Regional refined sea wave coupling numerical forecasting method based on non-static mode

Abstract

The invention discloses a regional refined sea wave coupling numerical forecasting method based on a non-static mode, and aims to improve forecasting accuracy. The method comprises the following steps: firstly, constructing a regional refined sea wave coupling numerical forecasting system consisting of a data collection subsystem, a preprocessing subsystem, a data assimilation subsystem, a coupling mode calculation subsystem and a post-processing subsystem; the data collection subsystem obtains related data of atmosphere, ocean and sea waves; the preprocessing subsystem generates atmosphere, ocean and sea wave mode grid files and interpolates atmosphere, ocean and sea wave forecast fields to obtain preprocessed data, namely an atmosphere initial field, an atmosphere boundary field, an atmosphere forced field, an ocean initial field, an ocean boundary field and a sea wave boundary field. The data assimilation subsystem assimilates the preprocessed data to obtain assimilated data; and the coupling mode calculation subsystem performs integration on assimilated data to obtain forecast fields of atmosphere, ocean and sea waves. And the post-processing subsystem generates a forecast product according to each forecast field. The method is small in forecasting error and high in forecasting accuracy.

Description

Technical Field

[0001] The present invention relates to the field of meteorological and ocean numerical forecasting, and particularly relates to a regional refined air-sea-wave coupled numerical forecasting method. Background Art

[0002] The interaction between the atmosphere and the ocean plays a leading role in the generation, extinction and evolution process of the global weather and climate system. In reality, the interaction between the atmosphere and the ocean is realized through various dynamic and thermodynamic processes at the air-sea interface. However, at present, the forecasting still relies on independent atmospheric models or ocean circulation models. Therefore, there are problems of mismatch and imbalance in the fluxes between the air-sea interfaces. In addition, waves affect the heat and momentum transfer between the ocean and the atmosphere through different wave states, which is also a key influencing factor that cannot be ignored. Therefore, it is necessary to couple the atmospheric, ocean circulation and wave models together to form an organic system in which the atmospheric, ocean circulation and wave models can instantaneously exchange information with each other, so as to synchronously, objectively and quantitatively produce an integrated forecast of meteorological and hydrological elements.

[0003] Considering that there are some air-sea interaction phenomena occurring at limited spatial scales (such as ocean basin scales) and short time scales (generally less than 1 week) between the atmosphere and the ocean, such as tropical cyclones and mid-latitude ocean explosive cyclones, strong offshore winds and low-level jets, coastal storm surges and sea fogs, ocean frontogenesis and upper ocean variability, etc., whose dynamic and thermodynamic mechanisms are mainly controlled by the meso-scale air-sea interaction process. In addition, the coastal marine environment in China is complex, with rich multi-scale dynamic processes. The forecasting accuracy of elements such as underwater sound, sea current, temperature, salinity, and density needs to be improved, and there are still great difficulties in forecasting ocean phenomena such as thermoclines, internal waves, and meso-scale eddies. Therefore, the refined air-sea-wave coupling technology is an important means to improve the level of ocean environmental forecasting. Developing a regional refined air-sea-wave coupled numerical forecasting method can strengthen the understanding of multi-scale, especially meso-scale ocean dynamic processes, quickly and accurately predict the changes of ocean dynamic processes around islands, and improve the early warning and detection ability of high-impact weather on the sea surface, meeting the urgent needs of ocean hydrological forecasting and guarantee in the new era and new stage.

[0004] Existing regional numerical meteorological and hydrological forecasting methods mostly adopt a single forecasting mode. For example, a regional atmospheric model (such as the Weather Research and Forecasting Model, WRF) is used alone. The global atmospheric forecast field is used as the boundary field of the regional atmospheric model, and the initial field after assimilating atmospheric observation data is used as the initial condition of the regional atmospheric model. The regional atmospheric model is driven to run, and the model calculation results are output and post-processed to generate regional atmospheric forecast products. Similarly, the regional hydrostatic ocean model also uses the forecast fields of the global ocean and atmosphere to produce the initial field and boundary field of the regional ocean model, and drives the regional ocean model to perform integral calculations. Currently, there are two types of regional ocean models: hydrostatic models and non-hydrostatic models. Since the hydrostatic ocean model has a lower computational complexity and is more suitable for simulating ocean states at low resolutions, the current ocean operational forecasting mainly adopts the regional hydrostatic ocean model (such as the Princeton Ocean Model, POM). However, with the continuous improvement of the computing power of high-performance computers and the need for forecasting fine mesoscale processes (internal waves, mesoscale eddies, etc.) in the ocean, the hydrostatic ocean model is no longer suitable for refined ocean forecasting. Therefore, the main problems existing in the current single forecasting mode are as follows: 1. The ocean and atmospheric component models operate independently. The variables at the air-sea interface are either initial fixed values, or statistical parameterized estimated values, or offline updated using low-frequency (3 to 6 hours or even longer) file data. This will cause the imbalance problem of variables at the air-sea interface, resulting in large forecasting errors. 2. Most of the ocean models in the existing research-type coupled models are hydrostatic models with a resolution of 1 / 12° (about 9 km), and it is difficult to accurately reflect complex ocean processes such as internal waves and internal tides in the ocean under extremely high resolution requirements (such as 1 km and below), and the model has inaccurate simulations of mesoscale processes. 3. The existing research-type coupled models often adopt a global smoothing method to process the terrain, artificially reducing the fineness of the ocean terrain and not being refined enough in the treatment of rivers and waterways, resulting in inaccurate forecasting of physical processes in nearshore shallow waters. 4. The existing research-type coupled models do not consider assimilating multi-source observation data, especially lacking the coupled assimilation technology of the ocean and atmosphere. Therefore, the initial conditions are inaccurate, and it is only suitable for scientific research on certain cases and not applicable to quasi-real-time operational coupled forecasting methods. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a regional refined air-sea-wave coupled numerical prediction method based on a non-hydrostatic atmospheric model, a non-hydrostatic ocean model, and a wave spectrum model of the full physical process, aiming at the problems of large prediction errors in the existing regional numerical meteorological and hydrological prediction methods, inaccurate and imprecise prediction of physical processes in nearshore shallow waters, and inapplicability to quasi-real-time operational coupled prediction. The method can reduce prediction errors, especially improve the accuracy of predicting physical processes in nearshore shallow waters, and comprehensively consider the coupled assimilation technology of multi-source observation data to improve the accuracy of initial conditions, and support high-performance parallel computing platforms for quasi-real-time operational coupled prediction.

[0006] The technical solution of the present invention is as follows:

[0007] In the first step, a regional refined air-sea-wave coupled numerical prediction system is constructed. The regional refined air-sea-wave coupled numerical prediction system consists of a data collection subsystem, a preprocessing subsystem, a data assimilation subsystem, a coupled model calculation subsystem, and a post-processing subsystem.

[0008] The data collection subsystem is connected to a dedicated line database (usually installed on the servers of a high-performance computing center), a preprocessing subsystem, and a data assimilation subsystem, and reads configuration files, global atmospheric forecast field data (GRIB format) for the current day's forecast, global ocean wave forecast field data, global ocean forecast field data, high-precision terrain data, tidal data, real-time atmospheric observation data, and real-time ocean observation data from the dedicated line database. For the global atmospheric forecast field data, it is clipped according to the forecast area's longitude and latitude (read from the configuration file), and the three-dimensional air temperature, geopotential height, relative humidity, and wind vector field variable files are merged into a global atmospheric forecast field variable file. Then, according to the date and data source, a regional atmospheric forecast field is generated and saved in the standard GRIB format; for the global ocean forecast field data, it is clipped according to the forecast area range, and the sea surface temperature, salinity, sea surface height, and flow field variable files are merged into a global ocean forecast field variable file, and a regional ocean forecast field is obtained and saved in the standard NetCDF format; for the global ocean wave forecast field data, it is clipped and archived according to the forecast area range, and the significant wave height, wave period, wavelength, and wave direction variables are merged to obtain a regional ocean wave forecast field and saved in the standard NetCDF format. The real-time atmospheric observation data is classified and processed. The real-time atmospheric observation data is renamed as an atmospheric observation file according to the data source and archived in ASCII or PREPBUFR format, and the observation error file in the atmospheric observation file is archived in text document format; the real-time ocean observation data is classified and processed. The real-time ocean observation data is renamed as an ocean observation file according to the data source or type and archived in NetCDF format, and the climatological statistical results of the ocean model in the ocean observation file are archived in chronological order. The generated regional atmospheric forecast field, regional ocean wave forecast field, regional ocean forecast field, as well as the tidal data and high-precision terrain data obtained from the dedicated line database are sent to the preprocessing subsystem, and the atmospheric observation file and ocean observation file are sent to the data assimilation subsystem. The content of the configuration file includes: working directory, forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, number of parallel nodes, map projection method, forecast time (i.e., the start date t s and the end date t e ) of the integration, grid resolution and number of grid points of the atmospheric model module, grid resolution and number of grid points of the ocean model module, grid resolution and number of grid points of the ocean wave model module, integration time step (Ta) of the atmospheric model module, integration time step (To) of the ocean model module, integration time step (Tw) of the ocean wave model module, atmospheric coupling exchange time step (Tca), ocean coupling exchange time step (Tco), ocean wave coupling exchange time step (Tcw), output path and output frequency, assimilation area, assimilation time window, and variable names for assimilation. The configuration file is sent to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem.

[0009] The preprocessing subsystem is connected to the data collection subsystem, the data assimilation subsystem, and the coupled model calculation subsystem, and consists of a regional ocean-atmosphere-wave grid generation module, a high-precision terrain processing module, a wave thermal start preprocessing module, an atmospheric preprocessing module, and an ocean preprocessing module.

[0010] The regional ocean-atmosphere-wave grid generation module is connected to the data collection subsystem, the coupled model calculation subsystem, the data assimilation subsystem, the high-precision terrain processing module, the atmospheric preprocessing module, the ocean preprocessing module, and the wave thermal start preprocessing module. The regional ocean-atmosphere-wave grid generation module reads the longitude and latitude of the forecast area from the configuration file sent by the data collection subsystem, receives the locally smoothed terrain data from the high-precision terrain processing module, performs vertical interpolation and land-sea marking on the atmospheric terrain data in the locally smoothed terrain data to generate an atmospheric model grid file; performs interpolation and land-sea marking on the ocean terrain data in the locally smoothed terrain data, marks and processes the positions of river channels to generate an ocean model grid file and a wave model grid file. Sends the atmospheric model grid file to the atmospheric preprocessing module; sends the ocean model grid file to the ocean preprocessing module; sends the wave model grid file to the wave thermal start preprocessing module.

[0011] The high-precision terrain processing module is connected to the data collection subsystem and the regional ocean-atmosphere-wave grid generation module, obtains high-precision terrain data from the data collection subsystem, classifies and archives and locally smooths the high-precision terrain data to obtain locally smoothed terrain data suitable for the regional refined ocean-atmosphere-wave coupled numerical prediction system, and sends the locally smoothed terrain data to the regional ocean-atmosphere-wave grid generation module.

[0012] The wave thermal start preprocessing module is connected to the data collection subsystem and the regional ocean-atmosphere-wave grid generation module, obtains the wave model grid file from the regional ocean-atmosphere-wave grid generation module, receives the regional wave forecast field from the data collection subsystem, interpolates the regional wave forecast field according to the wave model grid file to generate a wave model boundary field, and sends the wave model boundary field and the wave model grid file to the coupled model calculation subsystem.

[0013] The atmospheric preprocessing module is connected to the data collection subsystem, the regional ocean-atmosphere-wave grid generation module, and the data assimilation subsystem, obtains the regional atmospheric forecast field from the data collection subsystem, obtains the atmospheric model grid file from the regional ocean-atmosphere-wave grid generation module, decodes the regional atmospheric forecast field, interpolates the decoded atmospheric forecast field onto the atmospheric model grid file to obtain an atmospheric initial field and an atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem.

[0014] The ocean preprocessing module is connected to the data collection subsystem and the regional ocean-atmosphere-wave grid generation module. It obtains the regional ocean forecast field, tidal data, and regional atmosphere forecast field from the data collection subsystem, receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module, performs grid vertical interpolation preprocessing on the regional ocean forecast field and tidal data to obtain the ocean initial field and ocean boundary field of the regional ocean model, and sends the ocean model grid file, ocean initial field, and ocean boundary field to the data assimilation subsystem; interpolates the regional atmosphere forecast field onto the ocean model grid file to obtain the atmospheric forcing field of the ocean model and sends the atmospheric forcing field to the data assimilation subsystem; sends the ocean model grid file to the coupled model calculation subsystem.

[0015] The data assimilation subsystem is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an atmospheric data assimilation module and an ocean data assimilation module. The atmospheric data assimilation module is connected to the data collection subsystem, the atmospheric preprocessing module of the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an atmospheric data preprocessing sub-module, a background error covariance sub-module, and an atmospheric assimilation sub-module. The atmospheric data preprocessing sub-module is connected to the data collection subsystem and the atmospheric assimilation sub-module, obtains the atmospheric observation file from the data collection subsystem, performs data preprocessing and quality control on the atmospheric observation file to generate an atmospheric observation temporary data file, and sends the atmospheric observation temporary data file to the atmospheric assimilation sub-module. The background error covariance sub-module is connected to the atmospheric assimilation sub-module, uses the background field information in the background error covariance sub-module to calculate the background error covariance, generates a background error covariance matrix, and sends the background error covariance matrix to the atmospheric assimilation sub-module; the atmospheric assimilation sub-module is connected to the data preprocessing sub-module, the background error covariance sub-module, and the preprocessing subsystem, obtains the atmospheric initial field and atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the background error covariance matrix from the background error covariance sub-module, obtains the atmospheric observation temporary data file from the atmospheric data preprocessing sub-module, performs three-dimensional variational assimilation on the atmospheric initial field and atmospheric boundary field to generate the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem.

[0016] The ocean data assimilation module is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an ocean data preprocessing sub-module and a four-dimensional variational sub-module. The ocean data preprocessing sub-module is connected to the data collection subsystem and the four-dimensional variational sub-module, receives ocean observation files from the data collection subsystem, performs data preprocessing and quality control on the ocean observation files, generates ocean observation temporary data files, and sends them to the four-dimensional variational sub-module; the four-dimensional variational sub-module is connected to the ocean preprocessing module, the ocean data preprocessing sub-module, and the coupled model calculation subsystem of the preprocessing subsystem, receives the ocean model grid file, the ocean initial field, the boundary field, and the atmospheric forcing field from the ocean preprocessing module, receives the ocean observation temporary data file from the ocean data preprocessing sub-module, performs variational assimilation calculations on the ocean observation temporary data file, obtains the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the ocean model module of the coupled model calculation subsystem.

