping of wind field adjustment factors can be achieved using linear interpolation consistent with the corresponding staggered direction. Any adjustment factor on the first staggered grid is determined by the wind field adjustment factors of the adjacent mass grid points on either side of that location, and any adjustment factor on the second staggered grid is determined by the wind field adjustment factors of the adjacent mass grid points on either side of that location. Boundary locations can be handled using nearest neighbor extension, one-sided interpolation, or existing pattern boundary processing methods, the latter belonging to existing technologies for staggered grid data processing.

[0085] Meanwhile, the first horizontal wind speed component With the second horizontal wind speed component Located respectively Grid and The grids, and the grid points of the two do not coincide. (This is in the context of...) Grid position and The corresponding composite wind speed is obtained at each grid location. In this embodiment, four-point equal-weight cross interpolation is used. Specifically, in the first horizontal wind speed component... corresponding At the grid location, around that location four The second horizontal wind speed component at the grid points is taken as the arithmetic mean to obtain the cross-interpolated component. ; second horizontal wind speed component corresponding At the grid location, around that location four The arithmetic mean of the first horizontal wind speed component at each grid point is used to obtain the cross-interpolated component. Each quantity is determined according to the following formula:

[0086] (6)

[0087] (7)

[0088] (8)

[0089] (9)

[0090] In the formula, For the second horizontal wind speed component Four equally weighted cross-interpolation components at grid locations. The first horizontal wind speed component is Four equally weighted cross-interpolation components at the grid location; the weights of the four adjacent grid points are all equal. ; and They are respectively Grid position and Composite wind speed at grid locations.

[0091] Based on the wind field adjustment factor obtained from the inverse mapping and the composite wind speed from the staggered grid, the two horizontal wind speed components are synchronously adjusted. The adjusted first horizontal wind speed component... Determined according to the following formula:

[0092] (10)

[0093] Adjusted second horizontal wind speed component Determined according to the following formula:

[0094] (11)

[0095] In the formula, The first horizontal wind speed component before adjustment; The second horizontal wind speed component before adjustment; This refers to the adjusted first horizontal wind speed component; This is the adjusted second horizontal wind speed component; The first staggered grid adjustment factor; This is the adjustment factor for the second staggered grid. The wind speed is the composite wind speed of the first staggered grid. 10.0 represents the composite wind speed of the second staggered grid; 10.0 represents the preset target wind speed in m / s.

[0096] When the first interlaced grid adjustment factor When the value is 1.0, the adjusted first horizontal wind speed component remains the original first horizontal wind speed component; when the second staggered grid adjustment factor is 1.0, When the value is 1.0, the adjusted second horizontal wind speed component remains the same as the original second horizontal wind speed component. For target areas where the wind field adjustment factor is less than 1.0, the two horizontal wind speed components are coordinated and adjusted according to their original values, the corresponding staggered grid adjustment factors, and the staggered grid composite wind speed, thereby reducing the disturbance to the original wind field direction information, velocity gradient, and spatial shear relationship caused by the unified assignment method.

[0097] In the program implementation, for non-target locations where the wind field adjustment factor is equal to 1.0, the original horizontal wind speed component is directly copied, and the ratio term in equation (10) or equation (11) is not calculated; for locations where the wind field adjustment factor is less than 1.0, since the corresponding grid points meet the identification condition that the composite wind speed is greater than or equal to 10 m / s, the composite wind speed of the first staggered grid is calculated. Second staggered grid combined wind speed All values ​​are positive. This avoids the problem of the denominator being 0.

[0098] After completing the adjustment, the first horizontal wind speed component Second horizontal wind speed component Write back the met_em file. Opening the schema data file, reading variables, matching dimensions, overwriting variables, and closing the file can be achieved using existing NetCDF data processing techniques.

[0099] After writing the adjusted horizontal wind speed components back to the model initial field file, run the real.exe program to generate the model initial conditions and boundary conditions required for the first integration. The generated model initial conditions and boundary conditions can use the existing wrfinput and wrfbdy files of the WRF model, and the relevant file structure and generation process adopt existing WRF preprocessing techniques.

