Atmospheric prediction method and device based on meteorological large model and electronic equipment

Abstract

The application provides a meteorological large model-based atmospheric prediction method and device and electronic equipment, and the method comprises the following steps: determining meteorological large model input data based on initial meteorological prediction data; calling a target meteorological large model, determining meteorological large model prediction data based on the meteorological large model input data; calling a target WRF preprocessing system, performing data fusion based on the initial meteorological prediction data and the meteorological large model prediction data to obtain fusion meteorological prediction data in a target region within a target prediction time range, wherein the fusion meteorological prediction data comprises initial boundary conditions in the target region within the target prediction time range; calling a target WRF model, and predicting target meteorological prediction data in the target region within the target prediction time range based on the fusion meteorological prediction data, wherein the target meteorological prediction data comprises target meteorological prediction fields in the target region within each target prediction time step in the target prediction time range. The embodiment of the application can improve the accuracy of meteorological prediction.

Description

Technical Field

[0001] This invention relates to the field of meteorological and environmental forecasting technology, and in particular to an atmospheric forecasting method, device, and electronic equipment based on a large meteorological model. Background Technology

[0002] Currently, air pollution poses a serious threat to public health, the ecological environment, and economic development, and accurate forecasting remains a major challenge in the field of ecological and environmental protection. At present, regional pollution process forecasting is typically achieved using numerical air quality (AQS) forecasting systems. These systems can be composed of global meteorological background fields (such as global forecast data from the GFS (Global Forecast System), regional meteorological models (such as the WRF (Weather Research and Forecasting Model)), regional emission data, and air quality models (such as NAQPMS (Nested Air Quality Prediction Modeling System) and CMAQ (Community Multiscale Air Quality Model)). Correspondingly, air pollution is influenced by both meteorological conditions and emissions. When regional emissions are relatively stable, the accuracy of regional meteorological forecasts directly determines the accuracy of pollution process forecasts. However, current forecasts of some key meteorological elements (such as wind speed and relative humidity) often show significant deviations, resulting in low accuracy and leading to misreporting or underreporting of pollution processes. Therefore, there is currently no good solution for improving the accuracy of weather forecasts. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide an atmospheric forecasting method, device, and electronic equipment based on a large meteorological model to solve the problem of low accuracy in related meteorological forecasts. That is, embodiments of the present invention can determine the forecast data of the large meteorological model through the target large meteorological model, and determine the fused meteorological forecast data through data fusion, so as to provide more accurate initial boundary conditions for subsequent meteorological forecasting processes, thereby obtaining more accurate target meteorological forecast data, which can effectively enhance the forecast accuracy of regional meteorological models and thus effectively improve the accuracy of meteorological forecasts.

[0004] According to one aspect of the present invention, an atmospheric forecasting method based on a large meteorological model is provided, the method comprising: Acquire initial meteorological forecast data within the initial region, where the initial region includes the target region; Based on the initial weather forecast data, the input data for the large-scale meteorological model is determined; and the target large-scale meteorological model is invoked, and based on the input data of the large-scale meteorological model, the forecast data of the large-scale meteorological model is determined. The target WRF preprocessing system is invoked to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data for the target forecast time range in the target area. The fused weather forecast data includes the initial boundary conditions for the target forecast time range in the target area. The target WRF model is invoked, and based on the fused meteorological forecast data, the target meteorological forecast data for the target forecast time range under the target area is forecasted. The target meteorological forecast data includes the target meteorological forecast field for each target forecast time step under the target area within the target forecast time range.

[0005] According to another aspect of the present invention, an atmospheric forecasting device based on a large meteorological model is provided, the device comprising: An acquisition unit is used to acquire initial meteorological forecast data within a target forecast time range in an initial region, wherein the initial region includes the target region. The processing unit is used to determine the input data of the meteorological big model based on the initial meteorological forecast data; and to call the target meteorological big model and determine the forecast data of the meteorological big model based on the input data of the meteorological big model. The processing unit is also used to call the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain the fused weather forecast data of the target forecast time range in the target area. The fused weather forecast data includes the initial boundary conditions of the target forecast time range in the target area. The processing unit is further configured to invoke the target WRF model and, based on the fused meteorological forecast data, forecast the target meteorological forecast data for the target forecast time range within the target area. The target meteorological forecast data includes the target meteorological forecast fields for each target forecast time step within the target forecast time range within the target area.

[0006] According to another aspect of the present invention, an electronic device is provided, the electronic device including a processor and a memory storing a program, wherein the program includes instructions that, when executed by the processor, cause the processor to perform the methods mentioned above.

[0007] According to another aspect of the present invention, a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the methods mentioned above is provided.

[0008] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein the computer program, when executed by a processor, is used to cause the computer to perform the methods mentioned above.

[0009] This invention embodiment can, after obtaining initial meteorological forecast data for the target forecast time range within an initial region, determine the input data for a large-scale meteorological model based on the initial meteorological forecast data; and then call the target large-scale meteorological model, determining the large-scale meteorological model forecast data based on the large-scale meteorological model input data. Then, the target WRF preprocessing system can be called to perform data fusion based on the initial meteorological forecast data and the large-scale meteorological model forecast data, obtaining fused meteorological forecast data for the target forecast time range within the target region. The fused meteorological forecast data includes the initial boundary conditions for the target forecast time range within the target region. Further, the target WRF model can be called to forecast target meteorological forecast data for the target forecast time range within the target region based on the fused meteorological forecast data. The target meteorological forecast data includes the target meteorological forecast fields for each target forecast time step within the target forecast time range within the target region. As can be seen, the embodiments of the present invention can determine the forecast data of the large meteorological model through the target meteorological model, and determine the fused meteorological forecast data through data fusion, so as to provide more accurate initial boundary conditions for the subsequent meteorological forecasting process, thereby obtaining more accurate target meteorological forecast data, which can effectively enhance the accuracy of model forecasts and thus effectively improve the accuracy of meteorological forecasts. Attached Figure Description

[0010] Further details, features, and advantages of the invention are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 A flowchart illustrating an atmospheric forecasting method based on a large meteorological model according to an exemplary embodiment of the present invention is shown. Figure 2 A schematic diagram of a forecasting process according to an exemplary embodiment of the present invention is shown; Figure 3 A flowchart illustrating another atmospheric forecasting method based on a large meteorological model according to an exemplary embodiment of the present invention is shown. Figure 4 A schematic diagram illustrating a comparative analysis of forecast results according to an exemplary embodiment of the present invention is shown; Figure 5 A schematic diagram illustrating another comparative analysis of forecast results according to an exemplary embodiment of the present invention is shown; Figure 6 A schematic diagram illustrating another comparative analysis of forecast results according to an exemplary embodiment of the present invention is shown; Figure 7A schematic block diagram of an atmospheric forecasting device based on a large meteorological model according to an exemplary embodiment of the present invention is shown; Figure 8 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present invention is shown. Detailed Implementation

[0011] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0012] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0013] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0014] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0015] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0016] It should be noted that the execution subject of the atmospheric forecasting method based on a large meteorological model provided in this embodiment of the invention can be one or more electronic devices, and this invention does not limit this. The electronic device can be a terminal (i.e., a client) or a server. Therefore, when the execution subject includes multiple electronic devices, and these multiple electronic devices include at least one terminal and at least one server, the atmospheric forecasting method based on a large meteorological model provided in this embodiment of the invention can be executed jointly by the terminal and the server. Accordingly, the terminal mentioned herein can include, but is not limited to: smartphones, tablets, laptops, desktop computers, smartwatches, smart voice interaction devices, smart home appliances, etc. The server mentioned herein can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, etc.

[0017] Based on the above description, this embodiment of the invention proposes an atmospheric forecasting method based on a large meteorological model. This method can be executed by the aforementioned electronic device (terminal or server); or, it can be executed jointly by a terminal and a server. For ease of explanation, the following description will use the execution of this large meteorological model-based atmospheric forecasting method by an electronic device as an example. Figure 1 As shown, this atmospheric forecasting method based on a large meteorological model may include the following steps S101-S104: S101, Obtain initial weather forecast data within the target forecast time range of the initial region, where the initial region includes the target region.

[0018] Optionally, the target forecast time range can be any time range, and this embodiment of the invention does not limit this; for example, the target forecast time range can be 3 days in the future, 7 days in the future, etc., starting from the target forecast time; correspondingly, the target forecast time can be any forecast time. Optionally, both the initial region and the target region can be any region, and this embodiment of the invention does not limit this; for example, the initial region can be global, and the target region can be one or more cities, etc. Optionally, the target forecast time can also be referred to as the forecast start time.

[0019] Optionally, the initial weather forecast data may include, but is not limited to, at least one of the following: IFS (Integrated Forecasting System, which is the operational integrated forecasting system of ECMWF (European Centre for Medium-Range Weather Forecasts)) forecast data and GFS (Global Forecast System) forecast data, etc.; this embodiment of the invention does not limit this. The accuracy of regional numerical weather prediction largely depends on two main factors: the sophistication of the model itself and the accuracy of the initial boundary conditions. Therefore, for a regional high-resolution model like WRF, the quality of the initial boundary conditions is even the most critical factor determining forecast accuracy. The initial boundary conditions may include initial conditions (i.e., the initial field, which may include the initial fields of each of the multiple target meteorological variables) and boundary conditions (which may include the boundary conditions of each target meteorological variable).

[0020] In this embodiment of the invention, the methods for obtaining initial weather forecast data may include, but are not limited to, the following: The first acquisition method: The electronic device stores the initial weather forecast data within the initial area for the target forecast time range in its own storage space. In this case, the electronic device can obtain the initial weather forecast data within the initial area for the target forecast time range from its own storage space.

[0021] The second acquisition method: The electronic device can download initial weather forecast data for the target forecast time range within the initial region from the global weather forecast system. Optionally, the electronic device may include a dynamic weather forecast data processing module. This module can be based on the Linux (an open-source operating system) platform, consisting of a Python (a programming language) main program and a Shell (an interpreter) scheduling script, to dynamically acquire and process the initial weather forecast data, ensuring that the processed data meets the data requirements driving the large-scale meteorological model (e.g., weather variables, vertical layers, and horizontal resolution). In other words, the electronic device can call the dynamic weather forecast data processing module to download the initial weather forecast data for the target forecast time range within the initial region, thereby achieving the acquisition of initial weather forecast data for the target forecast time range within the initial region. For example, taking IFS forecast data as the initial weather forecast data, the dynamic weather forecast data processing module can be the IFS weather forecast data dynamic processing module. This module primarily relies on earthkit.data, the official open-source library launched by ECMWF and positioned as a "one-stop interface for weather data" (ECMWF's unified data access layer for earth science data), to dynamically acquire and parse IFS GRIB2 (GRIdded Binary version 2, a data format) formatted data, and so on. Based on this, this embodiment of the invention can download initial weather forecast data in real time.