[0017] The coupled model calculation subsystem is connected to the preprocessing subsystem, the data assimilation subsystem, and the postprocessing subsystem, and consists of an atmospheric model module, an ocean model module, a wave model module, and a coupling module.

[0018] The coupling module uses a domestic coupler, C-Coupler 3.0, and is connected to the atmospheric model module (Weather Research and Forecasting Model, WRF), the ocean model module (Coastal and Regional Ocean Community model, CROCO), and the wave model module (ECWAM). Its function is to initialize the coupled model calculation subsystem and perform data exchange between the atmospheric model module, the ocean model module, and the wave model module according to the forecast time, the atmospheric coupling exchange step, the ocean coupling exchange step, and the wave coupling exchange step read from the configuration file sent by the data collection subsystem. The coupling module receives the configuration file from the data collection subsystem and performs initial parameter configuration on the coupled model calculation subsystem. When the integration calculation of the atmospheric model module reaches the atmospheric coupling exchange step preset in the configuration file, the coupling module obtains atmospheric variables from the atmospheric model module. The atmospheric variables include longwave radiation, shortwave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data. When the integration calculation of the ocean model module reaches the ocean coupling exchange step preset in the configuration file, the coupling module obtains ocean variables from the ocean model module. The ocean variables include sea surface temperature, sea surface current, water level, water depth, and bottom roughness data. When the integration calculation of the wave model module reaches the wave coupling exchange step preset in the configuration file, the coupling module obtains wave variables from the wave model module. The wave variables include significant wave height, wavelength, wave direction, and wave stress data. The coupling module performs data exchange between the atmospheric model module, the ocean model module, and the wave model module respectively.

[0019] The atmospheric model module is connected to the atmospheric data assimilation module of the data assimilation subsystem, the post-processing subsystem, and the coupling module. It receives the assimilated atmospheric analysis field (as the initial condition of the numerical equations in the atmospheric model module) and the assimilated atmospheric boundary field (as the boundary condition of the numerical equations in the atmospheric model module) from the atmospheric data assimilation module of the data assimilation subsystem, and performs numerical integration calculation on the non-hydrostatic atmospheric numerical equations in the atmospheric model module based on the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to obtain atmospheric variables and the atmospheric forecast field. It sends the atmospheric forecast field to the post-processing subsystem and transfers the longwave radiation, shortwave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data in the atmospheric variables to the coupling module. Through the coupling module, it transfers the longwave radiation, shortwave radiation, latent heat, sensible heat, wind stress, evaporation, and precipitation data to the ocean model module and transfers the 10-meter wind data to the wave model module. It receives ocean variables (including sea surface temperature and sea surface current) and wave variables (including significant wave height, wavelength, wave direction, and wave stress data) from the coupling module, updates the atmospheric variables, and achieves the purpose of two-way coupling (i.e., pairwise variable exchange between the ocean, atmosphere, and waves) at the ocean-atmosphere and wave-atmosphere interfaces.

[0020] The ocean model module is connected to the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem, the ocean data assimilation module of the data assimilation subsystem, the post - processing subsystem, and the coupling module. It receives the ocean model grid file from the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem, and receives the assimilated ocean analysis field (as the initial condition of the numerical equations in the ocean model module) and the assimilated ocean boundary field (as the boundary condition of the numerical equations in the ocean model module) from the ocean data assimilation module of the data assimilation subsystem. According to the assimilated ocean analysis field and the assimilated ocean boundary field, it performs numerical integration calculations on the non - hydrostatic ocean numerical equations in the ocean model module to obtain ocean variables and an ocean forecast field. It sends the ocean forecast field to the post - processing subsystem, and transfers the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data in the ocean variables to the coupling module. Through the coupling module, it transfers the sea surface temperature and sea surface current data to the atmospheric model module, and transfers the sea surface current, water level, water depth, and bottom roughness data to the wave model module. It receives the long - wave radiation, short - wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10 - meter wind, significant wave height, wave length, wave direction, and wave stress data in the atmospheric variables through the coupling module, updates the ocean variables, and achieves the purpose of two - way coupling at the ocean - atmosphere and wave - ocean interfaces.

[0021]

[0022] ​The post - processing subsystem is connected to the coupled - mode calculation subsystem and consists of the post - processing module for regional atmospheric model output, the post - processing module for regional ocean model output, and the post - processing module for regional wave model output. The post - processing module for regional atmospheric model output receives the atmospheric forecast field from the atmospheric model module of the coupled - mode calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea - level pressure, air temperature at 2 - meter height, relative humidity at 2 - meter height, visibility, U - wind component at 10 - meter height, and V - wind component at 10 - meter height in the atmospheric forecast field. These components are interpolated onto standard isobaric surfaces and saved as the first - part NetCDF file of the atmospheric forecast field. And it extracts the accumulated precipitation and humidity at each vertical layer in the atmospheric forecast field, calculates the accumulated precipitation at 3 - hour grid points and the radar reflectivity diagnostic analysis quantity and saves them as the second - part NetCDF file of the atmospheric forecast field. The above two - part NetCDF files of the atmospheric forecast field constitute the atmospheric analysis file. The post - processing module for regional ocean model output receives the ocean forecast field from the ocean model module of the coupled - mode calculation subsystem, extracts the ocean temperature, flow field, and salinity variables in the ocean forecast field, interpolates them onto isobaths from 0m to 1000m depth, and obtains the sea - surface temperature, sea - surface flow direction, sea - surface flow velocity, ocean temperature, salinity, and flow field data at each layer and saves them as the first - part NetCDF file of the ocean forecast field. Using the interpolated ocean temperature and salinity at each layer above, it calculates the ocean density and sound speed at each layer and obtains the second - part NetCDF file of the ocean forecast field. The above two - part NetCDF files of the ocean forecast field constitute the ocean analysis file. The post - processing module for regional wave model output receives the wave forecast field from the wave model module of the coupled - mode calculation subsystem, extracts the significant wave height, wave direction, and period variables in the wave forecast field, and saves them as independent NetCDF files respectively to obtain the wave analysis file. The atmospheric analysis file, ocean analysis file, and wave analysis file constitute the high - precision forecast products output by the regional refined air - sea - wave coupled numerical prediction system.

[0023] In the second step, the data collection subsystem obtains the configuration file, and the data related to the atmosphere, ocean, and waves from the dedicated - line database, and sends the data related to the atmosphere, ocean, and waves to the pre - processing subsystem and the data assimilation subsystem respectively, and sends the configuration file to the pre - processing subsystem, the data assimilation subsystem, and the coupled - mode calculation subsystem. The method is as follows:

[0024] 2.1 Data collection subsystem The data collection subsystem obtains the configuration file from the dedicated line database, reads the longitude and latitude of the forecast area therein, and obtains the global atmospheric forecast field data with a starting forecast time of the current day and a forecast duration of 10 days from the dedicated line database. The global atmospheric forecast field data is output every 3 hours, and each output is saved separately as a data file. Therefore, the global atmospheric forecast field data is a collection of a set of data files. Due to the excessive amount of data in the global atmospheric forecast field data, in order to reduce the storage pressure, the global atmospheric forecast field data is first trimmed according to the longitude and latitude of the forecast area. The temperature, pressure, humidity, and wind field variable files at the same time step are merged into a regional atmospheric forecast field at a single time step. The regional atmospheric forecast field is sequentially saved in the standard GRIB format according to the date and data source, and the regional atmospheric forecast field is sent to the atmospheric preprocessing module of the preprocessing subsystem.

[0025] 2.2 Data collection subsystem The data collection subsystem obtains the global ocean forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of one day from the dedicated line database. Each output of the global ocean forecast field data outputs sea surface temperature, salinity, sea surface height, and flow field variables and saves them separately as sea surface temperature, salinity, sea surface height, and flow field variable data files. Therefore, the global ocean forecast field data is a collection of a set of data files (that is, there are 10 time steps of data, and each time step has four independent small files of sea surface temperature, salinity, sea surface height, and flow field vector respectively). It is trimmed according to the forecast area range, the sea surface temperature, salinity, sea surface height, and flow field variable files at the same time step are merged into a regional ocean forecast field, and are sequentially saved as a set of standard NetCDF formats according to the date and data source, and the regional ocean forecast field is sent to the ocean preprocessing module of the preprocessing subsystem.

[0026] 2.3 Data collection subsystem The data collection subsystem obtains the global wave forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of one day from the dedicated line database. Each output of the global wave forecast field data outputs significant wave height, wave period, wavelength, and wave direction variables and saves them separately as significant wave height, wave period, wavelength, and wave direction variable data files. Therefore, the global wave forecast field data is a collection of a set of data files. It is trimmed according to the forecast area range, the significant wave height, wave period, wavelength, and wave direction variable files are merged into a regional wave forecast field, and are saved in the standard NetCDF format according to the date and data source, and the regional wave forecast field is sent to the wave thermal startup preprocessing module of the preprocessing subsystem.

[0027] 2.4 The data collection subsystem obtains high-precision terrain data and tidal data from the dedicated line database. The high-precision terrain data, such as ETOP2 dataset and GEO dataset, include bathymetry, rivers, lakes, forests, grasslands, ice and snow surfaces, deserts, land surfaces, and corresponding soil temperature and soil moisture information. The tidal data comes from the TPXO tidal data provided by Oregon State University (OSU), including four semi-diurnal tides: M2, S2, N2, K2, four diurnal tides K1, O1, P1, Q1, and fractional tides such as Mf, Mm, M4, MS4, MN4, with a resolution of 0.25°×0.25°. The high-precision terrain data and tidal data are sent to the ocean preprocessing module of the preprocessing subsystem;

[0028] 2.5 The data collection subsystem obtains real-time atmospheric observation data and real-time oceanic observation data from the dedicated line database, processes the real-time atmospheric observation data into binary files, renames them according to the data source and date (Universal Time, i.e., UTC), and archives them in NetCDF format to obtain atmospheric observation files; processes the real-time oceanic observation data into binary files, renames them according to the data source, and archives them in NetCDF format to obtain oceanic observation files. The atmospheric observation files are sent to the atmospheric data assimilation module of the data assimilation subsystem, and the oceanic observation files are sent to the oceanic data assimilation module of the data assimilation subsystem.

[0029] 2.6 The data collection subsystem obtains configuration files from the dedicated line database and sends the configuration files to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem.

[0030] In the third step, the regional air-sea-wave grid generation module of the preprocessing subsystem receives the configuration file from the data collection subsystem, reads the configuration file parameters from the configuration file (including forecast time, forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, map projection method, number of grid points in the atmospheric model, number of grid points in the ocean model, number of grid points in the wave model, and grid resolutions of the atmospheric model module, ocean model module, and wave model module), receives the locally smoothed terrain data from the high-precision terrain processing module of the preprocessing subsystem, and generates an atmospheric model grid file, an ocean model grid file, and a wave model grid file. The method is as follows:

[0031] 3.1 The regional air-sea-wave grid generation module reads the forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, map projection method, number of grid points in the atmospheric model, and grid resolution of the atmospheric model module from the configuration file, generates an initial grid for the atmospheric model, and saves it as a NetCDF format file;

[0032] 3.2 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the ocean model prediction area, the map projection method, the number of ocean model grid points, and the grid resolution of the ocean model module from the configuration file, generates the initial ocean model grid, and saves it as a NetCDF - format file;

[0033] 3.3 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the prediction area and the number of wave model grid points from the configuration file, generates the initial wave model grid, and saves it as a data file with the suffix.grb according to the format requirements of the wave model;

[0034] 3.4 The high - precision terrain processing module receives high - precision terrain data from the data collection subsystem, linearly filters and smooths the ocean points and grid points with large terrain gradients (i.e., large water depth changes) in the high - precision terrain data, marks rivers and waterways, classifies and archives terrain data of different land types and different resolutions, obtains the locally smoothed atmospheric and oceanic terrain data, and sends the locally smoothed terrain data to the regional ocean - atmosphere - wave grid generation module.

[0035] 3.5 The regional ocean - atmosphere - wave grid generation module receives the smoothed atmospheric and oceanic terrain data from the high - precision terrain processing module, and processes the initial atmospheric model grid, the initial ocean model grid, and the initial wave model grid respectively according to the type of terrain, the land - sea boundary, and the type of river waterways in the terrain data:

[0036] 3.5.1 Assign values to the land - sea masks in the initial atmospheric model grid, the initial ocean model grid, and the initial wave model grid, mark ocean points as 1 and land points as 0, and at the same time mark waterways at the river estuaries, with the waterway position as 1 and the coastal land as 0;

[0037] 3.5.2 Interpolate the altitude of the smoothed atmospheric terrain data, different land types, including rivers, lakes, forests, grasslands, ice - snow surfaces, deserts, land surfaces, and the corresponding soil temperature and soil humidity information into the initial atmospheric model grid to generate an atmospheric model grid file;

[0038] 3.5.3 Interpolate the water depth, land - sea boundary, and river waterway information of the smoothed oceanic terrain data onto the initial ocean model grid and the initial wave model grid respectively to generate an ocean model grid file and a wave model grid file;

[0039] 3.6 The regional ocean - atmosphere - wave grid generation module sends the atmospheric model grid file to the atmospheric pre - processing module, sends the ocean model grid file to the ocean pre - processing module, and sends the wave grid file to the wave warm - start pre - processing module.

[0040] Step 4: The atmospheric preprocessing module of the preprocessing subsystem receives the configuration file and the regional atmospheric forecast field from the data collection subsystem, and receives the atmospheric model grid file from the regional ocean-atmosphere-wave grid generation module. It decodes and interpolates the regional atmospheric forecast field to obtain the atmospheric initial field and the atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem. The method is as follows:

[0041] 4.1 The atmospheric preprocessing module receives the configuration file, the regional atmospheric forecast field from the data collection subsystem, and the atmospheric model grid file from the regional ocean-atmosphere-wave grid generation module;

[0042] 4.2 The atmospheric preprocessing module links the regional atmospheric forecast field to the working directory. Since the file name of the regional atmospheric forecast field contains time, symbols, and variable identifiers, the file name is too long and difficult to process in batches. Therefore, in order to facilitate batch processing of all regional atmospheric forecast field files, first rename all the data in the regional atmospheric forecast field in the form of a soft link according to the file type and English letters, in alphabetical order by data format (GRIBFILE).