[0100] The task management program is then initialized. This program can be implemented using Python and generates a `namelist.input` file corresponding to the current integration period based on the simulation start time, simulation end time, and integration period length. During the first integration period, the restart parameter in the `namelist.input` file is set to `.false`; after the first integration period, the restart parameter is set to `.true`.

[0101] The task management program submits numerical weather model integration tasks (running the wrf.exe program) to the computing cluster via the sbatch interface of the SLURM job scheduling system, and records the job identifier returned by the SLURM job scheduling system. After task submission, the task management program queries the job status corresponding to the job identifier via the sacct interface at preset polling intervals.

[0102] When the job status is RUNNING, it indicates that the integration task for the current integration period is still running, and the task management program continues to query the job status according to the preset polling interval. When the job status is COMPLETED, it indicates that the integration task for the current integration period has been completed normally, and the task management program obtains the wrfrst restart file output for the current integration period.

[0103] When a job status is FAILED, NODE_FAIL, or TIMEOUT, the job is classified as an abnormal interruption. FAILED indicates that the integration task has failed, NODE_FAIL indicates that the compute node has failed, and TIMEOUT indicates that the task execution time has exceeded the set time limit. The task management program waits 30 seconds after detecting an abnormal interruption status before triggering the abnormal recovery process.

[0104] If an abnormal interruption occurs during the first integration period, the restart parameter in the namelist.input file is kept at .false, and the numerical weather model integration task for the current integration period is resubmitted based on the already generated model initial conditions and boundary conditions. If an abnormal interruption occurs during any integration period after the first integration period, the restart parameter is set to .true, and the numerical weather model integration task for the current integration period is resubmitted based on the restart file of the previous completed integration period with wind field adjustment.

[0105] The restart file from the previous completed integration period, after wind field adjustment, is the valid mode input file for the current integration period. Therefore, the anomaly recovery process will not reprocess all completed simulation periods, but will re-execute from the valid input file corresponding to the current integration period. After resubmitting the task, the job status is queried again through the sacct interface until the current integration period is completed normally.

[0106] After the current integration period completes normally, the task management program retrieves the wrfrst restart file output for the current integration period and determines whether the current simulation time has reached the preset simulation end time. If the current simulation time has reached the preset simulation end time, the numerical meteorological model simulation ends; if the current simulation time has not yet reached the preset simulation end time, the wrfrst restart file output for the current integration period is used as the pending mode data file for the next integration period, and the program returns to execute low-level jet stream identification and wind field adjustment.

[0107] When the data file to be processed is the wrfrst restart file, read the first horizontal wind speed component from the previous time layer. Second horizontal wind speed component The first horizontal wind speed component of the current time layer Second horizontal wind speed component and the first horizontal wind speed component located on the mass grid. Second horizontal wind speed component .

[0108] in, and This is to restart the horizontal wind speed components of the staggered grid from the previous time layer in the file; and To restart the horizontal wind speed components of the staggered grid at the current time layer in the file; and This refers to the horizontal wind speed components calculated by the model and located on the quality grid in the wrfrst restart file. For different versions of numerical weather models, the specific names of the variables can correspond to the actual variable names in the model restart file, but the grid location and physical meaning of each variable remain consistent.

[0109] The task management program dynamically generates the NCL wind field processing script corresponding to the current integration period based on the current simulation time. This script can be named modify_uvrst_{time}.ncl, where time represents the simulation time identifier corresponding to the current integration period.

[0110] The dynamically generated wind farm processing script reads the wrfrst restart file. , , , , and The above wind speed variables are then subjected to mass grid mapping, coordinate transformation, low-level jet stream target grid point identification, wind field adjustment factor generation, staggered grid reverse mapping, and wind speed component coordination adjustment.

[0111] for , , and The processing is performed using adjustment factors on the first and second interlaced grids. For and The mass grid wind field adjustment factor is used. Perform the corresponding processing. Let... and The mass grid synthesized wind speed is ,but:

[0112] (12)

[0113] First, in the program... Perform non-zero protection; when When, the denominator will be used for calculation. Set the value to m / s. Then, according to the following formula... and Adjustments are needed:

[0114] (13)

[0115] (14)

[0116] In the formula, and These are the adjusted horizontal wind speed components of the mass grid. Because... and Using the same quality grid scaling factor for adjustment allows it to be coordinated with the adjustment process of the staggered grid dynamic variables.