[0022] Optionally, the electronic device can download initial weather forecast data daily to generate rolling weather forecasts.

[0023] S102, based on the initial weather forecast data, determine the input data for the large-scale meteorological model; and call the target large-scale meteorological model, based on the input data of the large-scale meteorological model, determine the forecast data of the large-scale meteorological model.

[0024] Optionally, the electronic device can invoke a dynamic weather forecast data processing module to determine the input data for the large-scale weather model based on the initial weather forecast data. The module can then process the initial weather forecast data to obtain the input data for the large-scale weather model. It should be understood that this embodiment of the invention can drive the target large-scale weather model using forecast data such as IFS (Information Free Surface Data), thereby enabling real-time weather forecasting. This effectively solves the problem of weather forecast lag caused by large-scale weather models driven by ERA5 (fifth-generation atmospheric reanalysis) data.

[0025] Optionally, when determining the input data for the large meteorological model based on initial weather forecast data, the electronic device can determine at least one model meteorological variable and model grid partitioning indication information corresponding to the target large meteorological model. Optionally, the at least one model meteorological variable may include, but is not limited to, at least one of the following: 2m temperature (which can be represented as T2M), the u component (which can be represented as U10) and v component (which can be represented as V10) of 10m wind, mean sea level pressure (which can be represented as MSLP), geopotential (which can be represented as Z), specific humidity (which can be represented as Q), temperature (which can be represented as T), and the u and v components of wind, etc.; this embodiment of the invention does not limit this. Wherein, due east is the positive direction of the u-axis, and due north is the positive direction of the v-axis. Optionally, the model grid partitioning indication information may include, but is not limited to, at least one of the following: vertical layer number and horizontal resolution, etc.; this embodiment of the invention does not limit this. Based on this, the model grid partitioning indication information can be used to indicate multiple model grid points. Optionally, one large meteorological model may correspond to one model grid partitioning indication information.

[0026] Accordingly, the electronic equipment can determine the input data of the model to be processed from the initial weather forecast data according to at least one model meteorological variable. The input data of the model to be processed may include the variable data of each of the at least one model meteorological variable. Optionally, the variable data of a model meteorological variable may include the initial meteorological variable field of the corresponding model meteorological variable at the target start time of the target forecast time range. Here, the target start time of the target forecast time range can also be simply referred to as the start time of the target forecast time range, that is, the start time of the target forecast time range can be the target start time.

[0027] In one implementation, for any model meteorological variable among at least one model meteorological variable, the initial meteorological variable field of any model meteorological variable at the target start time can be determined from the initial meteorological forecast data. The initial meteorological variable field of any model meteorological variable at the target start time may include all grid point values ​​of any model meteorological variable in the initial meteorological forecast data at the target start time. A grid point value of a meteorological variable may refer to the variable value of the corresponding meteorological variable at a grid point.

[0028] In another implementation, the initial weather forecast data may include first initial weather forecast data and second initial weather forecast data; for example, the first initial weather forecast data may be IFS forecast data, and the second initial weather forecast data may be GFS forecast data, etc. Based on this, when determining the model input data to be processed from the initial weather forecast data according to at least one model meteorological variable, the model input data to be processed can be determined from the first initial weather forecast data according to at least one model meteorological variable; in other words, for any model meteorological variable among the at least one model meteorological variable, the initial meteorological variable field of any model meteorological variable at the target forecast time can be determined from the first initial weather forecast data. Optionally, the first initial weather forecast data and the second initial weather forecast data may be the same or different, and this embodiment of the invention does not limit this.

[0029] Furthermore, the electronic device can perform data transformation on the input data of the model to be processed according to the model grid division indication information to obtain the input data of the large meteorological model, thereby determining the input data of the large meteorological model. Based on this, the spatial resolution of the input data of the model to be processed can be converted into the spatial resolution indicated by the model grid division indication information. Optionally, for any model grid point indicated by the model grid division indication information, the model meteorological grid point data of each of the M grid points closest to any model grid point can be determined from the input data of the model to be processed (the model meteorological grid point data of a grid point may include the variable values ​​of each model meteorological variable in at least one model meteorological variable under the corresponding grid point), and the model meteorological grid point data of each of the M grid points is weighted and summed to obtain the model meteorological grid point data of any model grid point (that is, for any model meteorological variable in at least one model meteorological variable, the variable values ​​of any model meteorological variable under each of the M grid points can be weighted and summed to obtain the variable value of any model meteorological variable under any model grid point), thereby determining the input data of the large meteorological model, which may include: the model meteorological grid point data of each model grid point indicated by the model grid division indication information; or, the model meteorological grid point data of the grid point closest to any model grid point can be determined from the input data of the model to be processed, and the determined model meteorological grid point data is used as the model meteorological grid point data of any model grid point, etc.; the embodiments of the present invention do not limit this. Optionally, M can be a positive integer, and the specific value of M is not limited in this embodiment of the invention. It should be noted that the weight of each grid point in the M grid points is not limited in this embodiment of the invention. For example, the model meteorological grid data of each grid point in the M grid points can be averaged to achieve a weighted sum (in which case the weight of each grid point can be the same), or the weight of each grid point in the M grid points can be determined according to the distance between each grid point in the M grid points and any model grid point (e.g., it can be negatively correlated with the distance), and so on.

[0030] Here, the model meteorological grid data for any model grid point can refer to the model meteorological grid data for any model grid point at the target reporting time. The model meteorological grid data for any model grid point at the target reporting time can include the variable values ​​of each model meteorological variable at that model grid point at the target reporting time. Optionally, the variable value of a meteorological variable at a grid point at a given time step can also be referred to as the variable value of the corresponding meteorological variable at the corresponding time step and the corresponding grid point; the variable value of a meteorological variable at a grid point can also be referred to as the variable value of the corresponding meteorological variable at the corresponding grid point, and so on.

[0031] Optionally, the target meteorological large model can be any meteorological large model, and this embodiment of the invention does not limit this; for example, the target meteorological large model can be any of the following meteorological large models: FourCastNet (Fourier Forecasting Neural Network, which can adaptively use Fourier neural operators to complete 14-day global forecasts in seconds), GraphCast (a global medium-term AI (Artificial Intelligence) weather forecasting model based on graph neural network (GNN), which can grid the earth into a multi-level graph and generate 10-day high-precision forecasts using only a single GPU (Graphics Processing Unit), Pangu (an AI meteorological large model that uses hierarchical spatiotemporal coding to achieve the same level of forecast accuracy as the European Centre for Medium-Range Weather Forecasts, requiring only 1 / 1000 of the computing power), Fuxi (another AI meteorological large model that has achieved forecasting capabilities comparable to or even exceeding those of traditional numerical weather prediction on some key meteorological indicators). Optionally, the electronic device may also include a large-scale meteorological model forecasting module. In this case, the electronic device can use the large-scale meteorological model forecasting module to call a target large-scale meteorological model and determine the forecast data based on the input data of the large-scale meteorological model. Optionally, the large-scale meteorological model forecasting module can integrate multiple large-scale meteorological models, thereby identifying a target large-scale meteorological model from among them, calling the target large-scale meteorological model, and determining the forecast data based on the input data of the target large-scale meteorological model. It should be understood that large-scale meteorological models greatly shorten the forecast generation time through end-to-end mapping relationships, making it possible to achieve rapid forecast updates; furthermore, by mining historical patterns, they demonstrate the potential to extend forecast lead time; and in addition, they achieve significant improvements in the forecast accuracy of certain extreme weather events and refined elements (such as precipitation, stationary wind fields, etc.).

[0032] Optionally, in embodiments of the present invention, weather forecasts can be performed by setting external parameters such as input driving data (i.e., meteorological big model input data), forecast start time (e.g., year, month, day, hour) and forecast time (e.g., hours), output path and file name, and the meteorological big model to be called (i.e., setting the target meteorological big model). Among these, the forecast start time and forecast time can be used to indicate the target forecast time range.

[0033] Optionally, the meteorological large model forecast data can be written into the output file in the GRIB format (GRIdded Binary, binary grid data format), such as the meteorological large model forecast data can be in the GRIB2 format. Optionally, the meteorological large model forecast data may include the model meteorological forecast fields (also referred to as meteorological large model forecast fields) at each time step (also referred to as the prediction time step length) within the target forecast time range. The model meteorological forecast field at one time step may include: at the corresponding time step, the variable forecast values of each model meteorological variable at each model grid point; optionally, the model meteorological forecast fields at different time steps can be written into different files (i.e., a file can be created for each time step), or the model meteorological forecast fields at all time steps can be written into one file, and the embodiments of the present invention do not limit this.

[0034] Based on this, the electronic device can input the meteorological large model input data into the target meteorological large model, so as to output the meteorological large model forecast data through the target meteorological large model. Optionally, the meteorological large model forecast data may include the meteorological forecast data in the initial area within the target forecast time range. Optionally, the meteorological large model forecast data may include the meteorological forecast data of each model grid point at each time step within the target forecast time range. The meteorological forecast data of one model grid point at one time step may include the variable values of each model meteorological variable at the corresponding model grid point at the corresponding time step.

[0035] S103, call the target WRF preprocessing system, perform data fusion based on the initial meteorological forecast data and the meteorological large model forecast data, and obtain the fused meteorological forecast data in the target area within the target forecast time range. The fused meteorological forecast data includes the initial boundary conditions in the target area within the target forecast time range.

[0036] Among them, the target WRF preprocessing system can be constructed based on the WPS (WRF preprocessing system, the supporting preprocessing system of the meteorological model) of the WRF. Optionally, the target WRF preprocessing system may include a format decoding module (also denoted as the ungrib module) and a meteorological grid fusion module (also denoted as the metgrib module).

[0037] Based on this, when calling the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data within the target forecast time range of the target area, the format decoding module can be called to parse the initial weather forecast data and the weather big model forecast data respectively to obtain the initial weather forecast parsing data and the weather big model forecast parsing data; and the meteorological grid fusion module can be called to perform data fusion based on the initial weather forecast parsing data and the weather big model forecast parsing data to obtain fused weather forecast data within the target forecast time range of the target area. Among them, the fusion priority of meteorological big model forecast analysis data is higher than that of initial meteorological forecast analysis data. That is, for any target meteorological variable among multiple target meteorological variables (i.e., including all meteorological variables involved in the initial meteorological forecast data), if at least one model meteorological variable includes the target meteorological variable, the data of the target meteorological variable in the meteorological big model forecast analysis data can be fused into the fused meteorological forecast data. If at least one model meteorological variable does not include the target meteorological variable, the data of the target meteorological variable in the initial meteorological forecast analysis data can be fused into the fused meteorological forecast data.

[0038] In this embodiment of the invention, the aforementioned multiple target meteorological variables can be all meteorological variables required to drive the target WRF model. Optionally, the multiple target meteorological variables may include, but are not limited to, at least one of the following: temperature, U and V components of wind, geopotential height (also referred to as geopotential), relative humidity (or specific humidity), pressure, surface pressure, mean sea-level pressure, skin temperature, 2-meter temperature, 2-meter relative humidity, 10-meter U and V components of wind, soil temperature, soil moisture, soil thickness (or terrain height), etc.; this embodiment of the invention does not limit the scope of these variables.