[0043] 4.3 The atmospheric preprocessing module makes a variable table according to the variable (wind speed, wind direction, temperature, geopotential height variable) type and name of the global atmospheric forecast field data, and decodes and interpolates the global atmospheric forecast field through WPS (WRF Preprocessing System). The variable table records the GRIB encoding, variable name, variable unit, and variable layer information of the global atmospheric forecast field data, which are the basis for the subsequent variable interpolation processing of the regional atmospheric model.

[0044] 4.4 The atmospheric preprocessing module reads the longitude and latitude of the forecast area from the configuration file, extracts the required variables (wind speed, wind direction, temperature, geopotential height variable) from the GRIB encoding of the global atmospheric forecast field according to the variable table, and writes them into an intermediate format file to form intermediate format meteorological data;

[0045] 4.5 The atmospheric preprocessing module uses the interpolation algorithm in the WRF model to horizontally interpolate the intermediate format meteorological data onto the atmospheric model grid file to produce the atmospheric initial field and the atmospheric boundary field;

[0046] 4.6 The atmospheric preprocessing module sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem.

[0047] Step 5: The ocean preprocessing module of the preprocessing subsystem receives the configuration file, the regional ocean forecast field, the tidal data, and the regional atmospheric forecast field from the data collection subsystem, and receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module. It performs interpolation preprocessing on the regional ocean forecast field, the tidal data, and the regional atmospheric forecast field to obtain the atmospheric forcing field, the ocean initial field, and the ocean boundary field, and sends the ocean model grid file, the atmospheric forcing field, the ocean initial field, and the ocean boundary field to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem. The method is as follows:

[0048] 5.1 The ocean preprocessing module receives the configuration file, the regional ocean forecast field, the tidal data, and the regional atmospheric forecast field from the data collection subsystem, and receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module;

[0049] 5.2 The ocean preprocessing module screens the regional ocean forecast field according to the grid point positions in the grid file, determines the default data at the corresponding grid point positions as outliers, marks them as NAN (meaning invalid values), obtains the screened regional ocean forecast field, and saves it as a temporary file in NetCDF format;

[0050] 5.3 The ocean preprocessing module interpolates the screened regional ocean forecast field according to the ocean model grid file to generate the ocean initial field and the ocean boundary field;

[0051] 5.4 The ocean preprocessing module interpolates the regional atmospheric forecast field to generate the atmospheric forcing field;

[0052] 5.5 The ocean preprocessing module sends the atmospheric forcing field, the ocean model initial field, the ocean boundary field, and the ocean model grid file to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem.

[0053] Step 6: The wave warm start preprocessing module of the preprocessing subsystem obtains the regional wave forecast field from the data collection subsystem, receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module, interpolates the regional wave forecast field to obtain the wave boundary field, and sends the wave boundary field and the wave model grid file to the coupled model calculation subsystem. The method is as follows:

[0054] 6.1 The wave warm start preprocessing module receives the regional wave forecast field from the data collection subsystem and receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module;

[0055] 6.2 The wave preprocessing module interpolates the regional wave forecast field according to the wave model grid file to generate the wave boundary field;

[0056] 6.3 The wave hot start preprocessing module sends the wave boundary field and wave model grid files to the coupled mode calculation subsystem.

[0057] In the seventh step, the atmospheric data assimilation module of the data assimilation subsystem receives the atmospheric initial field and atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the atmospheric observation file and configuration file from the data collection subsystem, pre-processes the atmospheric observation file, obtains the atmospheric observation temporary data file, calculates the background error covariance of the simulation area to obtain the background error covariance matrix, and finally performs regional atmospheric three-dimensional variational assimilation to obtain the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled mode calculation subsystem. The method is:

[0058] 7.1 The atmospheric data pre-processing submodule receives the atmospheric observation files and configuration files from the data collection subsystem, converts the various observation data files in the atmospheric observation files into ASCII or PREPBUFR formats recognizable by the atmospheric assimilation submodule; saves the observation error files in the atmospheric observation files into text documents and sends them to the atmospheric assimilation submodule;

[0059] 7.2 The atmospheric data pre-processing submodule removes the observations that are not in the assimilation area and assimilation time window according to the assimilation area and assimilation time window in the configuration file, reorders the variable data in the atmospheric observation file according to time, calculates the air pressure or height from the observed variables using the fluid statics hypothesis, checks the vertical consistency and superadiabatic conditions of multi-layer observation variables, sets the observation error according to the observation error file in the atmospheric observation data, obtains the atmospheric observation temporary data file and sends it to the atmospheric assimilation submodule;

[0060] 7.3 Background error covariance submodule randomly generates a set of background field disturbances, standardizes each disturbance (ensuring that its mean is zero), generates standardized disturbances, and then uses the standardized disturbances to calculate the background error covariance matrix, the formula is: Where N is the number of disturbances (N is usually between 20 and 100, determined according to computing resources and application scenarios), is a randomly generated standardized disturbance, T represents the matrix transpose operation, and the background error covariance matrix B is sent to the atmospheric assimilation submodule;

[0061] 7.4 The atmospheric assimilation submodule receives the atmospheric initial field and atmospheric boundary field from the preprocessing subsystem, receives the atmospheric observation temporary data file from the atmospheric data preprocessing submodule, receives the background error covariance matrix from the background error covariance submodule, performs three-dimensional variational assimilation, obtains the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the atmospheric model module of the coupled model calculation subsystem.

[0062] Step 8: The ocean data assimilation module of the data assimilation subsystem receives the ocean observation file and the configuration file from the data collection subsystem, receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the ocean preprocessing module, performs regional ocean four-dimensional variational assimilation on the ocean observation file to obtain the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem. The method is as follows:

[0063] 8.1 The ocean data preprocessing submodule of the ocean data assimilation module receives the ocean observation file and the configuration file from the data collection subsystem, sets according to the assimilation area, assimilation time window, and variable names in the configuration file, eliminates the observations in the ocean observation file that are not within the assimilation area and assimilation time window, eliminates the values recorded as default values (such as 9999) in the ocean observation file, obtains the valid ocean observation file, reorders and integrates these valid ocean observation files according to the assimilation area, assimilation time window, and variable names, generates the ocean observation temporary data file, and sends the ocean observation temporary data file to the four-dimensional variational submodule;

[0064] 8.2 The four-dimensional variational submodule receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the preprocessing subsystem, receives the ocean observation temporary data file from the ocean data preprocessing submodule, performs incremental four-dimensional variational calculation on the ocean observation temporary data file to generate the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem. The specific method is as follows:

[0065] 8.2.1 The four-dimensional variational submodule performs grid sparsification processing on the ocean model grid file to obtain the low-resolution ocean grid file;

[0066] 8.2.2 The four-dimensional variational submodule interpolates the atmospheric forcing field, the ocean initial field, and the ocean boundary field to the low-resolution ocean grid file to obtain the low-resolution atmospheric forcing field, ocean initial field, and ocean boundary field;

[0067] 8.2.3 The four-dimensional variational submodule extracts the observation positions and times from the ocean observation file, and performs four-dimensional variational assimilation calculation at the observation time on the low-resolution atmospheric forcing field, ocean initial field, ocean boundary field, and ocean observation temporary data file at low resolution to obtain the low-resolution four-dimensional variational analysis increment, and stores the low-resolution four-dimensional variational analysis increment as a NetCDF format temporary file;

[0068] 8.2.4 The four-dimensional variational sub-module uses low-resolution four-dimensional variational analysis increments and ocean model grid files, and through linear interpolation, obtains high-resolution four-dimensional variational analysis increments, which are stored as temporary NetCDF format files;

[0069] 8.2.5 The four-dimensional variational sub-module adds the high-resolution four-dimensional variational analysis increments to the high-resolution background field in the ocean observation file (the high-resolution background field is the climatological statistical result of the ocean model and comes from the ocean observation file sent by the data collection subsystem), obtains the high-resolution assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem.

[0070] Step 9: The coupled model calculation subsystem receives the assimilated atmospheric analysis field, the assimilated atmospheric boundary field, the assimilated ocean analysis field, and the assimilated ocean boundary field from the data assimilation subsystem, receives the configuration file from the data collection subsystem, and receives the ocean model grid file, the wave model grid file, and the wave boundary field from the preprocessing subsystem. The coupled module initializes the subsystem parameters and starts the atmospheric model module (WRF), the ocean model module (CROCO), and the wave model module (ECWAM), integrates the state equations of the atmosphere, ocean, and waves respectively, and exchanges variables among the three model modules through the coupled module to obtain the forecast fields of the atmosphere, ocean, and waves and sends them to the post-processing subsystem. The specific method is as follows:

[0071] 9.1 The coupled module receives the configuration file from the data collection subsystem and initializes the working directory of the coupled model calculation subsystem, the allocation of parallel nodes, the start date (t s ) and end date (t e ) of the integration of the atmospheric model module, the ocean model module, and the wave model module, the integration time step (Ta) of the atmospheric model module, the integration time step (To) of the ocean model module, the integration time step (Tw) of the wave model module, the atmospheric coupling exchange time step (Tca), the ocean coupling exchange time step (Tco), the wave coupling exchange time step (Tcw), the output path, and the output frequency, and sends start signals to the atmospheric model module, the ocean model module, and the wave model module;

[0072] 9.2 After receiving the start signals respectively, the atmospheric model module, the ocean model module, and the wave model module perform parallel start-up integration calculations, output of coupled variables, and output of corresponding forecast fields. Among them, the atmospheric model module executes step 9.2.1, the ocean model module executes step 9.2.2, and the wave integration module executes step 9.2.3. The three model modules perform integration calculations in parallel.

[0073] 9.2.1 The atmospheric model module receives a start signal from the coupling module and the assimilated atmospheric analysis field and the assimilated atmospheric boundary field from the atmospheric data assimilation module of the data assimilation subsystem. It uses the assimilated atmospheric analysis field as the initial value of the state equation system in the atmospheric model module and the assimilated atmospheric boundary field as the boundary condition of the state equation system in the atmospheric model module, and starts the non-hydrostatic atmospheric model integration calculation to obtain the atmospheric forecast field. The method is as follows:

[0074] 9.2.1.1 Initialize the atmospheric model module, and the start date of integration is t s at 00:00 (UTC). At the initial moment, the integration step n = 0. When Ta is the integration step length of the atmospheric model module, n(Ta) is the current total integration time 0 of the atmospheric model module at the initial moment;

[0075] 9.2.1.2 Let n = n + 1. The atmospheric model module performs non-hydrostatic atmospheric model integration calculation to obtain the atmospheric forecast field, which is a NetCDF format file. As the number of integration steps n increases, the time of the integration result of the atmospheric model module advances backward from t s at 00:00 (UTC) by the time of each integration step Ta, and the integration results in the atmospheric forecast field increase continuously.

[0076] 9.2.1.3 Let n(Ta) = the total integration time of the atmospheric model module integration calculation. If n(Ta) = t e -t s , then the integration of the atmospheric model module ends. The atmospheric forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and go to step 9.3; if n(Ta) < t e -t s , go to step 9.2.1.4;

[0077] 9.2.1.4 If n(Ta) < t e -t s and nTa is an integer multiple of the atmospheric coupling exchange step length (Tca), then the atmospheric model module sends long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind to the coupling module; at the same time, the atmospheric model module receives the sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the atmospheric model module with the received sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data, and goes to step 9.2.1.2 to perform the non-hydrostatic atmospheric model integration for the n + 1 step; if n(Ta) < t e -t s and n(Ta) is not an integer multiple of Tca, no data output and exchange are performed, and directly go to step 9.2.1.2 to perform the integration calculation for the n + 1 time step.

[0078] 9.2.2 The ocean model module receives a start signal from the coupling module, an ocean model grid file from the ocean preprocessing module of the preprocessing subsystem, and the assimilated ocean analysis field (as the initial condition for the ocean state equations in the ocean model module) and the assimilated ocean boundary field (as the boundary condition for the ocean state equations in the ocean model module) from the ocean data assimilation module of the data assimilation subsystem, and starts the non-hydrostatic ocean model integration calculation to obtain the ocean forecast field.

[0079] 9.2.2.1 Initialize the ocean model module, and the start date of integration is t s At 00:00 (Universal Time), the initial integration step n = 0. When To is the integration step length of the ocean model module, n(To) is the current total integration time of the ocean model module at the initial moment, which is 0;

[0080] 9.2.2.2 Let n = n + 1, and the ocean model module performs non-hydrostatic ocean state numerical integration (i.e., non-hydrostatic ocean model integration) calculation to obtain the ocean forecast field, which is a NetCDF format file. As the integration step n increases, the time of the integration result of the ocean model module moves backward continuously from 00:00 of t s at an interval of To for each integration step, and the integration results in the ocean forecast field increase continuously.

[0081] 9.2.2.3 Let n(To) = the total integration time of the ocean model module integration calculation. If n(To) = t e -t s , then the integration of the ocean model module ends, the ocean forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and go to step 9.3; if n(To) < t e -t s , go to step 9.2.2.4;

[0082] 9.2.2.4 If n(To) < t e -t s and n(To) is an integer multiple of the ocean coupling exchange step length (Tco), then the ocean model module sends the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data to the coupling module; at the same time, the ocean model module receives the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the ocean model module with the received long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data, and goes to step 9.2.2.2 for the n + 1 step of non-hydrostatic ocean model integration; if n(To) < t e -t sAnd when n(To) is not an integer multiple of Tco, no data output or exchange is performed, and directly go to step 9.2.2.2 for the integration calculation of the (n + 1)-th time step.

[0083] After the wave model receives the start signal, the module receives the wave model grid file and the wave boundary field file from the wave warm start preprocessing module of the preprocessing subsystem, and starts the numerical integration calculation of the wave state.