[0117] After writing the adjusted horizontal wind speed components back to the wrfrst restart file, the script execution subroutine closes the restart file, and the task management program deletes the temporary NCL wind field processing script corresponding to the current integration period. The generation, execution, and deletion of temporary scripts are all automatically completed by the task management program according to the current simulation time, without the need for manual hourly modification.

[0118] After completing the restart file processing, the current simulation time is advanced by 60 minutes. The namelist.input file is regenerated based on the advanced simulation time, with the restart parameter set to .true. The processed restart file is then used as the mode input file for the next integration time period. This process is executed cyclically along the simulation timeline, forming a closed loop of wind field processing, task submission, status monitoring, anomaly recovery, or time period advancement, until the preset simulation end time is reached.

[0119] Optionally, the low-level jet stream identification threshold, preset target wind speed, and wind field adjustment factor can be set according to the research object. For example, the degree of weakening or strengthening can be determined based on the deviation between the wind profiler radar observation results and the model wind field. This embodiment uses a parameter combination of air pressure ≥700hPa, composite wind speed ≥10m / s, north-south wind speed component >0, target grid wind field adjustment factor of 0.2, and preset target wind speed of 10.0m / s.

[0120] To verify the method of this embodiment, a heavy rainstorm weather event in a certain region from June 17 to June 18, 2020, was simulated. The concentrated precipitation period was from 02:00 on June 17, 2020 to 14:00 on June 18, 2020. A control experiment, a sensitivity experiment using the method of this embodiment, and a fixed wind speed sensitivity experiment were set up for this weather event.

[0121] The control test does not artificially adjust the low-level jet stream wind field. The sensitivity test using the method of this embodiment identifies and adjusts the low-level jet stream according to equations (1) to (14) and the above dynamic control process. The fixed wind speed sensitivity test sets the grid point wind speed that meets the low-level jet stream identification condition to 12 m / s.

[0122] Depend on Figures 3 to 5 It can be seen that after adjusting the wind field that meets the low-level jet stream identification conditions in different ways, the simulated precipitation amount and precipitation location both respond, and the spatial morphology of precipitation obtained by different wind field adjustment methods are significantly different. Figure 4 The proposed scheme first identifies target grid points of the low-level jet stream on the mass grid and generates wind field adjustment factors. Then, these adjustment factors are back-mapped to the first and second staggered grids, respectively, to adjust the two horizontal wind speed components accordingly. In this scheme, the main rain belt shifts slightly northward, and heavy precipitation is mainly located near the low-level shear line; in contrast, Figure 5 The traditional scheme shown sets the grid wind speed to 12 m / s for all points that meet the conditions of low-level jet stream, which reduces wind shear inside the jet stream and causes the rainband to be distributed in a blocky pattern.

[0123] Furthermore, the time consumed by the NCL scripts for modifying the wrfrst wind field using the method described in this embodiment and the traditional method (modifying wind speed on the quality grid and then interpolating back to the WRF staggered grid to modify the wrfrst wind field) for one time run were statistically analyzed. The algorithm in this embodiment takes 5.69 seconds to adjust the low-level jet stream for one time run, while the traditional algorithm takes 14.69 seconds. The algorithm in this embodiment is approximately 2.58 times faster than the traditional algorithm. For long-term series simulations requiring continuous processing of multiple restart files on an hourly or shorter time interval basis, this single-time processing time difference can accumulate with the number of integration periods, thereby shortening the file processing waiting time between adjacent integration periods and improving the continuous operation efficiency of the low-level jet stream dynamic control closed loop. It should be noted that the above times only represent the running efficiency of the wind field processing script and do not represent the total running time of the entire numerical meteorological model integration task.

[0124] The corresponding test results are shown in Table 2 below.