[0039] Optionally, the electronic device can use a format decoding module to read the initial weather forecast data and the large-scale meteorological model forecast data separately, thereby parsing out the variables required to drive the WRF. This allows for the separate parsing of the initial weather forecast data and the large-scale meteorological model forecast data, resulting in parsed initial weather forecast data and parsed large-scale meteorological model forecast data. It should be noted that the parsing process requires preparing a Vtable file (a configuration file, also known as a meteorological variable lookup table or Vtable data interface file) necessary for reading and parsing the large-scale meteorological model forecast data. Optionally, the electronic device can also create a model meteorological variable lookup table according to the command when it detects a command to create a model meteorological variable lookup table for the target large-scale meteorological model. One lookup table is used to map the variable encoding of a background field to the WRF standard variable identifier. Optionally, a background field can be any of the global meteorological numerical model forecast dataset (such as IFS forecast data, GFS forecast data, etc.) and the large-scale meteorological model forecast data. Optionally, a user (such as an administrator) can perform a model meteorological variable reference table upload operation. In this case, the electronic device can detect a model meteorological variable reference table creation instruction for the target meteorological model. The model meteorological variable reference table indicated by the creation instruction can be the same as the one uploaded during the upload operation. Alternatively, a user can perform a model meteorological variable reference table creation operation to set up a model meteorological variable reference table. The electronic device can then use the model meteorological variable reference table set by the creation operation as the model meteorological variable reference table indicated by the creation instruction, and so on. This embodiment of the invention does not limit this. Optionally, when creating a model meteorological variable reference table according to the creation instruction, the model meteorological variable reference table indicated by the creation instruction can be created to achieve the creation of the model meteorological variable reference table according to the creation instruction. Optionally, the variable lookup table corresponding to the initial weather forecast data can be determined from the WRF model, such as the variable lookup table provided by the official WRF organization, like the variable lookup table corresponding to IFS forecast data and / or GFS forecast data, etc. Based on this, a variable lookup table can be used to parse the metadata of GRIB files (such as GRIB2 files), i.e., to parse the corresponding background field. For example, a model weather variable lookup table can be used to parse large-scale weather model forecast data, thereby mapping the variable codes of the large-scale weather model forecast data (such as GRIB2 codes) to WRF standard variable identifiers (such as WRF standard variable names), thus enabling the WRF model to recognize variables in GRIB2 and other files. The GRIB2 code can include field information under each GRIB2 data field.

[0040] Optionally, the model meteorological variable lookup table may include multiple sets of fields, which are not limited in this embodiment of the invention; for example, the multiple sets of fields may include a first set of fields, a second set of fields, and a third set of fields. The first set of fields can be used to describe how to identify data in the GRIB file (i.e., to indicate how to identify data in the GRIB file), the second set of fields can be used to describe how to identify data fields in the metgrid module (i.e., to indicate how to identify data fields in the metgrid module), the third set of fields can be used to provide specific information about the GRIB2 data, and so on. Optionally, the first group of fields may include, but is not limited to: GRIB1 Param (a field used to specify the GRIB code of the weather field), Level Type (a field representing the level type), From Level1 and To Level2 (From Level1 and To Level2 can be used to specify at which levels the field is found, such as for specifying the vertical level range), etc. The second group of fields may include, but is not limited to: metgridName (used to determine the variable name assigned to the weather field when the weather field is written to the intermediate file by the ungrib module (i.e., the variable name assigned to each meteorological variable in the weather field). The variable name assigned here must match the entries in METGRID.TBL (the core configuration table file of the metgrid program in the WPS system) so that the metgrid program can determine how to interpolate horizontally), Metgrid Units (a field used to specify the field units), and metgrid Description (a field used to specify the field description. If the field is not described, it will not be written to the intermediate file), etc. The third group of fields may include, but is not limited to: GRIB2 Discp (the subject of the GRIB2 data), GRIB2 Catgy (the parameter category of the GRIB2 data), GRIB2 The fields include Param (parameter number for GRIB2 data) and GRIB2 Level (field level for GRIB2 data). Optionally, the information in the first and third groups of fields can be found using the WRF-provided utilities g1print.exe (corresponding to GRIB1 format (GRIdded Binary version 1)) and g2print.exe (corresponding to GRIB2 format). g1print.exe and g2print.exe are lightweight command-line tools used in meteorological data processing to parse GRIB format files.

[0041] Based on this, when the format decoding module is invoked to parse the initial weather forecast data and the large-scale meteorological model forecast data respectively, and the initial weather forecast analysis data and the large-scale meteorological model forecast analysis data are obtained, the electronic device can determine the variable lookup table corresponding to the initial weather forecast data and the model meteorological variable lookup table; and invoke the format decoding module to parse the initial weather forecast data and the large-scale meteorological model forecast data respectively according to the variable lookup table corresponding to the initial weather forecast data and the model meteorological variable lookup table, and obtain the initial weather forecast analysis data and the large-scale meteorological model forecast analysis data.

[0042] Optionally, the meteorological grid fusion module can be used to horizontally interpolate the meteorological element fields (i.e., initial weather forecast analysis data and meteorological big model forecast analysis data) extracted by the format decoding module onto the forecast area (i.e., the target area). The meteorological grid fusion module can interpolate and fuse data from multiple data sources. When it is necessary to combine two or more complementary datasets to generate the complete initial boundary condition input data (i.e., initial boundary conditions) required for the WRF model, the meteorological grid fusion module can be used to implement the data fusion function. Optionally, the meteorological grid fusion module may include a meteorological field data name parameter (which can also be represented as fg_name). It should be noted that the meteorological field data names set in the meteorological field data name parameter are set in ascending order according to the corresponding fusion priority. For example, the meteorological field data name of the meteorological big model forecast analysis data is located after the meteorological field data name of the initial weather forecast analysis data. For instance, assuming the target meteorological big model is the Pangu meteorological big model, and the data to be fused includes GFS forecast data and meteorological big model forecast data, then the setting of the meteorological field data name parameter can be represented as "fg_name = 'gfs', "pangu" can be used to represent the name of the meteorological field data in the parsing results of GFS forecast data (i.e., global weather forecast data from GFS), and "pangu" can also represent the name of the meteorological field data in the parsing data of the Pangu meteorological big model (i.e., weather forecast data from the Pangu meteorological big model). This setting indicates that the fusion priority (or simply priority) of the weather forecast data from the meteorological big model is higher. That is, when two meteorological datasets contain the same meteorological variable, the variable data in the meteorological big model dataset (i.e., the weather big model forecast parsing data) will be used first, and so on.

[0043] Optionally, the fused meteorological forecast data may further include the fused background field of each initial forecast time step in the target forecast time range within the target forecast time range under the target region. Each initial forecast time step in the target forecast time range may be any initial forecast time step included in at least one initial forecast time step within the target forecast time range, and an initial forecast time step may be any forecast time step within the target forecast time range that is located after the target start time and has a time step size equal to the initial time step size; wherein, a forecast time step may be simply referred to as a time step. Optionally, the initial time step size may be the interval between any two adjacent initial forecast time steps; optionally, the initial time step size may be set according to experience or actual needs, and this embodiment of the invention does not limit this.

[0044] Based on this, when fusing initial weather forecast analysis data and weather big model forecast analysis data to obtain fused weather forecast data for the target forecast time range within the target area, for any initial forecast time step within the target forecast time range, any target meteorological variable among multiple target meteorological variables, and any grid point among multiple grid points corresponding to the target area (which may include multiple target grid points obtained by gridding the target area according to the target spatial resolution and / or boundary grid points of multiple target grid points, etc.), if any target meteorological variable is a meteorological variable in at least one weather big model variable, then from the weather big model forecast... The analysis data identifies the interpolation variable data (which may include at least one meteorological variable value) for any target meteorological variable at any initial forecast time step and any grid point. If any target meteorological variable is not a meteorological variable in at least one large meteorological model variable, the interpolation variable data for any target meteorological variable at any initial forecast time step and any grid point is determined from the initial meteorological forecast analysis data. Correspondingly, the interpolation variable data can be used to determine the variable value of any target meteorological variable at any initial forecast time step and any grid point, such as by performing mean calculation or weighted summation on the interpolation variable data (i.e., performing interpolation calculation). Optionally, the interpolation variable data may include the variable value of any target meteorological variable at any initial forecast time step at at least one grid point closest to any grid point, etc. It should be noted that the specific process of interpolation calculation is not limited in this embodiment. Optionally, the target spatial resolution can be set according to experience or actual needs, and this embodiment does not limit this.

[0045] Optionally, when the initial weather forecast data includes a first initial weather forecast data and a second initial weather forecast data, the electronic device can call the target WRF preprocessing system to perform data fusion on the second initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data within the target forecast time range of the target area. This enables the calling of the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data within the target forecast time range of the target area. Based on this, when calling the format decoding module to parse the initial weather forecast data and the weather big model forecast data respectively, the electronic device can call the format decoding module to parse the second initial weather forecast data and the weather big model forecast data respectively, obtaining the initial weather forecast analysis data and the weather big model forecast analysis data. At this time, the initial weather forecast analysis data can be the data obtained by parsing the second initial weather forecast data. The initial weather forecast analysis data here can also be called the second initial weather forecast analysis data or the analysis data of the second initial weather forecast data, etc. Correspondingly, when fusing the second initial weather forecast data and the weather big model forecast data to obtain the fused weather forecast data with the target forecast time range in the target area, the second initial weather forecast analysis data and the weather big model forecast analysis data can be fused to obtain the fused weather forecast data with the target forecast time range in the target area. Based on this, the initial weather forecast data used to determine the input data of the model to be processed and the initial weather forecast data used for data fusion can be different, so that multiple data sources can be incorporated to improve the accuracy of weather forecasts.

[0046] It should be understood that when the initial weather forecast data includes only one forecast data (such as IFS forecast data or GFS forecast data), the input data of the large meteorological model and the fused weather forecast data are both determined based on the same meteorological background field, such as both being determined based on IFS forecast data, and so on.

[0047] S104, invoke the target WRF model, based on the fused meteorological forecast data, forecast the target meteorological forecast data within the target area within the target forecast time range, the target meteorological forecast data includes the target meteorological forecast field within the target area for each target forecast time step within the target forecast time range.

[0048] It should be noted that the specific construction of the target WRF mode is not limited in the embodiments of the present invention.