[0084] 9.2.3.1 Initialize the wave model module. At 00:00 (Universal Time) of the start date t of the integration, the initial integration step n = 0. When Tw is used as the integration step of the wave model module, n(Tw) is the current total integration time 0 of the wave model module at the initial moment. Check whether there is a wave initial field file hotfile.txt in the current working directory (this file is a temporary file generated after the previous integration. If the previous integration does not exist, there is no such file). If hotfile.txt exists, the wave model module reads hotfile.txt as the initial condition of the wave spectrum equation in the wave model module at n = 0, and go to step 9.2.3.2; if hotfile.txt does not exist, all the initial values of the wave height, wavelength, wave period, wave direction parameter, and wave energy spectrum in the wave spectrum equation of the wave model module are initialized to 0, and go to step 9.2.3.2; s

[0085] 9.2.3.2 Let n = n + 1, and the wave model module performs the numerical integration calculation of the wave state to obtain the wave forecast field, which is a NetCDF format file. As the integration step n increases, the time of the integration result of the wave model module moves backward continuously from 00:00 of t at each integration step of Tw, and the integration results in the wave forecast field increase continuously. s

[0086] 9.2.3.3 Let n(Tw) = the total integration time of the wave model module. If n(Tw) = t e -t s e , then the integration of the wave model module ends, the wave forecast field is sent to the post-processing subsystem, the integration completion signal is sent to the coupling module. At the same time, a wave warm start text file is generated and saved as the temporary file hotfile.txt in the current working directory, which is used as the initial state field in the wave model module when the next coupled forecast is started (the main content of the hotfile.txt file is the wave state information, including wave height, wavelength, wave period, wave direction parameter, wave energy spectrum; boundary information, including the sea surface wind field; grid information, including the spatial resolution and layout information of the grid), and go to step 9.3. If n(Tw) < t e -t s , go to step 9.2.3.4;​

[0087] 9.2.3.4 If n(Tw) < t e -t s and n(Tw) is an integer multiple of the ocean wave coupling exchange step length (Tcw), the ocean wave model module sends the significant wave height, wavelength, wave direction, and wave stress data to the coupling module; at the same time, the ocean model module receives the 10-meter wind, sea surface current, water level, water depth, and bottom roughness data sent by the coupling module, replaces the corresponding variables in the ocean wave model module with the received 10-meter wind, sea surface current, water level, water depth, and bottom roughness data, and proceeds to step 9.2.3.2 for the (n + 1)-step ocean wave model integration; if n(Tw) < t e -t s and n(Tw) is not an integer multiple of Tcw, no data output and exchange are performed, and directly proceed to step 9.2.3.2 for the integration calculation at the (n + 1)th time step.

[0088] 9.3 When the coupling module receives the integration completion signals sent by the atmosphere, ocean, and ocean wave model modules, the coupling model calculation process ends and the working log text file log.txt is output to save the basic information during the system operation, which is convenient for future inspection or query records in case of errors. The content of the log.txt file includes the number of parallel cores of the three models, the initial state of the atmosphere model module, the initial state of the ocean model module, the initial state of the ocean wave model module, the integration time, the names of the coupling variables, and the integration end signal.

[0089] Step 10, the post-processing subsystem receives the atmosphere forecast field, ocean forecast field, and ocean wave forecast field from the coupling model calculation subsystem, performs interpolation or diagnostic quantity calculation on the atmosphere forecast field, ocean forecast field, and ocean wave forecast field, and generates regional refined air-sea-wave coupled numerical forecast products. The method is as follows:

[0090] 10.1 The regional atmosphere model output post-processing module receives the atmosphere forecast field from the atmosphere model module of the coupling model calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea level pressure, 2-meter air temperature, 2-meter relative humidity, visibility, 10-meter U wind component, and 10-meter V wind component in the atmosphere forecast field, interpolates the components to the standard isobaric surfaces, saves them as the first part of the atmosphere forecast field NetCDF file, and extracts the accumulated precipitation and humidity at each vertical layer in the atmosphere forecast field, calculates the accumulated precipitation and radar reflectivity diagnostic analysis quantity at each 3-hour grid point and saves them as the second part of the atmosphere forecast field NetCDF file. The above two parts of the atmosphere forecast field NetCDF files constitute the atmosphere analysis file;

[0091] 10.2 The post - processing module for regional ocean model output receives the ocean forecast fields from the ocean model module of the coupled model calculation subsystem, extracts the ocean temperature, flow field, and salinity variables from the ocean forecast fields, interpolates them onto the isobaths from 0m to 1000m depth, obtains sea surface temperature, sea surface flow direction, sea surface flow velocity, ocean temperature, salinity, and flow field data at each layer, and saves them as the first - part ocean forecast field NetCDF file. Using the ocean temperature and salinity at each layer interpolated above, calculates the density and sound speed at each layer of the ocean, obtains the second - part ocean forecast field NetCDF file, and the above two - part ocean forecast field NetCDF files constitute the ocean analysis file;

[0092] 10.3 The post - processing module for regional wave model output receives the wave forecast fields from the wave model module of the coupled model calculation subsystem, extracts the significant wave height, wave direction, wave speed, and period from the wave forecast fields, and saves them as independent NetCDF files respectively, obtaining the wave analysis file. The atmospheric analysis file, ocean analysis file, and wave analysis file form the high - precision forecast products output by the regional refined air - sea - wave coupled numerical prediction system.

[0093] The following beneficial effects can be achieved by using the present invention:

[0094] (1) Based on a domestic coupler, the present invention designs coupling variables respectively in the ninth step, integrates the non - hydrostatic atmospheric model, integrates the non - hydrostatic ocean model, and integrates the wave model, which can make full use of the advantages of the non - hydrostatic model in depicting the refined atmospheric and ocean processes at a resolution from 1km to 100m, and significantly improves the fineness of the atmospheric and ocean forecast products in the key - concerned areas.

[0095] (2) In the ninth step of the present invention, by using the advantage of the coupler to only exchange and update the key variables at the interfaces of the non - hydrostatic atmosphere, non - hydrostatic ocean, and wave models, the originally relatively closed calculation methods for separately calculating the internal process changes of the atmosphere, ocean, and waves themselves can become an integrated system that can perform high - frequency two - way transmission and update variables at the interfaces of the atmosphere, ocean, and wave models (steps 9.2.1.4, step 9.2.2.4, and step 9.2.3.4), achieving the purpose of timely and reasonable description of air - sea interaction, thus improving the accuracy of coupled prediction. At the same time, it can minimize the impact on the integrity of the single - model framework, ensuring the relative independence of each model module and being easy to upgrade and maintain.

[0096] (3) In the eighth and ninth steps of the present invention, the atmospheric three - dimensional variational assimilation and ocean four - dimensional variational assimilation methods with multi - source observational data are added in the initial field production, respectively improving the accuracy of the initial fields of the atmospheric model module and the non - hydrostatic ocean model module, and effectively improving the prediction accuracy of the regional refined air - sea - wave coupled numerical prediction system.

[0097] (4) In step 3.4, the processing of ultra-high resolution terrain and river waterways is added, enabling the regional refined ocean-atmosphere-wave coupled numerical prediction system to have the refined prediction ability for meso-scale and small-scale ocean phenomena and nearshore ocean processes.

[0098] (5) In the fourth step, by marking and smoothing the atmospheric and ocean terrain data, the stability of the regional refined ocean-atmosphere-wave coupled numerical prediction system is effectively improved, and the number of calculation non-convergences is significantly reduced.

[0099] (6) In step 9.2.3.3, the wave heat start text file (hotfile.txt) generated by the wave model module in the ocean-atmosphere coupling model calculation subsystem is used as the initial wave field for the next prediction, realizing the warm start of the wave model module and significantly improving the prediction effect of the sea state. The warm start time of the traditional prediction method is shortened from 1.5 hours to 30 minutes, and the calculation time is reduced by 67%. Description of the Drawings

[0100] Figure 1 It is the logical structure diagram of the regional refined ocean-atmosphere-wave coupled numerical prediction system of the present invention.

[0101] Figure 2 It is the logical structure diagram of the preprocessing subsystem.

[0102] Figure 3 It is the logical structure diagram of the data assimilation subsystem.

[0103] Figure 4 It is the logical structure diagram of the coupling model calculation subsystem.

[0104] Figure 5 It is the logical structure diagram of the post-processing subsystem.

[0105] Figure 6 It is the overall flow chart of the present invention. Detailed Implementation Modes

[0106] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0107] As Figure 6 shown, the present invention includes the following steps:

[0108] In the first step, a regional refined ocean-atmosphere-wave coupled numerical prediction system is constructed. The regional refined ocean-atmosphere-wave coupled numerical prediction system, as Figure 1 shown, is composed of a data collection subsystem, a preprocessing subsystem, a data assimilation subsystem, a coupling model calculation subsystem, and a post-processing subsystem.

[0109] The data collection subsystem is connected to a dedicated line database (usually installed on the servers of a high-performance computing center), a preprocessing subsystem, and a data assimilation subsystem. It reads configuration files, global atmospheric forecast field data (GRIB format) for the current day's forecast, global ocean wave forecast field data, global ocean forecast field data, high-precision terrain data, tidal data, real-time atmospheric observation data, and real-time ocean observation data from the dedicated line database. For the global atmospheric forecast field data, it is clipped according to the forecast area's longitude and latitude (read from the configuration file), and the three-dimensional air temperature, geopotential height, relative humidity, and wind vector field variable files are merged into a global atmospheric forecast field variable file. Then, according to the date and data source, a regional atmospheric forecast field is generated and saved in the standard GRIB format. For the global ocean forecast field data, it is clipped according to the forecast area range, and the sea surface temperature, salinity, sea surface height, and flow field variable files are merged into a global ocean forecast field variable file, and a regional ocean forecast field is obtained and saved in the standard NetCDF format. For the global ocean wave forecast field data, it is clipped and archived according to the forecast area range, and the significant wave height, wave period, wavelength, and wave direction variables are merged to obtain a regional ocean wave forecast field and saved in the standard NetCDF format. The real-time atmospheric observation data is classified and processed. The real-time atmospheric observation data is renamed as an atmospheric observation file according to the data source and archived in ASCII or PREPBUFR format, and the observation error file in the atmospheric observation file is archived in text document format. The real-time ocean observation data is classified and processed. The real-time ocean observation data is renamed as an ocean observation file according to the data source or type and archived in NetCDF format, and the climatological statistical results of the ocean model in the ocean observation file are archived in chronological order. The generated regional atmospheric forecast field, regional ocean wave forecast field, regional ocean forecast field, as well as the tidal data and high-precision terrain data obtained from the dedicated line database are sent to the preprocessing subsystem, and the atmospheric observation file and ocean observation file are sent to the data assimilation subsystem. The content of the configuration file includes: working directory, forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, number of parallel nodes, map projection method, forecast time (i.e., the start date of integration t s and the end date of integration t e ), the grid resolution and number of grid points of the atmospheric model module, the grid resolution and number of grid points of the ocean model module, the grid resolution and number of grid points of the ocean wave model module, the integration time step of the atmospheric model module (Ta), the integration time step of the ocean model module (To), the integration time step of the ocean wave model module (Tw), the atmospheric coupling exchange time step (Tca), the ocean coupling exchange time step (Tco), the ocean wave coupling exchange time step (Tcw), output path and output frequency, assimilation area, assimilation time window, and variable names for assimilation. The configuration file is sent to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem.

[0110] The preprocessing subsystem is connected to the data collection subsystem, the data assimilation subsystem, and the coupled model calculation subsystem, and consists of a regional air-sea-wave grid generation module, a high-precision terrain processing module, a sea-wave warm start preprocessing module, an atmospheric preprocessing module, and an ocean preprocessing module.

[0111] As Figure 2 shown, the regional air-sea-wave grid generation module is connected to the data collection subsystem, the coupled model calculation subsystem, the data assimilation subsystem, the high-precision terrain processing module, the atmospheric preprocessing module, the ocean preprocessing module, and the sea-wave warm start preprocessing module. The regional air-sea-wave grid generation module reads the longitude and latitude of the forecast area from the configuration file sent by the data collection subsystem, receives the locally smoothed terrain data from the high-precision terrain processing module, performs vertical interpolation and land-sea marking on the atmospheric terrain data in the locally smoothed terrain data, and generates an atmospheric model grid file; performs interpolation and land-sea marking on the ocean terrain data in the locally smoothed terrain data, marks and processes the positions of river channels, and generates an ocean model grid file and a sea-wave model grid file. Sends the atmospheric model grid file to the atmospheric preprocessing module; sends the ocean model grid file to the ocean preprocessing module; sends the sea-wave model grid file to the sea-wave warm start preprocessing module.

[0112] The high-precision terrain processing module is connected to the data collection subsystem and the regional air-sea-wave grid generation module, obtains high-precision terrain data from the data collection subsystem, classifies and archives and locally smooths the high-precision terrain data to obtain locally smoothed terrain data suitable for the regional refined air-sea-wave coupled numerical prediction system, and sends the locally smoothed terrain data to the regional air-sea-wave grid generation module.

[0113] The sea-wave warm start preprocessing module is connected to the data collection subsystem and the regional air-sea-wave grid generation module, obtains the sea-wave model grid file from the regional air-sea-wave grid generation module, receives the regional sea-wave forecast field from the data collection subsystem, interpolates the regional sea-wave forecast field according to the sea-wave model grid file to generate a sea-wave model boundary field, and sends the sea-wave model boundary field and the sea-wave model grid file to the coupled model calculation subsystem.

[0114] The atmospheric preprocessing module is connected to the data collection subsystem, the regional air-sea-wave grid generation module, and the data assimilation subsystem, obtains the regional atmospheric forecast field from the data collection subsystem, obtains the atmospheric model grid file from the regional air-sea-wave grid generation module, decodes the regional atmospheric forecast field, interpolates the decoded atmospheric forecast field onto the atmospheric model grid file to obtain an atmospheric initial field and an atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem.

[0115] The ocean preprocessing module is connected to the data collection subsystem and the regional ocean-atmosphere-wave grid generation module. It obtains the regional ocean forecast field, tidal data, and regional atmosphere forecast field from the data collection subsystem, receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module, performs grid vertical interpolation preprocessing on the regional ocean forecast field and tidal data to obtain the ocean initial field and ocean boundary field of the regional ocean model, and sends the ocean model grid file, ocean initial field, and ocean boundary field to the data assimilation subsystem; interpolates the regional atmosphere forecast field onto the ocean model grid file to obtain the atmospheric forcing field of the ocean model, and sends the atmospheric forcing field to the data assimilation subsystem; sends the ocean model grid file to the coupled model calculation subsystem.

[0116] As Figure 3 shown, the data assimilation subsystem is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an atmospheric data assimilation module and an ocean data assimilation module. The atmospheric data assimilation module is connected to the data collection subsystem, the atmospheric preprocessing module of the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an atmospheric data preprocessing sub-module, a background error covariance sub-module, and an atmospheric assimilation sub-module. The atmospheric data preprocessing sub-module is connected to the data collection subsystem and the atmospheric assimilation sub-module, obtains the atmospheric observation file from the data collection subsystem, performs data preprocessing and quality control on the atmospheric observation file to generate an atmospheric observation temporary data file, and sends the atmospheric observation temporary data file to the atmospheric assimilation sub-module. The background error covariance sub-module is connected to the atmospheric assimilation sub-module, uses the background field information in the background error covariance sub-module to calculate the background error covariance, generates a background error covariance matrix, and sends the background error covariance matrix to the atmospheric assimilation sub-module; the atmospheric assimilation sub-module is connected to the data preprocessing sub-module, the background error covariance sub-module, and the preprocessing subsystem, obtains the atmospheric initial field and atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the background error covariance matrix from the background error covariance sub-module, obtains the atmospheric observation temporary data file from the atmospheric data preprocessing sub-module, performs three-dimensional variational assimilation on the atmospheric initial field and atmospheric boundary field to generate the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem.