[0125] Table 2 Comparison of different wind field treatment methods

[0126]

[0127] The above experiments verified this embodiment from the perspectives of spatial response to precipitation and single-time file processing efficiency. After the experiments were completed, the restart file output by the current integration period automatically entered the wind field processing process of the next integration period according to the simulation time; when the integration task was abnormally interrupted, the task was resubmitted based on the valid mode input file corresponding to the current integration period, without changing the previously completed integration results. Thus, the integration periods are continuously connected through restart files, and the entire simulation process is jointly controlled by the current simulation time and the preset simulation end time.

[0128] Optionally, for numerical meteorological models employing other staggered grid structures, the variable reading, file writing-back, and task submission methods can be adapted based on the actual grid locations of the horizontal wind speed component and mass variable in the model, as well as the initial field file, restart file, and task startup interface. Corresponding model file processing, boundary location interpolation, and job scheduling can utilize the existing processing methods of the numerical meteorological model and the computing cluster.

[0129] In summary, this embodiment uniformly identifies target grid points for the low-level jet stream on the mass grid and back-maps the wind field adjustment factors generated from the identification results to the staggered grids containing the two horizontal wind speed components, enabling coordinated adjustment of the two horizontal wind speed components at their respective native grid locations. Simultaneously, by restarting file loop processing, monitoring the status of the integration task, and resubmitting abnormal tasks, the low-level jet stream adjustment can be continuously implemented along the simulation time axis. This implementation method reduces the data processing volume caused by repeated interpolation of the complete wind field, minimizes the disturbance of wind field adjustment to wind direction, wind speed gradient, and shear structure, and improves the continuous operation capability of long-term sensitivity experiments.

[0130] Example 2

[0131] This embodiment provides a dynamic control system for low-altitude jet streams using an interlaced grid.

[0132] like Figure 2 As shown, the system includes a wind field processing module, a task management module, an anomaly recovery module, and a time period advancement module, and is connected to a computing cluster running a numerical meteorological model.

[0133] The wind farm processing module is connected to the task management module. It is used to acquire the data file of the mode to be processed and send the written data file to the task management module. The wind farm processing module includes an initial wind farm processing unit and a restart wind farm processing unit.

[0134] The initial field processing unit is used to read and modify the model initial field file in NetCDF format. In this embodiment, the model initial field file is a met_em file. The initial field processing unit reads the horizontal wind speed components on the first and second staggered grids, generates a mass grid wind field as described in Embodiment 1, identifies the target grid points of the low-level jet stream, generates a wind field adjustment factor, maps the wind field adjustment factor back to the two staggered grids, and writes the adjusted horizontal wind speed components back to the model initial field file.

[0135] The restart field processing unit is used to read and modify the wrfrst restart file in NetCDF format. The restart field processing unit reads the horizontal wind speed components of the interleaved grid of the previous time layer and the current time layer, as well as the horizontal wind speed components on the quality grid, and processes the horizontal wind speed components in accordance with the method described in Example 1.

[0136] The task management module connects to the wind field processing module, the anomaly recovery module, the time period advancement module, and the computing cluster. The task management module receives model data files output by the wind field processing module, generates a model configuration file for the current integration period, submits the numerical meteorological model integration task for the current integration period to the computing cluster, and receives the operation status and integration results returned by the computing cluster.

[0137] The task management module comprises a configuration generation unit, a task submission unit, and a status acquisition unit, which are connected in sequence. The output of the configuration generation unit is connected to the input of the task submission unit. The task submission unit is connected to the computing cluster through a job scheduling interface, and the status acquisition unit is connected to the computing cluster through a job scheduling interface.

[0138] The configuration generation unit generates the namelist.input file based on the current simulation time, simulation end time, and integration period length. During the initial integration, the configuration generation unit sets the restart parameter to .false; for subsequent integrations, it sets the restart parameter to .true.

[0139] The task submission unit receives the namelist.input file generated by the configuration generation unit and the model data file output by the wind farm processing module. When the wind farm processing module outputs the initial model field file, it first runs the real.exe program based on the processed initial model field file to generate the initial and boundary conditions of the model, and then submits the first integration task running the wrf.exe program to the computing cluster through the sbatch interface. When the wind farm processing module outputs a restart file, it uses the processed restart file as the restart input for the corresponding integration period and submits the integration task running the wrf.exe program through the sbatch interface. The task submission unit saves the job identifier returned by the job scheduling system and sends the job identifier to the status acquisition unit.