[0049] Optionally, each target forecast time step in the target forecast time range can be any target forecast time step included in at least one target forecast time step within the target forecast time range. A target forecast time step can be any forecast time step within the target forecast time range that is located after the target start time and has a time step size equal to the target time step size. Optionally, the target time step size (i.e., the interval between any two adjacent target forecast time steps) can be set according to experience or actual needs, and this embodiment of the invention does not limit this. Optionally, the at least one initial forecast time step and the at least one target forecast time step can be the same or different, and this embodiment of the invention does not limit this; that is, the initial time step size between any two adjacent initial forecast time steps and the target time step size between any two adjacent target forecast time steps can be the same or different, and this embodiment of the invention does not limit this.

[0050] Optionally, for different global meteorological background fields (such as GFS forecast data, IFS forecast data, etc.), the electronic equipment can also acquire multiple fusion schemes and process and fuse the data according to each fusion scheme to obtain fused meteorological forecast data under each fusion scheme. Then, the target WRF model is invoked, and meteorological forecasts are made based on the fused meteorological forecast data under each fusion scheme, thereby obtaining the target meteorological forecast data under each fusion scheme (i.e., steps S101-S104 are executed according to each fusion scheme). Furthermore, the electronic equipment can also acquire observation data and evaluate the target meteorological forecast data under each fusion scheme based on the observation data, selecting the fusion scheme with the best evaluation result from multiple fusion schemes as the target meteorological forecast scheme. Based on this, subsequent meteorological forecasts can all be made according to the target meteorological forecast scheme.

[0051] Based on this, the embodiments of the present invention conducted a multi-scheme fusion comparison experiment: ① IFS data-driven meteorological big model forecast results + IFS soil data fusion-driven WRF, that is, the initial meteorological forecast data may only include IFS forecast data; ② GFS data-driven meteorological big model forecast results + GFS soil data fusion-driven WRF, that is, the initial meteorological forecast data may only include GFS forecast data; ③ IFS data-driven meteorological big model forecast results + GFS soil data fusion-driven WRF, that is, the initial meteorological forecast data may include a first initial meteorological forecast data and a second initial meteorological forecast data, the first initial meteorological forecast data may be IFS forecast data, the second initial meteorological forecast data may be GFS forecast data, and so on. Therefore, within the same region and time period, weather forecasts are conducted using the various fusion schemes described above. By quantitatively comparing and evaluating the forecasting effectiveness of each scheme for core meteorological elements such as 2m temperature, 2m humidity, and 10m wind field, the "IFS-driven meteorological large model forecast results + GFS soil data fusion-driven WRF" (i.e., the third scheme) is ultimately selected as the target weather forecast scheme (i.e., the optimal weather forecast scheme). In other words, the target weather forecast scheme can be used to indicate: obtaining initial weather forecast data including first and second initial weather forecast data for weather forecasting. Based on this, the embodiments of the present invention can select the best scheme through testing multiple sets of multi-source meteorological background field fusion data, that is, weather forecasting using IFS-driven meteorological large model forecast results + GFS soil data fusion-driven WRF, thereby achieving the best weather forecasting effect.

[0052] It should be noted that this embodiment of the invention integrates meteorological forecast data to drive a target WRF model for regional meteorological forecasting, verifying the forecasting effect of the atmospheric forecasting method based on a large meteorological model proposed in this embodiment. This method can meet the timeliness requirements of operational meteorological forecasts and improve forecast accuracy. Furthermore, the output three-dimensional meteorological field (i.e., target meteorological forecast data) can provide high-fidelity meteorological conditions for subsequent air quality forecasts, thereby improving the forecasting effect of regional air pollution processes. In other words, this embodiment of the invention can further couple and drive a regional air quality model (i.e., an air quality model) to achieve regional pollution process forecasting, ultimately improving the forecasting accuracy of regional pollution processes. For example, as shown... Figure 2 As shown, this embodiment of the invention can fuse initial weather forecast data and large-scale meteorological model forecast data, and then generate initial boundary conditions by fusing the weather forecast data, which can effectively improve the accuracy of the initial boundary conditions. Correspondingly, weather forecasts can be performed by calling the target WRF model using more accurate initial boundary conditions, thereby achieving more accurate weather forecasts. Furthermore, air quality models can be driven by highly accurate target weather forecast data, further improving the accuracy of air quality forecasts. Figure 2The geographic grid generation module in the WRF preprocessing system (WPS) can also be represented as the geogrid module, which can be the first core module of the WRF preprocessing system (WPS); the core configuration file can also be represented as namelist.wps, which can be the core master configuration file of the WRF preprocessing system (WPS), or the WPS parameter list file; optionally, the initial weather forecast data can be IFS forecast data, etc., and the global weather forecast data for data fusion can be IFS, GFS, etc.

[0053] This invention embodiment can, after obtaining initial meteorological forecast data for the target forecast time range within an initial region, determine the input data for a large-scale meteorological model based on the initial meteorological forecast data; and then call the target large-scale meteorological model, determining the large-scale meteorological model forecast data based on the large-scale meteorological model input data. Then, the target WRF preprocessing system can be called to perform data fusion based on the initial meteorological forecast data and the large-scale meteorological model forecast data, obtaining fused meteorological forecast data for the target forecast time range within the target region. The fused meteorological forecast data includes the initial boundary conditions for the target forecast time range within the target region. Further, the target WRF model can be called to forecast target meteorological forecast data for the target forecast time range within the target region based on the fused meteorological forecast data. The target meteorological forecast data includes the target meteorological forecast fields for each target forecast time step within the target forecast time range within the target region. As can be seen, the embodiments of the present invention can determine the forecast data of the large meteorological model through the target meteorological model, and determine the fused meteorological forecast data through data fusion, so as to provide more accurate initial boundary conditions for the subsequent meteorological forecasting process, thereby obtaining more accurate target meteorological forecast data, which can effectively enhance the forecast accuracy of regional meteorological models and improve the accuracy of meteorological forecasts.

[0054] Based on the above description, this embodiment of the invention also proposes a more specific atmospheric forecasting method based on a large meteorological model. Accordingly, this atmospheric forecasting method based on a large meteorological model can be executed by the aforementioned electronic device (terminal or server); or, this atmospheric forecasting method based on a large meteorological model can be executed jointly by a terminal and a server. For ease of explanation, the following description will use the execution of this atmospheric forecasting method based on a large meteorological model by an electronic device as an example; please refer to [link to relevant documentation]. Figure 3 The atmospheric forecasting method based on the large meteorological model may include the following steps S301-S306: S301, Obtain initial weather forecast data within the target forecast time range of the initial region, where the initial region includes the target region.

[0055] S302, based on the initial weather forecast data, determine the input data for the large-scale meteorological model; and call the target large-scale meteorological model, based on the input data of the large-scale meteorological model, determine the forecast data of the large-scale meteorological model.

[0056] S303, call the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data for the target forecast time range in the target area. The fused weather forecast data includes the initial boundary conditions for the target forecast time range in the target area.

[0057] Based on this, embodiments of the present invention can solve the problem that the output variables of the meteorological large model may be insufficient and thus unable to directly drive the regional meteorological model WRF by fusing the meteorological large model with the background field data of conventional global numerical model forecasts (such as GFS forecast data).

[0058] The target forecast time range may include multiple time steps to be traversed, including the target start time and at least one target forecast time step. Optionally, the fused meteorological forecast data can support the determination of the fused background field of each of the multiple time steps to be traversed in the target area. A fused background field includes the variable background field of each of the multiple target meteorological variables. At least one target forecast time step includes all target forecast time steps within the target forecast time range that are after the target start time, and each target forecast time step within the target forecast time range is a target forecast time step among at least one target forecast time step. Optionally, the fused background field of a time step to be traversed in the target area may be empty or not, and this embodiment of the invention does not limit this.

[0059] S304, invoke the target WRF pattern, traverse each of the multiple time steps to be traversed except the last time step to be traversed, and take the currently traversed time step as the current time step.

[0060] S305, for any target meteorological variable among multiple target meteorological variables, determine the variable background field of any target meteorological variable at the current time step from the fused meteorological forecast data, and make a meteorological forecast based on the variable forecast field and variable background field of any target meteorological variable at the current time step, so as to obtain the variable forecast field of any target meteorological variable at the next time step to be traversed, so as to realize the forecast of the target meteorological forecast field in the target area at the next time step to be traversed. The target meteorological forecast field in the target area at a time step includes the variable forecast fields of each target meteorological variable at the corresponding time step among multiple target meteorological variables.

[0061] Optionally, the background field of any target meteorological variable at a time step to be traversed may or may not be empty, and the embodiments of the present invention do not limit this.

[0062] Optionally, if the current time step is the target reporting time, the background field of any target meteorological variable at the current time step can be the initial field of any target meteorological variable, that is, the initial field of any target meteorological variable in the initial boundary conditions, and the initial field of each target meteorological variable is the fused background field of the target reporting time in the target area included in the fused meteorological forecast data; in this case, the forecast field of any target meteorological variable at the current time step is also the initial field of any target meteorological variable.

[0063] Optionally, if the current time step is neither the target start time nor an initial forecast time step, the background field of any target meteorological variable at each of the P interpolation forecast time steps can be determined from the fused meteorological forecast data (e.g., the background field of any target meteorological variable at one interpolation forecast time step can be determined from the fused background field of the corresponding interpolation forecast time step in the target area). Furthermore, based on the background fields of any target meteorological variable at each interpolation forecast time step, the background field of any target meteorological variable at the current time step can be determined, thereby enabling the determination of the background field of any target meteorological variable at the current time step from the fused meteorological forecast data. The variable background field at the previous time step (also known as the variable background field of any target meteorological variable at the current time step based on fused meteorological forecast data); for example, the variable background fields of any target meteorological variable at each interpolation forecast time step can be weighted and summed (e.g., for any grid point among multiple grid points, the variable values ​​of any grid point in the variable background fields at each interpolation forecast time step can be weighted and summed) to obtain the variable background field of any target meteorological variable at the current time step; where P can be a positive integer, and when the value of P is 1, the determined variable background field can be used as the variable background field of any target meteorological variable at the current time step. Optionally, the aforementioned P interpolation forecast time steps may include the P initial forecast time steps with the shortest interval to the current time step; that is, the background field of any target meteorological variable at each of the P initial forecast time steps with the shortest interval to the current time step can be determined from the fused meteorological forecast data, so as to determine the background field of any target meteorological variable at the current time step based on the determined background field, and so on.

[0064] It should be noted that the embodiments of the present invention do not limit the weight of the background field of any target meteorological variable at each interpolation forecast time step. For example, the weight of the background field of any target meteorological variable at each interpolation forecast time step (the weight of the background field of a meteorological variable at one time step can be simply referred to as the weight of the background field at the corresponding time step) can all be 1 / P, meaning the weights can be the same; or, when P is 2, the current time step is located between P interpolation forecast time steps, and the weight of the background field at the interpolation forecast time step closest to the current time step can be: the interval between the current time step and the interpolation forecast time step closest to the current time step / the background field time interval (i.e., the initial forecast duration). In this case, the weight of the background field at the two interpolation time steps is equal to the weight of the background field at each interpolation time step. The weight of the background variable field in another interpolation time step within the forecast time step can be: 1 - the weight of the background variable field in the forecast time step closest to the current time step; or, the initial weight of the background variable field in any interpolation time step can be: the interval between any interpolation time step and the current time step / the sum of the intervals between all interpolation time steps and the current time step. In this case, the normalized result between the initial weights of the background variable fields in each interpolation time step can be used as the weight of the background variable field in each interpolation time step, and so on; the embodiments of the present invention do not limit this. It should be understood that the embodiments of the present invention do not limit the specific value of P.