[0117] The ocean data assimilation module is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem, and consists of an ocean data preprocessing sub-module and a four-dimensional variational sub-module. The ocean data preprocessing sub-module is connected to the data collection subsystem and the four-dimensional variational sub-module, receives ocean observation files from the data collection subsystem, performs data preprocessing and quality control on the ocean observation files, generates ocean observation temporary data files, and sends them to the four-dimensional variational sub-module; the four-dimensional variational sub-module is connected to the ocean preprocessing module of the preprocessing subsystem, the ocean data preprocessing sub-module, and the coupled model calculation subsystem, receives the ocean model grid file, the ocean initial field, the boundary field, and the atmospheric forcing field from the ocean preprocessing module, receives the ocean observation temporary data file from the ocean data preprocessing sub-module, performs variational assimilation calculations on the ocean observation temporary data file, obtains the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the ocean model module of the coupled model calculation subsystem.

[0118] As Figure 4 shown, the coupled model calculation subsystem is connected to the preprocessing subsystem, the data assimilation subsystem, and the postprocessing subsystem, and consists of an atmospheric model module, an ocean model module, a wave model module, and a coupling module.

[0119] The coupling module uses the domestic coupler C-Coupler3.0 and is connected to the atmospheric model module, the ocean model module, and the wave model module. Its function is to initialize the coupled model calculation subsystem and perform data exchange between the atmospheric model module, the ocean model module, and the wave model module according to the forecast time, the atmospheric coupling exchange step, the ocean coupling exchange step, and the wave coupling exchange step read from the configuration file sent by the data collection subsystem. The coupling module receives the configuration file from the data collection subsystem and performs initial parameter configuration on the coupled model calculation subsystem; when the integration calculation of the atmospheric model module reaches the preset atmospheric coupling exchange step in the configuration file, the coupling module obtains the atmospheric variables from the atmospheric model module. The atmospheric variables include long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data; when the integration calculation of the ocean model module reaches the preset ocean coupling exchange step in the configuration file, the coupling module obtains the ocean variables from the ocean model module. The ocean variables include sea surface temperature, sea surface current, water level, water depth, and bottom roughness data; when the integration calculation of the wave model module reaches the preset wave coupling exchange step in the configuration file, the coupling module obtains the wave variables from the wave model module. The wave variables include significant wave height, wavelength, wave direction, and wave stress data; the coupling module performs data exchange between the atmospheric model module, the ocean model module, and the wave model module respectively.

[0120] The atmospheric model module is connected to the atmospheric data assimilation module, the post-processing subsystem, and the coupling module of the data assimilation subsystem. It receives the assimilated atmospheric analysis field (as the initial condition of the numerical equations in the atmospheric model module) and the assimilated atmospheric boundary field (as the boundary condition of the numerical equations in the atmospheric model module) from the atmospheric data assimilation module of the data assimilation subsystem. Based on the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, it performs numerical integration calculations on the non-hydrostatic atmospheric numerical equations in the atmospheric model module to obtain atmospheric variables and an atmospheric forecast field. It sends the atmospheric forecast field to the post-processing subsystem and transfers the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data in the atmospheric variables to the coupling module. Through the coupling module, it transfers the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, and precipitation data to the ocean model module and transfers the 10-meter wind data to the wave model module. It receives ocean variables (including sea surface temperature and sea surface current) and wave variables (including significant wave height, wavelength, wave direction, and wave stress data) from the coupling module, updates the atmospheric variables, and achieves the purpose of two-way coupling (i.e., pairwise exchange of variables among the ocean, atmosphere, and waves) at the ocean-atmosphere and wave-atmosphere interfaces.

[0121] The ocean model module is connected to the regional ocean-atmosphere-wave grid generation module of the preprocessing subsystem, the ocean data assimilation module of the data assimilation subsystem, the post-processing subsystem, and the coupling module. It receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module of the preprocessing subsystem and the assimilated ocean analysis field and the assimilated ocean boundary field from the ocean data assimilation module of the data assimilation subsystem. Based on the assimilated ocean analysis field and the assimilated ocean boundary field, it performs numerical integration calculations on the non-hydrostatic ocean numerical equations in the ocean model module to obtain ocean variables and an ocean forecast field. It sends the ocean forecast field to the post-processing subsystem and transfers the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data in the ocean variables to the coupling module. Through the coupling module, it transfers the sea surface temperature and sea surface current data to the atmospheric model module and transfers the sea surface current, water level, water depth, and bottom roughness data to the wave model module. It receives the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data in the atmospheric variables from the coupling module, updates the ocean variables, and achieves the purpose of two-way coupling at the ocean-atmosphere and wave-ocean interfaces.

[0122] The wave model module is connected to the regional air-sea wave grid generation module of the preprocessing subsystem, the wave warm start preprocessing module of the preprocessing subsystem, the coupling module, and the postprocessing subsystem. It receives the wave model grid file from the regional air-sea wave grid generation module of the preprocessing subsystem and the wave boundary field from the wave warm start preprocessing module. It performs numerical integration calculations based on the wave boundary field and the static initial field with a default initial value of zero (or the warm start file hotfile.txt generated by the previous forecast) to obtain wave variables and the wave forecast field. It sends the wave forecast field to the postprocessing subsystem and transfers the significant wave height, wave length, wave direction, and wave stress data in the wave variables to the coupling module. Through the coupling module, it transfers the significant wave height, wave length, wave direction, and wave stress data to the atmospheric model module and the ocean model module. It receives the 10-meter wind, sea surface current, water level, water depth, and bottom roughness data in the ocean variables through the coupling module to update the wave variables, achieving the purpose of two-way coupling at the ocean-wave and wave-ocean interfaces.

[0123] Such as Figure 5As shown in the figure, the post-processing subsystem is connected to the coupled model calculation subsystem and consists of an atmospheric model output post-processing module for the regional atmospheric model, an ocean model output post-processing module for the regional ocean model, and a wave model output post-processing module for the regional wave model. The atmospheric model output post-processing module for the regional atmospheric model receives the atmospheric forecast fields from the atmospheric model module of the coupled model calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea level pressure, air temperature at 2 m height, relative humidity at 2 m height, visibility, U wind component at 10 m height, and V wind component at 10 m height from the atmospheric forecast fields, interpolates the components onto standard isobaric surfaces respectively, saves them as the first part of the atmospheric forecast field NetCDF file, and extracts the accumulated precipitation and humidity at each vertical layer from the atmospheric forecast fields, calculates the accumulated precipitation at 3-hourly grid points, the radar reflectivity diagnostic analysis quantity and saves them as the second part of the atmospheric forecast field NetCDF file. The above two parts of the atmospheric forecast field NetCDF files constitute the atmospheric analysis file; the ocean model output post-processing module for the regional ocean model receives the ocean forecast fields from the ocean model module of the coupled model calculation subsystem, extracts the ocean temperature, flow field and salinity variables from the ocean forecast fields, interpolates them onto isobaths from 0 m to 1000 m depth, obtains the sea surface temperature, sea surface current direction, sea surface current velocity, ocean temperature, salinity and flow field data at each layer and saves them as the first part of the ocean forecast field NetCDF file, uses the interpolated ocean temperature and salinity at each layer above to calculate the ocean density and sound speed at each layer, obtains the second part of the ocean forecast field NetCDF file. The above two parts of the ocean forecast field NetCDF files constitute the ocean analysis file; the wave model output post-processing module for the regional wave model receives the wave forecast fields from the wave model module of the coupled model calculation subsystem, extracts the significant wave height, wave direction and period variables from the wave forecast fields, saves them as independent NetCDF files respectively, and obtains the wave analysis file. The atmospheric analysis file, ocean analysis file and wave analysis file constitute the high-precision forecast products output by the regional refined air-sea-wave coupled numerical forecasting system.

[0124] In the second step, the data collection subsystem obtains the configuration file, the data related to the atmosphere, ocean and waves from the dedicated line database, sends the data related to the atmosphere, ocean and waves to the preprocessing subsystem and the data assimilation subsystem respectively, and sends the configuration file to the preprocessing subsystem, the data assimilation subsystem and the coupled model calculation subsystem. The method is as follows:

[0125] 2.1 Data collection subsystem The data collection subsystem obtains the configuration file from the dedicated line database, reads the longitude and latitude of the forecast area therein, and obtains the global atmospheric forecast field data with a starting forecast time of the current day and a forecast duration of 10 days from the dedicated line database. The global atmospheric forecast field data is output every 3 hours, and each output is saved as a separate data file. Therefore, the global atmospheric forecast field data is a collection of a set of data files. Due to the excessive data volume of the global atmospheric forecast field data, in order to reduce the storage pressure, the global atmospheric forecast field data is first trimmed according to the longitude and latitude of the forecast area, and the temperature, pressure, humidity, and wind field variable files at the same time step are merged into a regional atmospheric forecast field at a single time step. The regional atmospheric forecast field is sequentially saved in the standard GRIB format according to the date and data source (for example, "ATMO_2025010100.grb", that is, the data is ATMO, and the time of the data is 00:00 on January 1, 2025), and the regional atmospheric forecast field is sent to the atmospheric preprocessing module of the preprocessing subsystem.

[0126] 2.2 Data collection subsystem The data collection subsystem obtains the global ocean forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of once a day from the dedicated line database. Each output of the global ocean forecast field data outputs sea surface temperature, salinity, sea surface height, and flow field variables and saves them as sea surface temperature, salinity, sea surface height, and flow field variable data files respectively. Therefore, the global ocean forecast field data is a collection of a set of data files (that is, there are 10 time steps of data, and each time step has four independent small files of sea surface temperature, salinity, sea surface height, and flow field vector respectively). It is trimmed according to the forecast area range, and the sea surface temperature, salinity, sea surface height, and flow field variable files at the same time step are merged into a regional ocean forecast field, and are sequentially saved in a set of standard NetCDF formats according to the date and data source, and the regional ocean forecast field is sent to the ocean preprocessing module of the preprocessing subsystem.

[0127] 2.3 Data collection subsystem The data collection subsystem obtains the global wave forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of once a day from the dedicated line database. Each output of the global wave forecast field data outputs significant wave height, wave period, wavelength, and wave direction variables and saves them as significant wave height, wave period, wavelength, and wave direction variable data files respectively. Therefore, the global wave forecast field data is a collection of a set of data files. It is trimmed according to the forecast area range, and the significant wave height, wave period, wavelength, and wave direction variable files are merged into a regional wave forecast field, and are saved in the standard NetCDF format according to the date and data source, and the regional wave forecast field is sent to the wave thermal start preprocessing module of the preprocessing subsystem.

[0128] 2.4 The data collection subsystem obtains high-precision terrain data and tidal data from the dedicated line database. The high-precision terrain data, such as ETOP2 dataset and GEO dataset, includes bathymetry, rivers, lakes, forests, grasslands, ice and snow surfaces, deserts, land surfaces, and corresponding soil temperature and soil moisture information. The tidal data comes from the TPXO tidal data provided by Oregon State University (OSU), including four semidiurnal tides: M2, S2, N2, K2, four diurnal tides K1, O1, P1, Q1, and fractional tides Mf, Mm, M4, MS4, MN4, with a resolution of 0.25°×0.25°. The high-precision terrain data and tidal data are sent to the ocean preprocessing module of the preprocessing subsystem;

[0129] 2.5 The data collection subsystem obtains real-time atmospheric observation data and real-time oceanic observation data from the dedicated line database, processes the real-time atmospheric observation data into binary files, renames them according to the data source and date (Universal Time, i.e., UTC), and archives them in NetCDF format (e.g., "KTYY_2025010100.txt", which means the data comes from KTYY and the time of the data is 00:00 on January 1, 2025) to obtain atmospheric observation files; processes the real-time oceanic observation data into binary files, renames them according to the data source, and archives them in NetCDF format to obtain oceanic observation files. The atmospheric observation files are sent to the atmospheric data assimilation module of the data assimilation subsystem, and the oceanic observation files are sent to the oceanic data assimilation module of the data assimilation subsystem.

[0130] 2.6 The data collection subsystem obtains configuration files from the dedicated line database and sends the configuration files to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem.

[0131] In the third step, the regional air-sea-wave grid generation module of the preprocessing subsystem receives the configuration file from the data collection subsystem, reads the configuration file parameters from the configuration file (including forecast time, forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, map projection method, number of grid points in the atmospheric model, number of grid points in the ocean model, number of grid points in the wave model, and grid resolution of the atmospheric model module, grid resolution of the ocean model module, and grid resolution of the wave model module), receives the locally smoothed terrain data from the high-precision terrain processing module of the preprocessing subsystem, and generates an atmospheric model grid file, an ocean model grid file, and a wave model grid file. The method is as follows:

[0132] 3.1 The regional air-sea-wave grid generation module reads the forecast area longitude and latitude, central longitude and latitude, reference longitude and latitude, map projection method, number of grid points in the atmospheric model, and grid resolution of the atmospheric model module from the configuration file, generates an initial grid for the atmospheric model, and saves it as a NetCDF format file;

[0133] 3.2 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the forecast area, map projection method, number of ocean model grid points, and grid resolution of the ocean model module from the configuration file, generates the initial ocean model grid, and saves it as a NetCDF - format file;

[0134] 3.3 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the forecast area and the number of wave model grid points from the configuration file, generates the initial wave model grid, and saves it as a data file with the suffix grb according to the format requirements of the wave model;

[0135] 3.4 The high - precision terrain processing module receives high - precision terrain data from the data collection subsystem, linearly filters and smooths the ocean points and grid points with large terrain gradients (i.e., large water depth changes) in the high - precision terrain data, marks rivers and waterways, classifies and archives terrain data of different land types and different resolutions, obtains the locally smoothed atmospheric and oceanic terrain data, and sends the locally smoothed terrain data to the regional ocean - atmosphere - wave grid generation module.