[0140] The status acquisition unit queries the job status of the integration task through the sacct interface and the job identifier. When the status is RUNNING, the status acquisition unit continues to poll; when the status is COMPLETED, it retrieves the wrfrst restart file output by the current integration period; and when the status is FAILED, NODE_FAIL, or TIMEOUT, it sends an exception interruption message to the exception recovery module.

[0141] The anomaly recovery module is connected to the task management module. It receives anomaly interruption information from the status acquisition unit and reads the valid mode input file and current mode configuration file corresponding to the current integration period. After a preset waiting time, the anomaly recovery module sends a resubmission command to the configuration generation unit and the task submission unit.

[0142] For the initial integration period, the anomaly recovery module keeps the restart parameter of the configuration generation unit at .false and triggers the task submission unit to resubmit the task based on the mode initial conditions and boundary conditions. For subsequent integration periods, the anomaly recovery module keeps or resets the restart parameter of the configuration generation unit to .true and triggers the task submission unit to resubmit the current integration period task based on the adjusted restart file output from the previous completed integration period.

[0143] The time-segment advancement module is connected to both the task management module and the wind field processing module. After obtaining the COMPLETED status, the status acquisition unit sends the completion status, the simulation time corresponding to the current integration period, and the output restart file to the time-segment advancement module.

[0144] The time-period advancement module determines whether the current simulation time has reached the preset simulation end time. When the preset simulation end time is reached, the time-period advancement module outputs a simulation end signal. When the preset simulation end time has not been reached, the time-period advancement module advances the current simulation time by one integration period length and sends the restart file output in the current integration period as the pending mode data file for the next integration period to the wind farm processing module.

[0145] After the wind farm processing module completes the restart file processing for the next integration period, it sends the processed restart file to the task management module. Thus, Figure 2 The time period advancement module and the task management module establish a completion status and restart file transfer relationship, and the time period advancement module and the wind farm processing module establish a data file transfer relationship for the next integral time period pending processing mode.

[0146] The restart field processing unit comprises a script generation subunit, a script execution subunit, and a script deletion subunit connected in sequence. The script generation subunit generates the corresponding NCL wind field processing script based on the current simulation time and the restart file name, and sends the script path to the script execution subunit.

[0147] The script execution subunit calls the NCL runtime environment to execute the wind farm processing script, which is related to the restart file. , , , , and The script performs reading, identification, adjustment, and writing back. After the script execution is complete and the file is closed upon restart, the script execution subunit sends an execution completion signal to the script deletion subunit.

[0148] The script deletion subunit deletes the temporary wind farm processing script corresponding to the current integration period based on the execution completion signal, and sends the path to the restart file after processing to the output of the wind farm processing module. If the temporary script fails to execute, the original restart file is retained, and the task management module can regenerate the script based on the execution log and execute it again.

[0149] The configuration generation unit, task submission unit, status acquisition unit, exception recovery module, and time-segment progression module can be implemented by the processor executing Python programs. The initial field processing unit and restart field processing unit can be implemented by the processor executing NCL data processing programs. The Python program and the NCL data processing program can interact with each other via file paths, command-line arguments, return codes, and log files.

[0150] The system can be deployed on the login node of a computing cluster, a standalone task management server, or a computer device that can access the shared file system of the computing cluster. The wind farm processing module, task management module, anomaly recovery module, and time period advancement module can be deployed on the same computer device or on multiple interconnected computing nodes.

[0151] When modules are deployed on different nodes, model data files and restart files can be transferred via a shared file system, while job status and control commands can be transmitted via a network communication interface. The shared file system, network communication protocols, and node authentication methods can all be implemented using existing technologies in high-performance computing systems.

[0152] During system operation, the wind field processing module first completes the low-level jet stream adjustment of the initial field of the model; for the first integration, the task management module runs real.exe based on the processed initial field of the model to generate the initial conditions and boundary conditions of the model, and then submits the first integration task and monitors the operation status; when the operation is abnormal, the abnormal recovery module triggers the resubmission of the current integration period; when the operation is completed but the simulation end time has not been reached, the period advancement module sends the restart file to the wind field processing module; the wind field processing module processes the restart file and sends it to the task management module again.