[0065] Alternatively, if the current time step is neither the target forecast start time nor an initial forecast time step, the background field of any target meteorological variable at the current time step can be empty, and so on. For example, the initial time step can be 6 hours, and the target time step can be 1 hour. In this case, the background field of any target meteorological variable determined every 6 hours at a time step to be traversed can be non-empty, meaning that every 6 hours, the current time step becomes an initial forecast time step, and so on.

[0066] Optionally, if the current time step is an initial forecast time step, the background field of any target meteorological variable at the current time step can be the background field of any target meteorological variable at an initial forecast time step. That is, the background field of any target meteorological variable at the current time step can be determined from the fused background field of the target area at the current time step. In other words, the electronic device can use the background field of any target meteorological variable included in the fused background field of the target area at the current time step as the background field of any target meteorological variable at the current time step.

[0067] Optionally, when forecasting weather based on the variable forecast field and variable background field of any target meteorological variable at the current time step, and obtaining the variable forecast field of any target meteorological variable at the next time step to be traversed at the current time step, for any target grid point in the target area (i.e., any target grid point among multiple target grid points, the target area may include multiple target grid points), the electronic device can calculate the relaxation forcing term of any target meteorological variable at the current time step and any target grid point based on the difference between the variable background value and the variable forecast value of any target meteorological variable at the current time step and any target grid point, and determine the variable forecast value of any target meteorological variable at the next time step to be traversed at the current time step and any target grid point based on the relaxation forcing term and variable forecast value of any target meteorological variable at the current time step and any target grid point, so as to realize the determination of the variable forecast field of any target meteorological variable at the next time step to be traversed at the current time step. Among them, the variable forecast field of a meteorological variable at a time step may include the variable forecast value of the corresponding meteorological variable at the corresponding time step and each target grid point, and the variable background field of a meteorological variable at a time step may include the variable background value of the corresponding meteorological variable at the corresponding time step and each target grid point.

[0068] Correspondingly, when determining the forecast value of any target meteorological variable at the next time step to be traversed and at any target grid point based on the relaxation forcing term and the variable forecast value at the current time step and at any target grid point, at least one physical forcing term in the WRF model can be determined based on the forecast value of any target meteorological variable at the current time step and at any target grid point (such as all physical forcing terms of the model generated by advection, Coriolis force, etc., which can also be referred to as at least one physical forcing term of any target meteorological variable at the current time step and at any target grid point, and one physical forcing term is a physical forcing term in the WRF model). Then, using at least one physical forcing term in the WRF model, the relaxation forcing term and the variable forecast value at the current time step and at any target grid point, the forecast value of any target meteorological variable at the next model integration time step and at any target grid point is calculated. Finally, based on the forecast value of any target meteorological variable at the current time step and at any target grid point, the forecast value of the variable is determined. The forecast value of a target meteorological variable is determined at the next model integration time step and at any target grid point in the current time step. For example, at least one physical forcing term in the WRF model and the relaxation forcing term of any target meteorological variable in the current time step and at any target grid point can be summed to obtain the unit change of any target meteorological variable. The product between the model integration time step (i.e., the time interval for the WRF model to perform one dynamic and physical process calculation, usually in seconds) and the unit change of any target meteorological variable can be used as the forecast value change of any target meteorological variable. Furthermore, the sum of the forecast value change of any target meteorological variable and the variable forecast value of any target meteorological variable in the current time step and at any target grid point can be used as the variable forecast value of any target meteorological variable in the next model integration time step and at any target grid point in the current time step. Optionally, at least one physical forcing term (also referred to as one or more physical forcing terms) may include, but is not limited to, advection term, Coriolis force term, and physical process term, etc., and the embodiments of the present invention do not limit this.

[0069] Optionally, the model integration time step can be set according to experience or actual needs, and this embodiment of the invention does not limit this. Optionally, when determining the variable forecast value of any target meteorological variable at the next model integration time step and at any target grid point based on the variable forecast value of any target meteorological variable at the next model integration time step and at any target grid point in the current time step, the variable forecast value of any target meteorological variable at the next model integration time step and at any target grid point in the current time step can be used as the variable forecast value of any target meteorological variable at the current model integration time step and at any target grid point, and the variable forecast value of any target meteorological variable at the current model integration time step and at any target grid point can be iteratively executed to determine the variable forecast value of any target meteorological variable at the current model integration time step and at any target grid point. The variable forecast values ​​at the next model integration time step and any target grid point of the previous model integration time step are used to determine the variable forecast values ​​at the next target grid point of the current model integration time step. This process continues until the current model integration time step becomes the next time step to be traversed, thereby determining the variable forecast values ​​at the next target grid point of the current model integration time step and any target grid point. (That is, the variable forecast values ​​at the next target grid point of the current model integration time step and any target grid point can be the variable forecast values ​​at the current model integration time step and any target grid point.) In other words, there can be at least one model integration time step between any two adjacent time steps to be traversed. The electronic device can, based on the variable forecast value of any target meteorological variable at the current time step and at any target grid point, sequentially forecast the variable forecast value of any target meteorological variable at each model integration time step and at any target grid point between the current time step and the next time step to be traversed, until the variable forecast value of any target meteorological variable at the next time step to be traversed and at any target grid point is obtained, and so on. For example, assuming there are three model integration time steps between any two adjacent time steps to be traversed, the variable forecast value of any target meteorological variable at the first, second, and third model integration time steps and at any target grid point between the current time step and the next time step to be traversed can be forecast sequentially. Then, based on the variable forecast value of any target meteorological variable at the third model integration time step and at any target grid point, the variable forecast value of any target meteorological variable at the next time step to be traversed and at any target grid point can be determined, and so on. It should be understood that in model forecasting, a target forecast time step can also be a model integration time step. For example, the corresponding model integration time step can be used as the target forecast time step at each target time step.Optionally, the process of determining the forecast value of any target meteorological variable at the next model integration time step and at any target grid point, based on the variable forecast value of any target meteorological variable at the current model integration time step and at any target grid point, can be the same as the process of calculating the forecast value of any target meteorological variable at the next model integration time step and at any target grid point (e.g., the relaxation forcing term of any target meteorological variable at the current model integration time step and at any target grid point can be calculated based on the difference between the background value and the forecast value of any target meteorological variable at the current model integration time step and at any target grid point, and then the forecast value of any target meteorological variable at the next model integration time step can be determined through this relaxation forcing term). (e.g., the predicted value of a variable at a model integration time step and any target grid point). Alternatively, the predicted value of a target meteorological variable at the next model integration time step and any target grid point can be determined solely based on the predicted value of the variable at the current model integration time step and any target grid point. For example, the predicted value of a target meteorological variable at the next model integration time step and any target grid point could be the sum of at least one physical forcing term of the target meteorological variable at the current model integration time step and any target grid point, and the predicted value of the variable at the current model integration time step and any target grid point, etc. The embodiments of the present invention do not limit this. Optionally, the background value of any target meteorological variable at the current model integration time step and at any target grid point can be determined from the background field of the target meteorological variable at the current model integration time step. Optionally, the method for determining the background field of any target meteorological variable at the current model integration time step can be the same as the method for determining the background field of any target meteorological variable at the current time step, which will not be repeated here in this embodiment of the invention. Optionally, when the background field of any target meteorological variable at the current model integration time step is empty, the background value of any target meteorological variable at the current model integration time step and at any target grid point can be zero or empty. In this case, the forecast value of any target meteorological variable at the next model integration time step and at any target grid point can be determined based on the forecast value of any target meteorological variable at the current model integration time step and at any target grid point, and so on.

[0070] Optionally, in other embodiments, at least one physical forcing term, a relaxation forcing term for any target meteorological variable at the current time step and at any target grid point, and the variable forecast value in the WRF mode can be used to calculate the variable forecast value for any target meteorological variable at the next time step to be traversed and at any target grid point in the current time step. In this case, the target forecast time step can correspond one-to-one with the model integration time step, that is, the target forecast duration can be the same as the model integration time step length. In other words, the target WRF mode can output the target meteorological forecast field for each model integration time step in the target forecast time range in the target area, etc.; the present invention does not limit this. It should be understood that when the target forecast time step and the model integration time step do not correspond one-to-one, the target WRF mode can output the target meteorological forecast field for each model integration time step in the target forecast time range in the target area, that is, only output the target meteorological forecast field for each target forecast time step in the target area in the target forecast time range, etc.

[0071] Based on this, the electronic device can determine the variable forecast value of any target meteorological variable at the next time step to be traversed and at any target grid point based on at least one physical forcing term in the WRF mode, the relaxation forcing term of any target meteorological variable at the current time step and at any target grid point, and the variable forecast value.

[0072] Optionally, when calculating the relaxation forcing term of any target meteorological variable at the current time step and any target grid point based on the difference between the background value and the forecast value of the variable at the current time step and any target grid point, the electronic device can determine the approximation weight of any target meteorological variable at the current time step and any target grid point, and determine the relaxation time-scale approximation strength of any target meteorological variable (also known as the control approximation strength of any target meteorological variable for the time scale). Thus, the approximation weight, the relaxation time-scale approximation strength, and the difference between the background value and the forecast value of the variable at the current time step and any target grid point can be multiplied to obtain the relaxation forcing term of any target meteorological variable at the current time step and any target grid point.

[0073] Optionally, the approximation weight of any target meteorological variable at the current time step and at any target grid point can be determined based on the time weight of any target meteorological variable at the current time step (which can be used to control the relaxation intensity of different forecast periods) and / or the spatial weight of any target meteorological variable at any target grid point (which can be used to control the relaxation intensity of different regions). For example, the approximation weight of any target meteorological variable at the current time step and at any target grid point can be the time weight of any target meteorological variable at the current time step, or the approximation weight of any target meteorological variable at the current time step and at any target grid point can be the spatial weight of any target meteorological variable at any target grid point, or the approximation weight of any target meteorological variable at the current time step and at any target grid point can be the product of the time weight of any target meteorological variable at the current time step and the spatial weight of any target meteorological variable at any target grid point, etc.; the embodiments of the present invention do not limit this. Optionally, the temporal weight of a meteorological variable at a time step and the spatial weight of a meteorological variable at a target grid point can both be set according to experience or actual needs, and this embodiment of the invention does not limit this. Correspondingly, the temporal weights of different meteorological variables at a time step and the spatial weights of different meteorological variables at a target grid point can be the same or different, and this embodiment of the invention does not limit this. Furthermore, the temporal weights of a meteorological variable at different time steps and the spatial weights of a meteorological variable at different target grid points can be the same or different, and this embodiment of the invention does not limit this. For example, when any target meteorological variable is wind speed, the spatial weight of any target meteorological variable at each target grid point can be the same, which can achieve "uniform relaxation of the entire domain" for the wind field, thereby ensuring the stability of large-scale circulation. When any target meteorological variable is temperature, the spatial weight of any target meteorological variable at any target grid point can be determined based on the distance between any target grid point and the center point of the target area. This spatial weight can be positively correlated with the distance, thereby achieving a "strong boundary, weak center" spatial weight for the temperature field, so as to retain local thermodynamic characteristics in the central region of the model, and so on. For example, the time weight of any target meteorological variable at the current time step can be determined based on the interval between the current time step and the target reporting time. For instance, it can be negatively correlated with the interval. In this case, the influence of the background field on the forecast field can be gradually reduced over time, thereby improving the forecast accuracy, and so on.