[0136] 3.5 The regional ocean - atmosphere - wave grid generation module receives the smoothed atmospheric and oceanic terrain data from the high - precision terrain processing module, and processes the initial atmospheric model grid, initial ocean model grid, and initial wave model grid according to the type of terrain, land - sea boundary, and type of river and waterway in the terrain data:

[0137] 3.5.1 Assign values to the land - sea masks in the initial atmospheric model grid, initial ocean model grid, and initial wave model grid, mark ocean points as 1 and land points as 0, and at the same time mark waterways at the river mouths, with the waterway position as 1 and the coastal land as 0;

[0138] 3.5.2 Interpolate the altitude of the smoothed atmospheric terrain data and information on different land types, including rivers, lakes, forests, grasslands, ice - snow surfaces, deserts, and land surfaces, corresponding soil temperature and soil moisture information into the initial atmospheric model grid to generate an atmospheric model grid file;

[0139] 3.5.3 Interpolate the water depth, land - sea boundary, and river - waterway information of the smoothed oceanic terrain data onto the initial ocean model grid and initial wave model grid respectively to generate an ocean model grid file and a wave model grid file;

[0140] 3.6 The regional ocean - atmosphere - wave grid generation module sends the atmospheric model grid file to the atmospheric pre - processing module, sends the ocean model grid file to the ocean pre - processing module, and sends the wave grid file to the wave warm - start pre - processing module.

[0141] Step 4: The atmospheric preprocessing module of the preprocessing subsystem receives the configuration file and the regional atmospheric forecast field from the data collection subsystem, and receives the atmospheric model grid file from the regional ocean-atmosphere-wave grid generation module. It decodes and interpolates the regional atmospheric forecast field to obtain the atmospheric initial field and the atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem. The method is as follows:

[0142] 4.1 The atmospheric preprocessing module receives the configuration file, the regional atmospheric forecast field from the data collection subsystem, and the atmospheric model grid file from the regional ocean-atmosphere-wave grid generation module;

[0143] 4.2 The atmospheric preprocessing module links the regional atmospheric forecast field to the working directory. Since the file name of the regional atmospheric forecast field contains time, symbols, and variable identifiers, the file name is too long and difficult to process in batches. First, in the form of a soft link, all the data in the regional atmospheric forecast field are renamed according to the file type and the English alphabet, in the data format (GRIBFILE) plus alphabetical order (for example, GRIBFILE.AAA represents the file ATMO_GG_2022083118.grb)

[0144] 4.3 The atmospheric preprocessing module makes a variable table according to the variable (wind speed, wind direction, temperature, geopotential height variable) type and name of the global atmospheric forecast field data, and decodes and interpolates the global atmospheric forecast field through WPS (WRF Preprocessing System: WRF Preprocessing System). The variable table records the GRIB encoding, variable name, variable unit, and variable layer number information of the global atmospheric forecast field data, which are the basis for the subsequent variable interpolation processing of the regional atmospheric model.

[0145] 4.4 The atmospheric preprocessing module reads the longitude and latitude of the forecast area from the configuration file, extracts the required variables (wind speed, wind direction, temperature, geopotential height variable) from the GRIB encoding of the global atmospheric forecast field according to the variable table, and writes them into the intermediate format file to form the intermediate format meteorological data;

[0146] 4.5 The atmospheric preprocessing module uses the interpolation algorithm in the WRF model to horizontally interpolate the intermediate format meteorological data onto the atmospheric model grid file to produce the atmospheric initial field and the atmospheric boundary field;

[0147] 4.6 The atmospheric preprocessing module sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem.

[0148] Step 5. The ocean preprocessing module of the preprocessing subsystem receives the configuration file, regional ocean forecast field, tidal data, and regional atmospheric forecast field from the data collection subsystem, and receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module. It performs interpolation preprocessing on the regional ocean forecast field, tidal data, and regional atmospheric forecast field to obtain the atmospheric forcing field, ocean initial field, and ocean boundary field, and sends the ocean model grid file, atmospheric forcing field, ocean initial field, and ocean boundary field to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem. The method is as follows:

[0149] 5.1 The ocean preprocessing module receives the configuration file, regional ocean forecast field, tidal data, and regional atmospheric forecast field from the data collection subsystem, and receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module;

[0150] 5.2 The ocean preprocessing module screens the regional ocean forecast field according to the grid point positions in the grid file, determines the default data at the corresponding grid point positions as outliers, marks them as NAN, obtains the screened regional ocean forecast field, and saves it as a temporary file in NetCDF format;

[0151] 5.3 The ocean preprocessing module interpolates the screened regional ocean forecast field according to the ocean model grid file to generate the ocean initial field and ocean boundary field;

[0152] 5.4 The ocean preprocessing module interpolates the regional atmospheric forecast field to generate the atmospheric forcing field;

[0153] 5.5 The ocean preprocessing module sends the atmospheric forcing field, ocean model initial field, ocean boundary field, and ocean model grid file to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem.

[0154] Step 6. The wave warm start preprocessing module of the preprocessing subsystem obtains the regional wave forecast field from the data collection subsystem, receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module, interpolates the regional wave forecast field to obtain the wave boundary field, and sends the wave boundary field and the wave model grid file to the coupled model calculation subsystem. The method is as follows:

[0155] 6.1 The wave warm start preprocessing module receives the regional wave forecast field from the data collection subsystem and receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module;

[0156] 6.2 The wave preprocessing module interpolates the regional wave forecast field according to the wave model grid file to generate the wave boundary field;

[0157] 6.3 The wave warm start preprocessing module sends the wave boundary field and the wave model grid file to the coupled model calculation subsystem.

[0158] Step 7: The atmospheric data assimilation module of the data assimilation subsystem receives the atmospheric initial field and the atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the atmospheric observation file and the configuration file from the data collection subsystem, preprocesses the atmospheric observation file to obtain an atmospheric observation temporary data file, calculates the background error covariance of the simulation area to obtain a background error covariance matrix, and finally performs regional three-dimensional variational assimilation to obtain the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem. The method is as follows:

[0159] 7.1 The atmospheric data preprocessing sub-module receives the atmospheric observation file and the configuration file from the data collection subsystem, converts various observation data files in the atmospheric observation file into ASCII or PREPBUFR formats recognizable by the atmospheric assimilation sub-module; saves the observation error file in the atmospheric observation file as a text document (such as obserr.txt) and sends it to the atmospheric assimilation sub-module;

[0160] 7.2 The atmospheric data preprocessing sub-module eliminates the observations in the atmospheric observation file that are not within the assimilation area and the assimilation time window according to the assimilation area and the assimilation time window in the configuration file, reorders the variable data in the atmospheric observation file according to time, calculates the pressure or height from the observed variables using the hydrostatic assumption, checks the vertical consistency and superadiabatic conditions of the multi-layer observed variables, sets the observation error according to the observation error file in the atmospheric observation data, obtains an atmospheric observation temporary data file and sends it to the atmospheric assimilation sub-module;

[0161] 7.3 The background error covariance sub-module randomly generates a set of background field perturbations, performs normalization processing on each perturbation (ensuring that its mean is zero) to generate normalized perturbations, and then calculates the background error covariance matrix using the normalized perturbations. The formula is: where N is the number of perturbations (N = 20 in the embodiment), is the randomly generated normalized perturbation, T represents the matrix transpose operation, and sends the background error covariance matrix B to the atmospheric assimilation sub-module;

[0162] 7.4 The atmospheric assimilation sub-module receives the atmospheric initial field and the atmospheric boundary field from the preprocessing subsystem, receives the atmospheric observation temporary data file from the atmospheric data preprocessing sub-module, receives the background error covariance matrix from the background error covariance sub-module, performs three-dimensional variational assimilation to obtain the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the atmospheric model module of the coupled model calculation subsystem.

[0163] Step 8: The ocean data assimilation module of the data assimilation subsystem receives the ocean observation file and the configuration file from the data collection subsystem, receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the ocean preprocessing module, performs regional ocean four-dimensional variational assimilation on the ocean observation file to obtain the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem. The method is as follows:

[0164] 8.1 The ocean data preprocessing sub-module of the ocean data assimilation module receives the ocean observation file and the configuration file from the data collection subsystem, sets according to the assimilation area, assimilation time window, and variable names in the configuration file, eliminates the observations in the ocean observation file that are not within the assimilation area and assimilation time window, eliminates the values recorded as default values (such as 9999) in the ocean observation file, obtains the effective ocean observation file, reorders and integrates these effective ocean observation files according to the assimilation area, assimilation time window, and variable names, generates the ocean observation temporary data file, and sends the ocean observation temporary data file to the four-dimensional variational sub-module;

[0165] 8.2 The four-dimensional variational sub-module receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the preprocessing subsystem, receives the ocean observation temporary data file from the ocean data preprocessing sub-module, performs incremental four-dimensional variational calculation on the ocean observation temporary data file to generate the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem. The specific method is as follows:

[0166] 8.2.1 The four-dimensional variational sub-module performs grid sparsification processing on the ocean model grid file to obtain the low-resolution ocean grid file;

[0167] 8.2.2 The four-dimensional variational sub-module interpolates the atmospheric forcing field, the ocean initial field, and the ocean boundary field to the low-resolution ocean grid file to obtain the low-resolution atmospheric forcing field, ocean initial field, and ocean boundary field;

[0168] 8.2.3 The four-dimensional variational sub-module extracts the observation positions and times from the ocean observation file, and performs four-dimensional variational assimilation calculation at the observation time on the low-resolution atmospheric forcing field, ocean initial field, ocean boundary field, and ocean observation temporary data file at low resolution to obtain the low-resolution four-dimensional variational analysis increment, and stores the low-resolution four-dimensional variational analysis increment as a NetCDF format temporary file;

[0169] 8.2.4 The four-dimensional variational sub-module uses low-resolution four-dimensional variational analysis increments and ocean model grid files, and obtains high-resolution four-dimensional variational analysis increments through linear interpolation, and stores them as temporary NetCDF format files;

[0170] 8.2.5 The four-dimensional variational sub-module adds the high-resolution four-dimensional variational analysis increments to the high-resolution background field in the ocean observation file (the high-resolution background field is the climatological statistical result of the ocean model, from the ocean observation file sent by the data collection subsystem), to obtain the high-resolution assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem.

[0171] Step 9, the coupled model calculation subsystem receives the assimilated atmospheric analysis field, the assimilated atmospheric boundary field, the assimilated ocean analysis field and the assimilated ocean boundary field from the data assimilation subsystem, receives the configuration file from the data collection subsystem, and receives the ocean model grid file, the wave model grid file and the wave boundary field from the preprocessing subsystem. The coupled module initializes the subsystem parameters and starts the atmospheric model module (WRF), the ocean model module (CROCO) and the wave model module (ECWAM), integrates the state equations of the atmosphere, ocean and waves respectively, and exchanges variables among the three model modules through the coupled module, to obtain the forecast fields of the atmosphere, ocean and waves and send them to the post-processing subsystem. The specific method is as follows:

[0172] 9.1 The coupled module receives the configuration file from the data collection subsystem, and initializes the working directory of the coupled model calculation subsystem, the allocation of parallel nodes, the integration start date (t s ) and the integration end date (t e ) of the atmospheric model module, the ocean model module and the wave model module, the integration time step (Ta) of the atmospheric model module, the integration time step (To) of the ocean model module, the integration time step (Tw) of the wave model module, the atmospheric coupling exchange time step (Tca), the ocean coupling exchange time step (Tco), the wave coupling exchange time step (Tcw), the output path and the output frequency, and sends start signals to the atmospheric model module, the ocean model module and the wave model module;

[0173] 9.2 After receiving the start signals respectively, the atmospheric model module, the ocean model module and the wave model module perform parallel startup integration calculations, output of coupled variables and output of corresponding forecast fields. Among them, the atmospheric model module executes step 9.2.1, the ocean model module executes step 9.2.2, and the wave integration module executes step 9.2.3, and the three model modules perform integration calculations in parallel.

[0174] 9.2.1 The atmospheric model module receives a start signal from the coupling module, and receives the assimilated atmospheric analysis field and the assimilated atmospheric boundary field from the atmospheric data assimilation module of the data assimilation subsystem. It uses the assimilated atmospheric analysis field as the initial value of the state equation system in the atmospheric model module, and uses the assimilated atmospheric boundary field as the boundary condition of the state equation system in the atmospheric model module, and starts the non-hydrostatic atmospheric model integration calculation to obtain the atmospheric forecast field. The method is as follows:

[0175] 9.2.1.1 Initialize the atmospheric model module, and the start date of integration is t s At 00:00 (Universal Time), at the initial moment, the integration step n = 0. When Ta is the integration step length of the atmospheric model module, n(Ta) is the current total integration time 0 of the atmospheric model module at the initial moment;

[0176] 9.2.1.2 Let n = n + 1, and the atmospheric model module performs non-hydrostatic atmospheric model integration calculation to obtain the atmospheric forecast field. The atmospheric forecast field is a NetCDF format file. As the number of integration steps n increases, the time of the integration result of the atmospheric model module advances backward continuously from 00:00 of t s (UTC) at a time of each integration step Ta, and the integration results in the atmospheric forecast field increase continuously.

[0177] 9.2.1.3 Let n(Ta) = the total integration time of the atmospheric model module integration calculation. If n(Ta) = t e -t s , then the integration of the atmospheric model module ends, the atmospheric forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and go to step 9.3; if n(Ta) < t e -t s , go to step 9.2.1.4;

[0178] 9.2.1.4 If n(Ta) < t e -t s and nTa is an integer multiple of the atmospheric coupling exchange step length (Tca), then the atmospheric model module sends long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind to the coupling module; at the same time, the atmospheric model module receives the sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, and replaces the corresponding variables in the atmospheric model module with the received sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data, and go to step 9.2.1.2 for the n + 1-step non-hydrostatic atmospheric model integration; if n(Ta) < t e -t s and n(Ta) is not an integer multiple of Tca, no data output and exchange are performed, and directly go to step 9.2.1.2 for the integration calculation of the n + 1 time step.

[0179] 9.2.2 The ocean model module receives a start signal from the coupling module, an ocean model grid file from the ocean preprocessing module of the preprocessing subsystem, and the assimilated ocean analysis field (as the initial condition for the ocean state equations in the ocean model module) and the assimilated ocean boundary field (as the boundary condition for the ocean state equations in the ocean model module) from the ocean data assimilation module of the data assimilation subsystem, and starts the non-hydrostatic ocean model integration calculation to obtain the ocean forecast field.

[0180] 9.2.2.1 Initialize the ocean model module, and the start date of integration is t s at 00:00 (Universal Time). At the initial moment, the integration step n = 0. When To is the integration step length of the ocean model module, n(To) is the current total integration time of the ocean model module at the initial moment, which is 0;

[0181] 9.2.2.2 Let n = n + 1, and the ocean model module performs non-hydrostatic ocean state numerical integration (i.e., non-hydrostatic ocean model integration) calculation to obtain the ocean forecast field, which is a NetCDF format file. As the integration step n increases, the time of the integration result of the ocean model module moves backward continuously from t s at 00:00 with a time of each integration step of To, and the integration results in the ocean forecast field increase continuously.