[0153] Through the aforementioned module locations and connections, wind field physical constraint processing and computing cluster task scheduling are both independent and interconnected. When an integration task in the computing cluster encounters an anomaly, it is not necessary to re-execute all completed integration periods; when the integration task completes normally, the output restart file can automatically enter the wind field processing process for the next integration period, thereby improving the continuous operation capability of long-sequence sensitivity experiments.

[0154] Example 3

[0155] This embodiment provides a computer device, including a processor, a memory, and a communication interface. The memory is connected to the processor and stores a computer program. The communication interface is connected to the processor and is used to communicate with a computing cluster running a numerical meteorological model. When the processor executes the computer program, it acquires the model data file to be processed, completes the reading of horizontal wind speed components, mass grid mapping, identification of low-level jet target grid points, generation of wind field adjustment factors, reverse mapping of staggered grids, and writing back of horizontal wind speed components. Based on the processed model data file, it determines the model input for the current integration period and submits the numerical meteorological model integration task for the current integration period to the computing cluster through the communication interface.

[0156] The processor is also used to acquire the job status of the integration task. When the job status is in an abnormal interruption state, it resubmits the integration task for the current integration period. When the job status is in a completed state and the preset simulation end condition is not met, it uses the restart file output by the current integration period as the pending mode data file for the next integration period and continues to execute the low-altitude jet stream control process until the preset simulation end condition is met. Thus, the computer equipment can realize the continuous execution of mode data processing, integration task submission, job status monitoring, abnormal recovery, and time period advancement.

Claims

1. A method for dynamic control of low-level jet streams using an interlaced grid, characterized in that, include: S1. Obtain the data file of the mode to be processed, and read the first horizontal wind speed component and the second horizontal wind speed component located on the first interlaced grid and the second interlaced grid respectively in the data file of the mode to be processed; S2. Map the first horizontal wind speed component and the second horizontal wind speed component to the mass grid to obtain the mass grid wind field. Determine the target grid point of the low-level jet from the mass grid according to the preset low-level jet identification conditions, and generate the wind field adjustment factor corresponding to each mass grid grid point. S3. Based on the positional correspondence between the mass grid and the first staggered grid and the second staggered grid, the wind field adjustment factor is reverse-mapped to the first staggered grid and the second staggered grid respectively. The first horizontal wind speed component and the second horizontal wind speed component are adjusted according to the wind field adjustment factor obtained by the reverse mapping, and the adjusted horizontal wind speed component is written back to the mode data file to be processed. S4. Determine the model input for the current integration period based on the written-back model data file, execute the numerical meteorological model integration task for the current integration period based on the model input, and monitor the operational status of the numerical meteorological model integration task. S5. When the job status is completed, obtain the restart file output during the current integration period and determine whether the preset simulation end conditions are met; When the preset simulation termination conditions are met, the numerical meteorological model simulation ends; when the preset simulation termination conditions are not met, the restart file is used as the model data file to be processed in the next integration period, and the process returns to execute S1. When the operation status is abnormally interrupted, the numerical meteorological model integration task for the current integration period is resubmitted based on the model input already determined for the current integration period, and the operation status after resubmission is monitored.

2. The method for dynamic control of low-altitude jet streams using staggered grids according to claim 1, characterized in that: The numerical weather model is WRF mode. The model data files to be processed are either the model initial field file or the restart file. The model initial field file is the met_em file, and the restart file is the wrfrst file. When the model data file to be processed is the model initial field file, after writing the adjusted horizontal wind speed components back to the model initial field file, run the real.exe program to generate the model initial conditions and boundary conditions required for the first integration. When the data file to be processed is a restart file, the adjusted horizontal wind speed component is written back to the restart file, and the restart file written back is used as the restart input for the current integration period; The numerical weather model integration task includes running the wrf.exe program.