[0074] Optionally, the relaxation timescale approximation intensity of any target meteorological variable can be set according to experience or actual needs, and this embodiment of the invention does not limit this. Correspondingly, the relaxation timescale approximation intensity of different meteorological variables can be the same or different, and this embodiment of the invention does not limit this. Optionally, the relaxation timescale approximation intensity of the first meteorological variable can be greater than that of the second meteorological variable. The first meteorological variable can be a large-scale circulation variable (such as wind speed, air pressure, etc.), and the second meteorological variable can be a thermal or water vapor variable. In this case, the relaxation intensity of the large-scale circulation variable can be stronger because the large-scale circulation variable is the basis of regional forecasting and needs to quickly converge to the background field to avoid the forecast field deviating from the large-scale situation. The relaxation intensity of the thermal or water vapor variable can be weaker because small- and medium-scale features such as temperature and humidity are more obvious (such as local heating and orographic precipitation), and excessive relaxation will suppress these refined features. Based on this, this embodiment of the invention can further improve forecast accuracy. For example, the relaxation timescale approximation intensity for upper-level wind fields can be 1 / 3600 s, and for temperature it can be 1 / 21600 s, and so on. Based on this, the relaxation timescale approximation intensity of any target meteorological variable directly controls the intensity of the model state's adjustment to the background field state. The greater the relaxation timescale approximation intensity, the stronger the relaxation forcing term, and the faster and more forcefully the model state is pulled towards the background field value. Conversely, the smaller the relaxation timescale approximation intensity, the slower and gentler the adjustment process. Therefore, its value needs to strike a balance between effectively correcting model errors and maintaining the model's own physical dynamics. Optionally, the relaxation timescale approximation intensity of any target meteorological variable can be less than or equal to 1 / model integration time step (where the model integration time step can be in seconds).

[0075] Optionally, the electronic device can also determine whether any target meteorological variable is a relaxed forcing variable. If any target meteorological variable is a relaxed forcing variable, it triggers the execution of the variable background field of any target meteorological variable at the current time step determined from the fused meteorological forecast data. If any target meteorological variable is not a relaxed forcing variable, it performs meteorological forecasting based on the variable forecast field of any target meteorological variable at the current time step, and obtains the variable forecast field of any target meteorological variable at the next time step to be traversed. At this time, the variable forecast value of any target meteorological variable at the next time step to be traversed and at any target grid point can be determined based on the summation result between at least one physical forcing term in the WRF mode and the variable forecast value of any target meteorological variable at the current time step and at any target grid point. For example, the variable forecast value of any target meteorological variable at the next model integration time step and at any target grid point can be: the summation result between at least one physical forcing term in the WRF mode and the variable forecast value of any target meteorological variable at the current time step and at any target grid point, and so on. Optionally, the electronic device can determine at least one relaxed forcing variable, and then determine whether any target meteorological variable is a relaxed forcing variable based on the at least one relaxed forcing variable. If any target meteorological variable is a meteorological variable among the at least one relaxed forcing variable, then the target meteorological variable can be determined to be a relaxed forcing variable; if any target meteorological variable is not a meteorological variable among the at least one relaxed forcing variable, then the target meteorological variable can be determined not to be a relaxed forcing variable. Alternatively, the electronic device can also determine the variable indicator of any target meteorological variable. When the variable indicator is a relaxed variable indicator, the target meteorological variable can be determined to be a relaxed forcing variable; when the variable indicator is a non-relaxed indicator or is empty, the target meteorological variable can be determined not to be a relaxed forcing variable, and so on. The specific method for determining whether any target meteorological variable is a relaxed forcing variable is not limited in the embodiments of the present invention. Optionally, at least one relaxed forcing variable, a relaxed variable indicator, and a non-relaxed indicator can all be set according to experience or actual needs, and the embodiments of the present invention do not limit this. For example, at least one relaxed forcing variable may include at least one model meteorological variable, that is, one model meteorological variable is one relaxed forcing variable. In this case, the forecast field can tend to the background field of the large meteorological model, and so on.

[0076] For example, for any time step (such as any time step to be traversed or any model integration time step) and any target meteorological variable, if the target meteorological variable is a relaxed forcing variable, and any time step is an initial forecast time step or a target forecast start time, then the variable background field of any target meteorological variable at any time step can be determined from the fused meteorological forecast data. That is, the variable background field of any target meteorological variable at any time step may not be empty. Alternatively, if any target meteorological variable is a relaxed forcing variable, then the variable background field of any target meteorological variable at any time step may not be empty. In this case, regardless of where any time step is located, the variable background field of any target meteorological variable can be determined. The background field of a meteorological variable at any time step can be determined by interpolation when the time step is neither an initial forecast time step nor the target start time. Alternatively, the background field of any target meteorological variable at the corresponding initial forecast time step can be determined when the time step is an initial forecast time step, thus achieving the determination of the background field of any target meteorological variable at any time step. Or, if any target meteorological variable is not a relaxed forcing variable, or if any time step is neither an initial forecast time step nor the target start time, the background field of any target meteorological variable at any time step can be empty, etc. The embodiments of the present invention do not limit this. Correspondingly, when the background field of any target meteorological variable is empty at any time step, a meteorological forecast can be made based on the forecast field of any target meteorological variable at any time step to obtain the forecast field of any target meteorological variable at the next time step; when the background field of any target meteorological variable is not empty at any time step, a meteorological forecast can be made based on the forecast field and background field of any target meteorological variable at any time step to obtain the forecast field of any target meteorological variable at the next time step, and so on.

[0077] Based on this, electronic devices can use nudging (an empirical data assimilation technique in WRF models) to continuously approximate the background field by introducing relaxation forcing terms into the forecast equations. In other words, by using nudging, weather forecasts can be made based on the forecast field and background field of any target meteorological variable at the current time step, and the forecast field of any target meteorological variable at the next time step to be traversed can be obtained. Compared with direct background field interpolation assimilation, this technique can better consider topographic and convection effects, making the assimilation results more meteorologically consistent, thereby strengthening the constraint effect of the background field of the large meteorological model on regional weather forecasts.

[0078] Optionally, in other embodiments, the electronic device can also enable nudging technology through the &fdda module (the core configuration block for Four-Dimensional Data Assimilation in WRF, mainly used to introduce large-scale reanalysis or forecast data during model integration, constraining simulation results to be closer to the real atmospheric state, etc.) in the namelist.input (parameter configuration file of the WRF model) of the target WRF model. For example, by customizing the &fdda parameter, each relaxed forcing variable can be assimilated throughout the entire boundary layer height, achieving efficient injection of the background field of the large meteorological model in the forecasting process, etc. It should be noted that the specific settings of the parameters in the embodiments of the present invention are not limited. For example, one or more relaxed forcing variables can be set, or each target meteorological variable can be a relaxed forcing variable, or the relaxation timescale approximation intensity of any target meteorological variable can be set, etc.

[0079] S306 After traversing all time steps except the last one, the target weather forecast field for each target forecast time step in the target area is obtained, so as to realize the calling of the target WRF mode and the forecast of the target weather forecast data in the target area based on the fused weather forecast data.

[0080] To further illustrate the beneficial effects of the atmospheric forecasting method based on a large meteorological model proposed in this invention, a three-layer nested simulation is used for experimental demonstration. The first layer represents my country and surrounding areas (resolution 27 km × 27 km), the second layer includes most of my country (resolution 9 km × 9 km), and the third layer represents the Hunan, Jiangxi, and Hubei regions of my country (resolution 3 km × 3 km). The simulation period for this invention is from December 29, 2024 to January 10, 2025. For example, the Pangu large meteorological model is used as an example for illustration. Figure 4 The figure shows the variation characteristics of three meteorological elements over time (hours) at a station in Changsha during the test period; where the unit of 2m temperature can be ℃ (degrees Celsius), the unit of 2m humidity (i.e., 2m relative humidity) can be %, and the unit of 10m wind speed can be m / s. Figure 4The vertical axis can represent temperature, relative humidity, and wind speed, respectively. Based on this, embodiments of the present invention can compare the 2m temperature forecasts and observed temperatures from two sets of experiments (i.e., the atmospheric forecasting method based on a large meteorological model proposed in the embodiments of the present invention, also known as the large meteorological model-driven forecasting method) and the GFS-driven model forecasting method (i.e., the method of weather forecasting by driving the WRF model with GFS forecast data, also known as traditional numerical forecasting or traditional numerical forecasting methods, etc.). The results show that the forecast driven by the large meteorological model is slightly better, capturing the overall trend more accurately, especially during nighttime low temperatures and rapid cooling processes, where it more closely approximates observations. After introducing the large meteorological model-driven optimization technology (i.e., forecasting using the present invention's forecasting method), the forecasting effect of 2m relative humidity is significantly improved. In particular, the large meteorological model-driven method can forecast high humidity weather conditions and more accurately depicts the diurnal variation trend of increasing humidity at night and decreasing it during the day. In the 10m wind speed forecast, the overall error of the large meteorological model-driven forecast is smaller, significantly improving the problem of overestimating wind speeds under weak wind conditions, and enhancing the reliability of pollution diffusion capabilities under stable weather conditions. Figure 4 The horizontal axis represents the date.

[0081] For example, such as Figure 5 As shown, the spatial distribution of the average forecast values ​​of surface meteorological elements over the test period, combined with the average values ​​of station observations (i.e., the average values ​​from December 29, 2024 to January 10, 2025), reveals that the forecasting effect driven by the meteorological large model (i.e., the forecasting effect of the atmospheric forecasting method based on the meteorological large model proposed in this embodiment) shows a comprehensive advantage in the Hunan-Jiangxi-Hubei region during the test period. The spatial distribution patterns of 2m temperature in the two sets of experimental forecasts are not significantly different; for 2m relative humidity forecasts, the forecasting system of the GFS-driven model is relatively low, while the spatial distribution of the forecast driven by the meteorological large model is more reasonable, and the location and intensity of high humidity are more consistent with observations, correcting the problem of significantly low humidity in the traditional driving method; furthermore, the meteorological large model drives the simulation of 10m wind speed better than the traditional meteorological field driving method, mainly in effectively improving the problem of overestimating wind speed forecasts, and the range of the stable atmospheric zone shrinking synchronously with observations, providing a more realistic atmospheric diffusion capacity background for subsequent air quality forecasts. Figure 5 The colored bars in the image can represent temperature, relative humidity, and wind speed, respectively.