[0182] 9.2.2.3 Let n(To) = the total integration time of the ocean model module integration calculation. If n(To) = t e -t s , then the integration of the ocean model module ends, the ocean forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and go to step 9.3; if n(To) < t e -t s , go to step 9.2.2.4;

[0183] 9.2.2.4 If n(To) < t e -t s and n(To) is an integer multiple of the ocean coupling exchange step length (Tco), then the ocean model module sends the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data to the coupling module; at the same time, the ocean model module receives the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, and replaces the corresponding variables in the ocean model module with the received long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data, and go to step 9.2.2.2 for the n + 1 step of non-hydrostatic ocean model integration; if n(To) < t e -t sAnd when n(To) is not an integer multiple of Tco, no data output or exchange is performed, and directly go to step 9.2.2.2 for the integration calculation of the (n + 1)-th time step.

[0184] After the wave model receives the start signal, the module receives the wave model grid file and the wave boundary field file from the wave warm start preprocessing module of the preprocessing subsystem, and starts the numerical integration calculation of the wave state.

[0185] 9.2.3.1 Initialize the wave model module. At 00:00 (Universal Time) of the start date t s of the integration, the initial integration step n = 0. When Tw is the integration step of the wave model module, n(Tw) is the current total integration time 0 of the wave model module at the initial moment. Determine whether there is a wave initial field file hotfile.txt in the current working directory (this file is a temporary file generated after the previous integration. If there was no previous integration, this file does not exist). If hotfile.txt exists, the wave model module reads hotfile.txt as the initial condition of the wave spectrum equation in the wave model module at n = 0, and go to step 9.2.3.2; if hotfile.txt does not exist, all the initial values of the wave height, wavelength, wave period, wave direction parameter, and wave energy spectrum in the wave spectrum equation of the wave model module are initialized to 0, and go to step 9.2.3.2;

[0186] 9.2.3.2 Let n = n + 1, and the wave model module performs the numerical integration calculation of the wave state to obtain the wave forecast field, which is a NetCDF format file. As the integration step n increases, the time of the integration result of the wave model module advances backward from 00:00 of t s at a time of each integration step Tw, and the integration results in the wave forecast field increase continuously.

[0187] 9.2.3.3 Let n(Tw) = the total integration time of the wave model module. If n(Tw) = t e - t s , then the integration of the wave model module ends. Send the wave forecast field to the post-processing subsystem, send the integration completion signal to the coupling module. At the same time, generate a wave warm start text file, save it as the temporary file hotfile.txt, and store it temporarily in the current working directory as the initial state field in the wave model module when the next coupled forecast starts (the main content of the hotfile.txt file is the wave state information, including wave height, wavelength, wave period, wave direction parameter, wave energy spectrum; boundary information, including the sea surface wind field; grid information, including the spatial resolution and layout information of the grid), and go to step 9.3. If n(Tw) < t e - t s , go to step 9.2.3.4;

[0188] 9.2.3.4 If n(Tw) < t e -t s and n(Tw) is an integer multiple of the ocean wave coupling exchange step length (Tcw), the ocean wave model module will send the significant wave height, wavelength, wave direction, and wave stress data to the coupling module; at the same time, the ocean model module receives the 10-meter wind, sea surface current, water level, water depth, and bottom roughness data sent by the coupling module, replaces the corresponding variables in the ocean wave model module with the received 10-meter wind, sea surface current, water level, water depth, and bottom roughness data, and proceeds to step 9.2.3.2 for the (n + 1)-step ocean wave model integration; if n(Tw) < t e -t s and n(Tw) is not an integer multiple of Tcw, no data output and exchange are performed, and the process directly proceeds to step 9.2.3.2 for the integration calculation at the (n + 1)-th time step.

[0189] 9.3 When the coupling module receives the integration completion signals sent by the atmospheric, oceanic, and ocean wave model modules, the coupling mode calculation process ends and a working log text file log.txt is output to save the basic information during the system operation, which is convenient for future inspection or query records in case of errors. The content of the log.txt file includes the number of parallel cores of the three models, the initial states of the atmospheric model module, the ocean model module, and the ocean wave model module, the integration time, the names of the coupling variables, and the integration end signal.

[0190] Step 10, the post-processing subsystem receives the atmospheric forecast field, ocean forecast field, and ocean wave forecast field from the coupling mode calculation subsystem, performs interpolation or diagnostic quantity calculation on the atmospheric forecast field, ocean forecast field, and ocean wave forecast field, and generates regional refined air-sea-wave coupled numerical forecast products. The method is as follows:

[0191] 10.1 The regional atmospheric model output post-processing module receives the atmospheric forecast field from the atmospheric model module of the coupling mode calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea level pressure, 2-meter air temperature, 2-meter relative humidity, visibility, 10-meter U wind component, and 10-meter V wind component in the atmospheric forecast field, interpolates the components onto the standard isobaric surfaces, saves them as the first part of the atmospheric forecast field NetCDF file, and extracts the accumulated precipitation and humidity at each vertical layer in the atmospheric forecast field, calculates the accumulated precipitation and radar reflectivity diagnostic analysis quantity at each 3-hour grid point and saves them as the second part of the atmospheric forecast field NetCDF file. The above two parts of the atmospheric forecast field NetCDF files constitute the atmospheric analysis file;

[0192] 10.2 The post - processing module for the regional ocean model output receives the ocean forecast fields from the ocean model module of the coupled model calculation subsystem, extracts the ocean temperature, flow field, and salinity variables from the ocean forecast fields, interpolates them onto the isobaths from 0 m to 1000 m depth, obtains the sea surface temperature, sea surface flow direction, sea surface flow velocity, ocean temperature, salinity, and flow field data at each layer, and saves them as the first part of the ocean forecast field NetCDF file. Using the interpolated ocean temperature and salinity at each layer above, calculates the ocean density and sound speed at each layer, and obtains the second part of the ocean forecast field NetCDF file. The above two parts of the ocean forecast field NetCDF files constitute the ocean analysis file;

[0193] 10.3 The post - processing module for the regional wave model output receives the wave forecast fields from the wave model module of the coupled model calculation subsystem, extracts the significant wave height, wave direction, wave speed, and period from the wave forecast fields, and saves them as independent NetCDF files respectively, obtaining the wave analysis file. The atmospheric analysis file, ocean analysis file, and wave analysis file form the high - precision forecast products output by the regional refined air - sea - wave coupled numerical prediction system.

[0194] To verify the effect of the present invention, the following experiments are carried out:

[0195] Hardware configuration requirements: Galaxy high - performance computer system, with Redhat 7.5 operating system, Intel compiler version 19.0.0.117 that supports C, C++, Fortran 77, and Fortran90 languages, and supports parallel computing environment. The microcomputer terminal operating system is Windows 10.

[0196] During the experiment, in the configuration file obtained in the second step, it is set that the current mode is in the directory, the forecast time is from March 10, 2025 to March 15, 2025, the forecast area longitude and latitude are 0 - 45°N, 103 - 150°E, the central longitude and latitude are 22.5°N, 123.0°N, Lambert projection, the atmospheric model grid resolution is 1 km, the number of grid points is (5463×5283), the step size of the atmospheric model module is 5 seconds; the ocean model grid resolution is 1 km, the number of grid points is (5463×5283), the step size of the ocean model module is 15 seconds; the wave model grid resolution is 1 km, the number of grid points is (5463×5283), the step size of the wave model module is 60 seconds; the atmospheric coupling exchange step size is 120 seconds, the ocean coupling exchange step size is 120 seconds, the wave coupling exchange step size is 120 seconds; the assimilation area is 0 - 45°N, 103 - 150°E, the assimilation time window is 6 hours, and the assimilated variable names are: atmospheric vorticity, divergence, temperature, surface pressure, specific humidity, sea surface height anomaly, sea surface temperature, and sea surface salinity; the number N of perturbations in step 7.3 is 20.

[0197] In the first step of the present invention, the regional refined air-sea-wave coupled numerical prediction system constructs a global atmospheric prediction field obtained from a dedicated line database as the GFS (resolution 0.25°, approximately 25 km) product data released by the US Environmental Prediction Center. The global ocean prediction field is the PSY (resolution 1 / 12°, approximately 9 km) released by the European Copernicus Marine Service Center. The global wave prediction field comes from the global wave prediction data (resolution 0.2°, approximately 20 km) released by the European Copernicus Marine Service Center. The tidal data comes from the TPXO tidal data provided by Oregon State University (OSU), including four semidiurnal tides: M2, S2, N2, K2, four diurnal tides K1, O1, P1, Q1, and information on partial tides such as Mf, Mm, M4, MS4, MN4, with a resolution of 0.25°×0.25°. The high-precision terrain is ETOP2 and Geo data packets (including elevation data such as USGS GMTED2010, SRTM, GTOPO30, ASTER GDEM; land use type data such as MODIS, USGS Land Use / Land Cover Data; water vegetation cover and soil type data). The real-time atmospheric observation data includes conventional surface observation data, including temperature, pressure, humidity, wind speed, wind direction, precipitation), and satellite remote sensing observation data, including microwave radiometer data (US AMSU-A, AMSU-B, HIRS, IASI and Chinese FY series), microwave scatterometer data. The real-time ocean observation data includes data such as sea surface height, sea surface temperature and salinity, and flow field observed by buoys, ships and satellites). The regional refined air-sea-wave coupled numerical prediction system performs the second to tenth steps on the data obtained from the dedicated line database to obtain regional atmospheric, oceanic, and wave coupled prediction products with a horizontal resolution of 1 km. The atmospheric prediction products include atmospheric analysis files, ocean analysis files, and wave analysis files. The atmospheric analysis file contains geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, and sea level pressure, air temperature at 2 m height, relative humidity at 2 m height, visibility, U wind component at 10 m height, V wind at 10 m height on the isobaric surfaces from 1000 hPa to 15 hPa in the atmospheric prediction field, as well as cumulative precipitation, radar reflectivity diagnostic analysis quantities at grid points every 3 hours. The ocean analysis file includes sea surface temperature, sea surface flow direction, sea surface flow velocity, temperature, salinity, flow field, density, and sound speed of each layer of the ocean from the sea surface to a depth of 1000 m. The wave analysis file includes significant wave height, wave direction, wave speed, and period.

[0198] As shown in Table 1, the Shanghai Typhoon Institute of the China Meteorological Administration launched a regional ocean-atmosphere-wave coupled typhoon forecasting system in 2019. The resolution of its atmospheric model is 27 km, which cannot meet the requirements of refined forecasting (see the literature: Lei Xiaotu et al., A new generation of regional ocean-atmosphere-wave coupled typhoon forecasting system, 2019). Currently, the resolution of the regional ocean-atmosphere coupled (without a coupled wave model, lacking wave-atmosphere and wave-ocean current coupling processes) operational forecasting system proposed by the China Meteorological Administration is 3 km for the atmosphere and 3 - 5 km for the ocean. In addition, for the Hurricane Analysis and Forecast System (HAFS) in the United States, the resolution of its atmospheric model is a three-layer nested grid of 18 / 6 / 2 km, and the resolution of the ocean model is 1 / 12° (about 9 km). It lacks ocean assimilation and does not have an online two-way coupling function for tides and waves (see the literature Andrew Hazelton et.al., 2022 real-time Hurricane forecasts from an experimental version of the Hurricane analysis and forecast system (HAFSV0.3S); Andrew Hazelton et al., 2022 real-time hurricane forecasts: forecasts based on the experimental version of the Hurricane Analysis and Forecast System (HAFSV0.3S), 2023). This system is mainly oriented towards hurricane forecasting and has no requirements for ocean environmental forecasting. Therefore, the resolution of its coupled ocean model module is about 9 km. In addition, for the Coupled Ocean-Atmosphere-Wave Prediction System (COAMPS) proposed by the US Naval Research Laboratory, although its local resolution can be close to 3 km, its ocean model uses a hydrostatic ocean NCOM model with a resolution of about 1 - 10 km. Additionally, both the French Meteorological Service and the Japan Meteorological Agency have launched operational regional ocean-atmosphere coupled operational forecasting systems. The French Meteorological Service has not announced the resolution of its regional products. However, from its coupled forecasting products for the Mediterranean region, the resolution of its atmospheric forecasting products is about 10 - 25 km, and the resolution of its ocean products is about 2 - 10 km. But it lacks the online coupling function for tides and waves and cannot depict the complete ocean-atmosphere-wave interaction. The operational regional coupled forecasting system of the Japan Meteorological Agency provides atmospheric forecasting products with a resolution of 5 - 20 km, and the ocean forecasting products output through a hydrostatic ocean model have a resolution of about 2 - 10 km, without an online coupled wave forecasting function.As shown in Table 1, the present invention uses a refined model grid with a resolution of 1 km, which can reflect more meso-scale and small-scale processes in the atmosphere and ocean compared to low-resolution forecasting systems, reducing the uncertainty caused by low-resolution models. At the same time, the high-frequency two-way coupling once every 2 minutes can also more accurately depict the interactions among the atmosphere, ocean, and waves. Especially in the forecasting of high-impact weather processes, it has significant advantages.

[0199] Table 1 Comparison of the forecasting effects of the forecasting methods of the present invention and the main regional forecasting systems

[0200]

[0201]

[0202] Studies have shown that compared with the hydrostatic ocean model (see the literature Xiaoyu Fan et.al. Comparison of the Coastal and Regional Ocean Community model (CROCO) and NCAR-LES in non-hydrostatic simulations; Xiaoyu Fan et al., Comparison of the Coastal and Regional Ocean Community model (CROCO) and NCAR-LES in non-hydrostatic simulations), the non-hydrostatic ocean model is more suitable for the simulation of meso-scale and small-scale ocean processes and internal ocean waves. Therefore, the use of the non-hydrostatic ocean model CROCO in the present invention is more suitable for high-resolution forecasting requirements and can more accurately simulate refined ocean processes such as internal waves and meso-scale eddies in the ocean.

[0203] Studies have shown that the third-generation wave model (Wave Watch III, WWIII) is mainly used in current operational wave forecasting. This model is generally more suitable for the simulation and forecasting of global waves, while the SWAN model in the present invention has more advantages in the simulation of nearshore waves. This is mainly because the SWAN model has more physical parameterizations for the friction effect of the seabed. When the seabed friction can be ignored, the simulation effect of the SWAN model on deep-water waves is also comparable to that of WWIII. Therefore, using the SWAN model in the present invention for wave forecasting is more accurate for coastal wave forecasting. In addition, the use of a high-resolution grid can also improve the refinement degree of wave forecasting results.