3. The method for dynamic control of low-altitude jet streams using staggered grids according to claim 1 or 2, characterized in that, In step S2, the first horizontal wind speed component and the second horizontal wind speed component are mapped to the mass grid to obtain the mass grid wind field, including: The first central wind speed component on the mass grid is calculated using the following formula. Second center wind speed component : ; ; The first central wind speed component is calculated according to the following formula. and the second central wind speed component Transform to Earth coordinate system: ; ; Calculate the composite wind speed using the following formula. : ; in, and The adjacent first horizontal wind speed components on the first staggered grid. and The adjacent second horizontal wind speed components on the second staggered grid. , For grid indexing, The map projection rotation angle. This represents the east-west wind speed component in the Earth coordinate system. This represents the north-south wind speed component in the Earth coordinate system. and, and Adjacent along the intersection direction of the first staggered grid, and Adjacent along the intersection direction of the second staggered grid.

4. The method for dynamic control of low-altitude jet streams using staggered grids according to claim 3, characterized in that, In step S2, determining the target grid points of the low-level jet stream according to preset low-level jet stream identification conditions and generating the wind field adjustment factor includes: Traverse the quality grid points in each mode layer; When the air pressure corresponding to the mass grid point is ≥700 hPa and the composite wind speed is... ≥10m / s and north-south wind speed components When the value is greater than 0, the mass grid point is determined as the target grid point of the low-level jet stream, and the wind field adjustment factor corresponding to the mass grid point is set to 0.

2. If a mass grid point does not simultaneously meet the above conditions, the wind field adjustment factor corresponding to that mass grid point is set to 1.

0.

5. The method for dynamic control of low-altitude jet streams using staggered grids according to claim 3, characterized in that, Step S3 includes: The wind field adjustment factor on the mass grid is interpolated to the first staggered grid and the second staggered grid, respectively, to obtain the adjustment factor of the first staggered grid. Second staggered grid adjustment factor ; The first interlaced grid represents the first horizontal wind speed component. Where The grid, the second staggered grid represents the second horizontal wind speed component. Where Grid; Using four-point equal-weight cross interpolation, the first horizontal wind speed component Where Grid position, based on the four surrounding it. The second horizontal wind speed component at the grid points constructs the cross-interpolation component. ; second horizontal wind speed component Where Grid position, based on the four surrounding it. The first horizontal wind speed component at the grid point constructs the cross-interpolation component. And calculate according to the following formula: ; ; ; ; In the four-point equal-weight cross interpolation, the weights of each adjacent grid point are all equal. According to the first horizontal wind speed component With cross-interpolation components Calculate the composite wind speed of the first staggered grid. Based on cross-interpolation components With the second horizontal wind speed component Calculate the composite wind speed of the second staggered grid. Adjust the first and second horizontal wind speed components according to the following formula: ; ; in, and These represent the first and second horizontal wind speed components before adjustment, respectively. for The second horizontal wind speed component on the grid is Four equally weighted cross-interpolation components at grid locations. for The first horizontal wind speed component on the grid is Four equally weighted cross-interpolation components at grid locations. and These are the composite wind speeds from the first staggered grid and the second staggered grid, respectively. and These are the adjusted first and second horizontal wind speed components, respectively, with 10.0 representing the preset target wind speed in m / s.

6. The method for dynamic control of low-level jet streams using staggered grids according to claim 2, characterized in that, In steps S4 and S5: Generate the namelist.input file corresponding to the current integration time period based on the current simulation time; During the initial points period, the restart parameter in the namelist.input file is set to .false., and during subsequent points periods, the restart parameter is set to .true. The numerical weather model integration task is submitted through the sbatch interface of the SLURM job scheduling system, and the corresponding job identifier is obtained. The job status of the numerical weather model integration task is queried based on the job identifier through the sacct interface of the SLURM job scheduling system according to a preset polling interval. The job status includes running status, completion status, task failure status, node failure status, and timeout status; When the job status is a task failure status, node failure status, or timeout status, after a preset waiting period, if the current integration period is the first integration period, the restart parameter is kept at .false., and the numerical meteorological model integration task for the current integration period is resubmitted based on the model initial conditions and boundary conditions. If the current integration period is not the first integration period, the restart parameter will be set to .true., and the numerical meteorological model integration task for the current integration period will be resubmitted based on the restart file of the previous completed integration period output and wind field adjustment.