[0082] For example, such as Figure 6As shown, taking Changsha City as an example, the comparison of the diurnal variation characteristics of AQI (Air Quality Index) observed and predicted by various methods demonstrates that the meteorological large-scale model-driven method (i.e., the forecasting method of this invention shown in the figure) is significantly better than traditional numerical forecasting (such as the GFS-driven model forecasting method) for pollution periods (i.e., when the AQI is high, such as AQI greater than 100). This advantage is particularly evident on polluted days. It is evident that the meteorological large-scale model-driven method in the Hunan-Jiangxi-Hubei region not only maintains the stability of traditional meteorological background field-driven methods in temperature field forecasting, but also achieves "dual correction" on the two key variables of relative humidity and wind speed, providing high-fidelity meteorological field conditions for subsequent accurate air quality forecasts, ultimately effectively improving the forecast accuracy of regional pollution processes. Figure 6 The horizontal axis represents the date.

[0083] Based on this, the present invention combines a large meteorological model with a WRF model to fully leverage the advantages of both, thereby improving the accuracy and practicality of weather forecasts. Specifically, the large meteorological model can efficiently uncover the underlying patterns in observational data, assisting the WRF model in better capturing localized and extreme weather events, thus enhancing forecast precision. Furthermore, the powerful data processing capabilities of the large meteorological model can optimize the data input to the WRF model, effectively improving the accuracy of initial boundary conditions and enhancing forecast accuracy. Moreover, the present invention aims to provide a high-quality meteorological background field (i.e., a fused background field) for the WRF model using the large meteorological model. By adding a forcing term to the WRF forecast equation and using grid-to-grid relaxation forcing terms to gradually approximate the fused meteorological forecast data (i.e., approximate the forecast results of the large meteorological model) within the forecast period, and further strengthening this forcing, the advantages of the large meteorological model are fully utilized, effectively improving the accuracy and practicality of weather forecasts and providing high-fidelity meteorological field conditions for accurate forecasting of regional pollution processes.

[0084] In summary, the atmospheric forecasting method based on a large meteorological model proposed in this invention can also be applied to regional pollution process forecasting (i.e., air quality forecasting). This leads to the development of a regional pollution process forecasting method coupled with a large meteorological model. This method deeply integrates the capabilities of the large meteorological model, effectively improving the efficiency and accuracy of regional meteorological and pollution forecasts. In other words, this invention can improve the accuracy of the global meteorological background field by relying on a large meteorological model, and can use updated large meteorological model input driving data (such as IFS forecast data) that meets operational forecasting requirements, enabling "daily rolling" meteorological and pollution forecasts and significantly improving forecast effectiveness and operational availability.

[0085] This invention embodiment can, after obtaining initial meteorological forecast data for a target forecast time range within an initial region, determine the input data for a large meteorological model based on the initial meteorological forecast data; and then call the target large meteorological model, determining the forecast data for the large meteorological model based on the input data. Then, the target WRF preprocessing system can be called to perform data fusion based on the initial meteorological forecast data and the forecast data from the large meteorological model, obtaining fused meteorological forecast data for the target forecast time range within the target region. The fused meteorological forecast data includes the initial boundary conditions for the target forecast time range within the target region. The target forecast time range may include multiple time steps to be traversed, including the target start time and at least one target forecast time step. The fused meteorological forecast data supports the determination of the fusion background field for each of the multiple time steps to be traversed within the target region. Based on this, the target WRF model can be called to traverse each of the multiple time steps to be traversed except the last one, and the currently traversed time step is taken as the current time step. Furthermore, for any one of multiple target meteorological variables, the background field of that target meteorological variable at the current time step can be determined from the fused meteorological forecast data. Based on the forecast field and background field of that target meteorological variable at the current time step, a meteorological forecast is performed, yielding the forecast field for the next time step to be traversed. This allows for the forecasting of the target meteorological forecast field for the next time step in the target region. The target meteorological forecast field for a time step in the target region includes the forecast fields for each of the multiple target meteorological variables at their respective time steps. After traversing all but the last of the multiple time steps to be traversed, the target meteorological forecast field for each target forecast time step in the target region is obtained. This enables the invocation of the target WRF model, based on the fused meteorological forecast data, to forecast the target meteorological forecast data for the target forecast time range within the target region. As can be seen, the embodiments of the present invention can strengthen the constraint effect of the background field of the meteorological big model (i.e. the background field predicted by the meteorological big model in the fused background field) on regional forecasts by relaxing the grid points of the meteorological forecast, effectively transfer the capabilities of the meteorological big model to the regional meteorological model, improve the regional meteorological forecast effect, especially solve the problems of low relative humidity forecast and high wind speed forecast, and thus significantly improve the accuracy of regional pollution process forecast.

[0086] Based on the description of the relevant embodiments of the atmospheric forecasting method based on the meteorological large model described above, this invention also proposes an atmospheric forecasting device based on the meteorological large model. This atmospheric forecasting device can be a computer program (including program code) running on an electronic device; such as... Figure 7As shown, the atmospheric forecasting device based on a large meteorological model may include an acquisition unit 701 and a processing unit 702. This atmospheric forecasting device based on a large meteorological model can perform... Figure 1 or Figure 3 The atmospheric forecasting method based on a large meteorological model shown, i.e., the atmospheric forecasting device based on a large meteorological model, can operate the above-mentioned units: Acquisition unit 701 is used to acquire initial meteorological forecast data within the initial region for the target forecast time range, wherein the initial region includes the target region; The processing unit 702 is further configured to determine the input data of the meteorological big model based on the initial meteorological forecast data; and to call the target meteorological big model to determine the forecast data of the meteorological big model based on the input data of the meteorological big model. The processing unit 702 is further configured to call the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data for the target forecast time range in the target area. The fused weather forecast data includes the initial boundary conditions for the target forecast time range in the target area. The processing unit 702 is further configured to invoke the target WRF model and, based on the fused meteorological forecast data, forecast the target meteorological forecast data for the target forecast time range within the target area. The target meteorological forecast data includes the target meteorological forecast fields for each target forecast time step within the target forecast time range within the target area.

[0087] In one implementation, when the processing unit 702 determines the input data for the large meteorological model based on the initial weather forecast data, it may specifically be used for: Determine at least one model meteorological variable and model grid partitioning indication information corresponding to the target meteorological large model; Based on the at least one model meteorological variable, the input data of the model to be processed is determined from the initial meteorological forecast data, wherein the input data of the model to be processed includes the variable data of each model meteorological variable in the at least one model meteorological variable; According to the model grid division instruction information, the input data of the model to be processed is transformed to obtain the input data of the meteorological large model, so as to determine the input data of the meteorological large model.

[0088] In another embodiment, the initial weather forecast data includes first initial weather forecast data and second initial weather forecast data. When the processing unit 702 determines the input data of the model to be processed from the initial weather forecast data according to the at least one model weather variable, it can specifically be used for: Based on the at least one model meteorological variable, determine the input data of the model to be processed from the first initial meteorological forecast data; When processing unit 702 calls the target WRF preprocessing system to perform data fusion based on the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data with the target forecast time range within the target area, it can be specifically used for: The target WRF preprocessing system is invoked to fuse the second initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data within the target area for the target forecast time range.

[0089] In another embodiment, the target WRF preprocessing system includes a format decoding module and a meteorological grid fusion module; when the processing unit 702 calls the target WRF preprocessing system to perform data fusion based on the initial meteorological forecast data and the meteorological large model forecast data to obtain the fused meteorological forecast data with the target forecast time range under the target area, it can be specifically used for: The format decoding module is invoked to parse the initial weather forecast data and the weather big model forecast data respectively, to obtain the initial weather forecast parsing data and the weather big model forecast parsing data. The meteorological grid fusion module is invoked to perform data fusion based on the initial meteorological forecast analysis data and the meteorological big model forecast analysis data to obtain fused meteorological forecast data for the target forecast time range in the target area. The fusion priority of the meteorological big model forecast analysis data is higher than that of the initial meteorological forecast analysis data.

[0090] In another embodiment, the processing unit 702 may also be used for: When a model meteorological variable lookup table creation instruction for the target meteorological large model is detected, the model meteorological variable lookup table is created according to the model meteorological variable lookup table creation instruction; wherein, a variable lookup table is used to map the variable code of a background field to the WRF standard variable identifier; When processing unit 702 calls the format decoding module to parse the initial weather forecast data and the large-scale meteorological model forecast data respectively, and obtains the initial weather forecast parsing data and the large-scale meteorological model forecast parsing data, it can be specifically used for: The variable lookup table corresponding to the initial weather forecast data and the model weather variable lookup table are determined; and the format decoding module is called to parse the initial weather forecast data and the large-scale meteorological model forecast data according to the variable lookup table corresponding to the initial weather forecast data and the model weather variable lookup table, respectively, to obtain the initial weather forecast parsing data and the large-scale meteorological model forecast parsing data.

[0091] In another embodiment, the target forecast time range includes multiple time steps to be traversed, the multiple time steps to be traversed include the target start time and at least one target forecast time step; the fused meteorological forecast data supports the determination of the fused background field of each of the multiple time steps to be traversed in the target area, and a fused background field includes the variable background field of each of the multiple target meteorological variables; when the processing unit 702 calls the target WRF mode and forecasts the target meteorological forecast data of the target forecast time range in the target area based on the fused meteorological forecast data, it can be specifically used for: Invoke the target WRF pattern, traverse each of the multiple time steps to be traversed except the last one, and take the currently traversed time step as the current time step; For any one of the multiple target meteorological variables, the background field of the variable for the target meteorological variable at the current time step is determined from the fused meteorological forecast data. Based on the forecast field and background field of the variable for the target meteorological variable at the current time step, a meteorological forecast is performed to obtain the forecast field of the variable for the next time step to be traversed at the current time step. This enables the forecasting of the target meteorological forecast field at the next time step to be traversed at the current time step in the target area. The target meteorological forecast field at a time step in the target area includes the forecast fields of each of the multiple target meteorological variables at the corresponding time step. After traversing all time steps except the last one, the target weather forecast field for each target forecast time step in the target area is obtained, so as to realize the invocation of the target WRF mode and, based on the fused weather forecast data, forecast the target weather forecast data for the target forecast time range in the target area.

[0092] In another embodiment, the processing unit 702 may also be used for: Determine whether any of the target meteorological variables is a relaxed forcing variable; If any of the target meteorological variables is the relaxed forcing variable, then the process of determining the variable background field of any target meteorological variable from the fused meteorological forecast data at the current time step is triggered. If any of the target meteorological variables is not the relaxed forcing variable, then a meteorological forecast is made based on the variable forecast field of any of the target meteorological variables at the current time step, and the variable forecast field of any of the target meteorological variables at the next time step to be traversed at the current time step is obtained.

[0093] According to one embodiment of the present invention, Figure 7 Each unit in the atmospheric forecasting device based on a large meteorological model shown can be individually or entirely merged into one or more other units, or one or more of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effect of the embodiments of the present invention. The above units are based on logical function division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of the present invention, any atmospheric forecasting device based on a large meteorological model may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0094] According to another embodiment of the present invention, it is possible to perform operations such as those described above by running on a general-purpose electronic device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). Figure 1 or Figure 3 The computer program (including program code) involved in each step of the corresponding method shown, to construct such... Figure 7 The invention describes an atmospheric forecasting device based on a large meteorological model, and a method for implementing such an embodiment. The computer program can be stored on, for example, a computer storage medium, loaded onto the aforementioned electronic device via the computer storage medium, and run therein.

[0095] Based on the description of the method and apparatus embodiments above, an exemplary embodiment of the present invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, which, when executed by the at least one processor, causes the electronic device to perform the method according to an embodiment of the present invention.

[0096] An exemplary embodiment of the present invention also provides a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to an embodiment of the present invention.

[0097] An exemplary embodiment of the present invention also provides a computer program product, including a computer program, wherein, when executed by a computer's processor, the computer program is used to cause the computer to perform a method according to an embodiment of the present invention.

[0098] refer to Figure 8 The present invention will now be described in the form of a structural block diagram of an electronic device 800 that can serve as a server or client of the present invention, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0099] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0100] Multiple components in electronic device 800 are connected to I / O interface 805, including: input unit 806, output unit 807, storage unit 808, and communication unit 809. Input unit 806 can be any type of device capable of inputting information to electronic device 800. Input unit 806 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 807 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 808 may include, but is not limited to, disks and optical discs. Communication unit 809 allows electronic device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0101] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above. For example, in some embodiments, the atmospheric forecasting method based on a large meteorological model can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via ROM 802 and / or communication unit 809. In some embodiments, the computing unit 801 can be configured to perform the atmospheric forecasting method based on a large meteorological model by any other suitable means (e.g., by means of firmware).

[0102] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0104] As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0105] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0106] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0107] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0108] Furthermore, it should be understood that the above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. An atmospheric forecasting method based on a large meteorological model, characterized in that, include: Acquire initial meteorological forecast data within the target forecast time range of the initial region, which includes the target region; Based on the initial weather forecast data, determine the input data for the large-scale meteorological model; It also invokes the target meteorological big data model, and determines the meteorological big data model forecast data based on the input data of the meteorological big data model; The target WRF preprocessing system is invoked. This system includes a format decoding module and a meteorological grid fusion module. Data fusion is performed based on initial weather forecast data and large-scale meteorological model forecast data to obtain fused weather forecast data for the target forecast time range within the target area. This includes: invoking the format decoding module to parse the initial weather forecast data and the large-scale meteorological model forecast data respectively, obtaining initial weather forecast parsing data and large-scale meteorological model forecast parsing data; and invoking the meteorological grid fusion module to fuse the initial weather forecast parsing data and the large-scale meteorological model forecast parsing data to obtain fused weather forecast data for the target forecast time range within the target area. The fusion priority of the large-scale meteorological model forecast parsing data is higher than that of the initial weather forecast parsing data. The meteorological forecast data is obtained by fusing data through the format decoding module and the meteorological grid fusion module according to the fusion priority, so that the meteorological large model forecast analysis data is preferentially fused into the fused meteorological forecast data. The fused meteorological forecast data includes the initial boundary conditions of the target forecast time range under the target region; the fused meteorological forecast data also includes the fused background field of each initial forecast time step in the target forecast time range under the target region; the target forecast time range includes multiple time steps to be traversed, which include the target start time and at least one target forecast time step; the fused meteorological forecast data supports the determination of the fused background field of each of the multiple time steps to be traversed under the target region, and a fused background field includes the variable background field of each of the multiple target meteorological variables; The process involves invoking the target WRF pattern and, based on fused meteorological forecast data, forecasting target meteorological forecast data within the target area over a specified time range. This includes: invoking the target WRF pattern, iterating through multiple time steps (excluding the last one) and using the currently iterated time step as the current time step; for any target meteorological variable among multiple target meteorological variables, determining the variable background field of that variable at the current time step from the fused meteorological forecast data, and performing a meteorological forecast based on the variable forecast field and variable background field of that variable at the current time step to obtain the forecast data for that target meteorological variable at the current time step. A variable forecast field for a time step to be traversed is used to forecast the target meteorological forecast field for the next time step to be traversed in the target area. The target meteorological forecast field for a time step in the target area includes the variable forecast fields of each of the multiple target meteorological variables at the corresponding time step, thereby achieving the relaxation approximation assimilation of the meteorological forecast grid. After traversing all the time steps except the last one, the target meteorological forecast fields for each target forecast time step in the target area are obtained. The target meteorological forecast data includes the target meteorological forecast fields for each target forecast time step in the target area within the target forecast time range.

2. The method according to claim 1, characterized in that, The process of determining the input data for the large-scale meteorological model based on initial weather forecast data includes: Determine at least one model meteorological variable and model grid partitioning indication information corresponding to the target meteorological large model; Based on the at least one model meteorological variable, the input data of the model to be processed is determined from the initial meteorological forecast data, wherein the input data of the model to be processed includes the variable data of each model meteorological variable in the at least one model meteorological variable; According to the model grid division instruction information, the input data of the model to be processed is transformed to obtain the input data of the meteorological large model, so as to determine the input data of the meteorological large model.

3. The method according to claim 2, characterized in that, The initial weather forecast data includes first initial weather forecast data and second initial weather forecast data. The step of determining the input data for the model to be processed from the initial weather forecast data according to the at least one model weather variable includes: Based on the at least one model meteorological variable, determine the input data of the model to be processed from the first initial meteorological forecast data; The process involves calling the target WRF preprocessing system to fuse the initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data for the target forecast time range within the target area, including: The target WRF preprocessing system is invoked to fuse the second initial weather forecast data and the weather big model forecast data to obtain fused weather forecast data within the target area for the target forecast time range.

4. The method according to claim 1, characterized in that, The method further includes: When a model meteorological variable lookup table creation instruction for the target meteorological large model is detected, the model meteorological variable lookup table is created according to the model meteorological variable lookup table creation instruction; wherein, a variable lookup table is used to map the variable code of a background field to the WRF standard variable identifier; The method of calling the format decoding module parses the initial weather forecast data and the large-scale meteorological model forecast data respectively, to obtain the initial weather forecast parsing data and the large-scale meteorological model forecast parsing data, including: The variable lookup table corresponding to the initial weather forecast data and the model weather variable lookup table are determined; and the format decoding module is called to parse the initial weather forecast data and the large-scale meteorological model forecast data according to the variable lookup table corresponding to the initial weather forecast data and the model weather variable lookup table, respectively, to obtain the initial weather forecast parsing data and the large-scale meteorological model forecast parsing data.

5. The method according to any one of claims 1-4, characterized in that, The method further includes: Determine whether any of the target meteorological variables is a relaxed forcing variable; If any of the target meteorological variables is the relaxed forcing variable, then the process of determining the variable background field of any target meteorological variable at the current time step from the fused meteorological forecast data is triggered. If any of the target meteorological variables is not the relaxed forcing variable, then a meteorological forecast is made based on the variable forecast field of any of the target meteorological variables at the current time step, and the variable forecast field of any of the target meteorological variables at the next time step to be traversed at the current time step is obtained.

6. An atmospheric forecasting device based on a large meteorological model, characterized in that, The device includes: The acquisition unit is used to acquire initial meteorological forecast data within the target forecast time range in the initial area, where the initial area includes the target area. The processing unit is used to determine the input data of the meteorological big model based on the initial meteorological forecast data; and to call the target meteorological big model and determine the forecast data of the meteorological big model based on the input data of the meteorological big model. The processing unit is also used to invoke the target WRF preprocessing system, which includes a format decoding module and a meteorological grid fusion module. Based on the initial weather forecast data and the large meteorological model forecast data, it performs data fusion to obtain fused weather forecast data within the target forecast time range for the target area. This includes: invoking the format decoding module to parse the initial weather forecast data and the large meteorological model forecast data respectively, obtaining initial weather forecast parsing data and large meteorological model forecast parsing data; and invoking the meteorological grid fusion module to perform data fusion based on the initial weather forecast parsing data and the large meteorological model forecast parsing data, obtaining fused weather forecast data within the target forecast time range for the target area. The fusion priority of the large meteorological model forecast parsing data is higher than that of the initial weather forecast parsing data. The fused meteorological forecast data is obtained by fusing data according to fusion priority through the format decoding module and the meteorological grid fusion module, so that the meteorological large model forecast analysis data is preferentially fused into the fused meteorological forecast data. The fused meteorological forecast data includes the initial boundary conditions of the target forecast time range under the target area; the fused meteorological forecast data also includes the fused background field of each initial forecast time step in the target forecast time range under the target area; the target forecast time range includes multiple time steps to be traversed, which include the target start time and at least one target forecast time step; the fused meteorological forecast data supports the determination of the fused background field of each time step in the target area under the multiple time steps to be traversed, and a fused background field includes the variable background field of each target meteorological variable among multiple target meteorological variables; The processing unit is also used to invoke the target WRF pattern and, based on the fused meteorological forecast data, forecast target meteorological forecast data within the target area for the target forecast time range. This includes: invoking the target WRF pattern, traversing each of the multiple time steps to be traversed except the last one, and taking the currently traversed time step as the current time step; for any target meteorological variable among the multiple target meteorological variables, determining the variable background field of any target meteorological variable at the current time step from the fused meteorological forecast data, and performing a meteorological forecast based on the variable forecast field and variable background field of any target meteorological variable at the current time step, to obtain the target meteorological variable at the current time step. The variable forecast field for the next time step to be traversed is obtained to forecast the target meteorological forecast field in the target area for the next time step to be traversed in the current time step. The target meteorological forecast field in the target area for a time step includes the variable forecast fields of each target meteorological variable in the corresponding time step, thereby realizing the relaxation approximation assimilation of meteorological forecast grid points. After traversing all the time steps except the last one to be traversed, the target meteorological forecast field in the target area for each target forecast time step is obtained. The target meteorological forecast data includes the target meteorological forecast field in the target area for each target forecast time step in the target forecast time range.

7. An electronic device, characterized in that, include: processor; as well as Stored program memory, The program includes instructions that, when executed by the processor, cause the processor to perform the method according to any one of claims 1-5.

8. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-5.

Images