7. The method for dynamic control of low-altitude jet streams using staggered grids according to claim 2, characterized in that, When the data file to be processed is a restart file, the horizontal wind speed component read in step S1 includes: The first horizontal wind speed component of the previous time layer Second horizontal wind speed component ; The first horizontal wind speed component of the current time layer Second horizontal wind speed component ;as well as The first horizontal wind speed component located on the mass grid Second horizontal wind speed component ; The wind field processing script is generated based on the current simulation time and corresponds to the current integration period. The wind field processing script reads and adjusts the horizontal wind speed components, writes the adjusted horizontal wind speed components back to the restart file, and deletes the wind field processing script after the write-back is completed. for and Let the mass grid wind field adjustment factor be... And calculate the composite wind speed of the mass grid. : ; when When the value is 0, it will be used for denominator calculation. The value is set to the preset positive wind speed protection value. According to the following formula and Coordination and adjustment: ; ; in, and These represent the first and second horizontal wind speed components of the adjusted mass grid, respectively.

8. A staggered grid low-altitude jet stream dynamic control system, characterized in that, include: The wind field processing module is used to acquire the data file of the mode to be processed, extract two horizontal wind speed components located on the first staggered grid and the second staggered grid respectively, map the two horizontal wind speed components to the quality grid to identify the target grid point of the low-level jet and generate the wind field adjustment factor, reverse map the wind field adjustment factor to the first staggered grid and the second staggered grid to adjust the two horizontal wind speed components, and write the adjusted horizontal wind speed components back to the data file of the mode to be processed. The task management module is connected to the wind field processing module and the computing cluster respectively. It is used to determine the model input for the current integration period based on the model data file to be processed after writing back, submit the numerical meteorological model integration task for the current integration period to the computing cluster based on the model input, and obtain the operation status of the integration task and the restart file output when the integration task is completed. The anomaly recovery module, connected to the task management module, is used to trigger the task management module to resubmit the numerical meteorological model integration task for the current integration period based on the mode input already determined for the current integration period when the operation status is in an abnormal interruption state. The time period advancement module is connected to the task management module and the wind farm processing module respectively. When the operation status is completed and the preset simulation end conditions are not met, it obtains the restart file of the integral task output from the task management module, uses the restart file as the pending mode data file for the next integral time period, and sends it to the wind farm processing module.

9. The staggered grid low-altitude jet stream dynamic control system according to claim 8, characterized in that: The wind farm processing module includes an initial field processing unit and a restart field processing unit. The initial field processing unit is used to read and modify the NetCDF format initial field file, and the restart field processing unit is used to read and modify the NetCDF format restart file. The task management module includes a configuration generation unit, a task submission unit, and a status acquisition unit connected in sequence. The configuration generation unit is used to generate a model configuration file corresponding to the current integration period based on the current simulation time. The task submission unit is used to submit the numerical meteorological model integration task to the computing cluster based on the model configuration file. The status acquisition unit is used to obtain the job status of the numerical meteorological model integration task and the restart file output when the task is completed through the job scheduling interface of the computing cluster. The configuration generation unit, task submission unit, and status acquisition unit are implemented by the processor executing Python programs, while the initial field processing unit and restart field processing unit are implemented by the processor executing NCL data processing programs. The computing cluster is configured with a SLURM job scheduling system. The task submission unit communicates with the SLURM job scheduling system through the sbatch interface, and the status acquisition unit communicates with the SLURM job scheduling system through the sacct interface. The restart field processing unit includes a script generation subunit, a script execution subunit, and a script deletion subunit connected in sequence. The script generation subunit is used to generate a wind field processing script corresponding to the current integration period based on the current simulation time. The script execution subunit is used to call the wind field processing script to modify the restart file. The script deletion subunit is used to delete the wind field processing script after the restart file is modified.

10. A computer device, characterized in that, Includes processor, memory, and communication interface; The memory is connected to the processor and stores computer programs; The communication interface is connected to the processor and is used to communicate with the computing cluster running the numerical weather model; When the processor executes the computer program, it causes the computer device to perform the staggered grid low-altitude jet dynamic control method according to any one of claims 1 to 7.