[0204] Since the warm start cycle of the SWAN mode module is added in the present invention, it can ensure that the initial field of sea waves is closer to the real situation (compared with the case where the initial field is all 0). If the warm start cycle is not added, the usual approach is to start the forecast at least two days earlier (with the forecast end time unchanged) so as to obtain a relatively real sea wave field through the coupled forecast of the first two days. However, this will waste a large amount of computing time. Therefore, using the sea wave warm start method of the present invention can greatly shorten the wall clock time required for the usual forecast.

[0205] In the third and fourth steps of the present invention, when generating the ocean model grid file, the ETOP2 topography is first smoothed and interpolated, which can ensure the stable integration of the ocean model module at relatively steep terrain. However, since the resolution of the ocean model grid (3 km) is higher than that of ETOP2 (about 3.7 km), the river channels within the longitude and latitude of the forecast area are marked and corresponding processing is carried out, which is more in line with the actual geographical information and is more conducive to the calculation at the river estuary; this method is more suitable for refined simulation than the case without corresponding terrain processing.

[0206] In the seventh step, through the data assimilation subsystem, multi-source satellite observation data can be assimilated, which has not been realized in the current research on air-sea-wave coupled models (such as Coupled Ocean-Atmosphere-Wave-Sediment Transport, COAWST), and the ocean data assimilation method has not been realized in the currently known regional air-sea-wave coupled operational forecasting system HAFS. Research shows that using data assimilation technology can significantly improve the accuracy of the initial fields of the atmospheric and ocean model modules.

[0207] In the ninth step, a domestic coupler is used to initialize the process of the coupled model calculation subsystem, especially to control the number of parallel computing nodes of the atmospheric model module, the ocean model module, and the sea wave model module, and different processes are respectively allocated to perform calculations independently. At the same time, the pairwise exchange of coupling physical quantities between the atmospheric, ocean, and sea wave component model modules can be quickly completed at a smaller coupling time step (such as 2 minutes), solving the problem that the boundary conditions of a single model in the current operational forecasting method cannot be updated quickly, so as to obtain a more accurate forecast result.

Claims

1. A regional refined air-sea-wave coupled numerical prediction method based on a non-hydrostatic model, characterized in that It includes the following steps: The first step is to construct a regional refined ocean - atmosphere - wave coupled numerical prediction system, which consists of a data collection subsystem, a pre - processing subsystem, a data assimilation subsystem, a coupled model calculation subsystem, and a post - processing subsystem; The data collection subsystem is connected to the dedicated line database, the preprocessing subsystem, and the data assimilation subsystem, and reads the configuration file, the global atmospheric forecast field data, the global ocean wave forecast field data, the global ocean forecast field data, the high-precision terrain data, the tidal data, the atmospheric real-time observation data, and the ocean real-time observation data reported on the same day from the dedicated line database; for the global atmospheric forecast field data, it is clipped according to the forecast area longitude and latitude read from the configuration file, the three-dimensional air temperature, geopotential height, relative humidity, and wind vector field variable files are merged into the global atmospheric forecast field variable file, and the regional atmospheric forecast field is generated according to the date and data source; for the global ocean forecast field data, it is clipped according to the forecast area range, the sea surface temperature, salinity, sea surface height, and flow field variable files are merged into the global ocean forecast field variable file to obtain the regional ocean forecast field; for the global ocean wave forecast field data, it is clipped and archived according to the forecast area range, and the significant wave height, wave period, wavelength, and wave direction variables are merged to obtain the regional ocean wave forecast field; the atmospheric real-time observation data is classified and processed, and the atmospheric real-time observation data is renamed as the atmospheric observation file according to the data source; the ocean real-time observation data is classified and processed, and the ocean real-time observation data is renamed as the ocean observation file according to the data source or type; the generated regional atmospheric forecast field, regional ocean wave forecast field, regional ocean forecast field, as well as the tidal data and high-precision terrain data obtained from the dedicated line database are sent to the preprocessing subsystem, and the atmospheric observation file and ocean observation file are sent to the data assimilation subsystem; the content of the configuration file includes: the working directory, the forecast area longitude and latitude, the central longitude and latitude, the reference longitude and latitude, the number of parallel nodes, the map projection method, the forecast time, i.e., the integration start date t s and the integration end date t e , the grid resolution and the number of grid points of the atmospheric model module, the grid resolution and the number of grid points of the ocean model module, the grid resolution and the number of grid points of the ocean wave model module, the integration time step Ta of the atmospheric model module, the integration time step To of the ocean model module, the integration time step Tw of the ocean wave model module, the atmospheric coupling exchange time step Tca, the ocean coupling exchange time step Tco, the ocean wave coupling exchange time step Tcw, the output path and output frequency, the assimilation area, the assimilation time window, and the variable names to be assimilated; the configuration file is sent to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem; The pre - processing subsystem is connected to the data collection subsystem, the data assimilation subsystem, and the coupled model calculation subsystem, and consists of a regional ocean - atmosphere - wave grid generation module, a high - precision terrain processing module, a sea - wave warm - start pre - processing module, an atmospheric pre - processing module, and an ocean pre - processing module; The regional ocean - atmosphere - wave grid generation module generates an atmospheric model grid file, an ocean model grid file, and a sea - wave model grid file; sends the atmospheric model grid file to the atmospheric pre - processing module; sends the ocean model grid file to the ocean pre - processing module; sends the sea - wave model grid file to the sea - wave warm - start pre - processing module; The high - precision terrain processing module generates locally smoothed terrain data and sends the locally smoothed terrain data to the regional ocean - atmosphere - wave grid generation module; The sea - wave warm - start pre - processing module generates a sea - wave model boundary field and sends the sea - wave model boundary field and the sea - wave model grid file to the coupled model calculation subsystem; The atmospheric pre - processing module generates an atmospheric initial field and an atmospheric boundary field and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem; The ocean pre - processing module generates an ocean initial field and an ocean boundary field for the regional ocean model, sends the ocean model grid file, the ocean initial field, and the ocean boundary field to the data assimilation subsystem; generates an atmospheric forcing field for the ocean model and sends the atmospheric forcing field to the data assimilation subsystem; sends the ocean model grid file to the coupled model calculation subsystem; The data assimilation subsystem consists of an atmospheric data assimilation module and an ocean data assimilation module; the atmospheric data assimilation module consists of an atmospheric data pre - processing sub - module, a background error covariance sub - module, and an atmospheric assimilation sub - module; the atmospheric data pre - processing sub - module generates an atmospheric observation temporary data file and sends the atmospheric observation temporary data file to the atmospheric assimilation sub - module; the background error covariance sub - module generates a background error covariance matrix and sends the background error covariance matrix to the atmospheric assimilation sub - module; the atmospheric assimilation sub - module generates an assimilated atmospheric analysis field and an assimilated atmospheric boundary field and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem; The ocean data assimilation module consists of an ocean data pre - processing sub - module and a four - dimensional variational sub - module; the ocean data pre - processing sub - module generates an ocean observation temporary data file and sends it to the four - dimensional variational sub - module; the four - dimensional variational sub - module generates an assimilated ocean analysis field and an assimilated ocean boundary field and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the ocean model module of the coupled model calculation subsystem; The coupled model calculation subsystem is connected to the pre - processing subsystem, the data assimilation subsystem, and the post - processing subsystem, and consists of an atmospheric model module, an ocean model module, a sea - wave model module, and a coupling module; The coupling module initializes the coupled model calculation subsystem and exchanges data among the atmospheric model module, the ocean model module, and the sea - wave model module; The atmospheric model module performs numerical integration on the non-hydrostatic atmospheric numerical equations in the atmospheric model module to obtain atmospheric variables and an atmospheric forecast field. The atmospheric forecast field is sent to the post-processing subsystem, and the atmospheric variables are passed to the coupling module and then to the wave model module through the coupling module. The atmospheric variables are updated according to the ocean variables and wave variables received from the coupling module, realizing two-way coupling at the ocean-atmosphere and wave-atmosphere interfaces, that is, pairwise variable exchange among the ocean, atmosphere, and waves. The ocean model module performs numerical integration on the non-hydrostatic ocean numerical equations in the ocean model module to obtain ocean variables and an ocean forecast field. The ocean forecast field is sent to the post-processing subsystem, and the ocean variables are passed to the coupling module. The ocean variables are updated by receiving atmospheric variables through the coupling module, realizing two-way coupling at the ocean-atmosphere and wave-ocean interfaces. The wave model module performs numerical integration to obtain wave variables and a wave forecast field. The wave forecast field is sent to the post-processing subsystem, and the wave variables are passed to the coupling module. The wave variables are updated by receiving atmospheric variables and ocean variables through the coupling module, realizing two-way coupling at the ocean-wave and wave-ocean interfaces. The post-processing subsystem is connected to the coupled model calculation subsystem and consists of the post-processing module for regional atmospheric model output, the post-processing module for regional ocean model output, and the post-processing module for regional wave model output. The post-processing module for regional atmospheric model output generates an atmospheric analysis file. The post-processing module for regional ocean model output generates an ocean analysis file. The post-processing module for regional wave model output generates a wave analysis file. The atmospheric analysis file, ocean analysis file, and wave analysis file constitute the high-precision forecast products output by the regional refined ocean-atmosphere-wave coupled numerical prediction system. In the second step, the data collection subsystem obtains the configuration file, and data related to the atmosphere, ocean, and waves from the dedicated line database, and sends the data related to the atmosphere, ocean, and waves to the preprocessing subsystem and the data assimilation subsystem respectively, and sends the configuration file to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem. In the third step, the regional ocean-atmosphere-wave grid generation module of the preprocessing subsystem receives the configuration file from the data collection subsystem, reads the configuration file parameters from the configuration file. The configuration file parameters include the forecast time, the longitude and latitude of the forecast area, the central longitude and latitude, the reference longitude and latitude, the map projection method, the number of grid points of the atmospheric model, the number of grid points of the ocean model, the number of grid points of the wave model, and the grid resolutions of the atmospheric model module, the ocean model module, and the wave model module. The locally smoothed terrain data is received from the high-precision terrain processing module of the preprocessing subsystem to generate the atmospheric model grid file, the ocean model grid file, and the wave model grid file. The method is as follows: 3.1 The regional ocean-atmosphere-wave grid generation module reads the longitude and latitude of the forecast area, the central longitude and latitude, the reference longitude and latitude, the map projection method, the number of grid points of the atmospheric model, and the grid resolution of the atmospheric model module from the configuration file to generate the initial grid of the atmospheric model and save it as a NetCDF format file. 3.2 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the ocean model forecast area, the map projection method, the number of ocean model grid points, and the grid resolution of the ocean model module from the configuration file, generates the initial ocean model grid, and saves it as a NetCDF - format file; 3.3 The regional ocean - atmosphere - wave grid generation module reads the longitude and latitude of the forecast area and the number of wave model grid points from the configuration file, generates the initial wave model grid, and saves it as a data file with the suffix.grb according to the format requirements of the wave model; 3.4 The high - precision terrain processing module receives high - precision terrain data from the data collection subsystem, performs linear filtering and smoothing on the ocean points and grid points with large terrain gradients in the high - precision terrain data, marks rivers and waterways, classifies and archives terrain data of different land types and different resolutions, obtains the locally smoothed atmospheric and oceanic terrain data, and sends the locally smoothed terrain data to the regional ocean - atmosphere - wave grid generation module; 3.5 The regional ocean - atmosphere - wave grid generation module receives the smoothed atmospheric and oceanic terrain data from the high - precision terrain processing module, and processes the initial atmospheric model grid, the initial ocean model grid, and the initial wave model grid respectively according to the type of terrain, the land - sea boundary, and the type of river and waterway in the terrain data, generating the atmospheric model grid file, the ocean model grid file, and the wave model grid file; 3.6 The regional ocean - atmosphere - wave grid generation module sends the atmospheric model grid file to the atmospheric pre - processing module, the ocean model grid file to the ocean pre - processing module, and the wave grid file to the wave warm - start pre - processing module; Fourthly, the atmospheric pre - processing module of the pre - processing subsystem receives the configuration file and the regional atmospheric forecast field from the data collection subsystem, receives the atmospheric model grid file from the regional ocean - atmosphere - wave grid generation module, decodes and interpolates the regional atmospheric forecast field to obtain the atmospheric initial field and the atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem; Fifthly, the ocean pre - processing module of the pre - processing subsystem receives the configuration file, the regional ocean forecast field, the tidal data, and the regional atmospheric forecast field from the data collection subsystem, receives the ocean model grid file from the regional ocean - atmosphere - wave grid generation module, interpolates and pre - processes the regional ocean forecast field, the tidal data, and the regional atmospheric forecast field to obtain the atmospheric forcing field, the ocean initial field, and the ocean boundary field, sends the ocean model grid file, the atmospheric forcing field, the ocean initial field, and the ocean boundary field to the data assimilation subsystem, and sends the ocean model grid file to the coupled - model calculation subsystem; Sixthly, the wave warm - start pre - processing module of the pre - processing subsystem obtains the regional wave forecast field from the data collection subsystem, receives the wave model grid file from the regional ocean - atmosphere - wave grid generation module, interpolates the regional wave forecast field according to the wave model grid file to obtain the wave boundary field, and sends the wave boundary field and the wave model grid file to the coupled - model calculation subsystem; Step 7: The atmospheric data assimilation module of the data assimilation subsystem receives the atmospheric initial field and the atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the atmospheric observation file and the configuration file from the data collection subsystem, preprocesses the atmospheric observation file to obtain the temporary atmospheric observation data file, calculates the background error covariance of the simulation area to obtain the background error covariance matrix, and finally performs regional three-dimensional variational assimilation of the atmosphere to obtain the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem; Step 8: The ocean data assimilation module of the data assimilation subsystem receives the ocean observation file and the configuration file from the data collection subsystem, receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the ocean preprocessing module, performs regional ocean four-dimensional variational assimilation on the ocean observation file to obtain the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem; The method is as follows: 8.1 The ocean data preprocessing sub-module of the ocean data assimilation module receives the ocean observation file and the configuration file from the data collection subsystem, sets according to the assimilation area, the assimilation time window, and the variable name to be assimilated in the configuration file, eliminates the observations in the ocean observation file that are not within the assimilation area and the assimilation time window, eliminates the values recorded as default values in the ocean observation file, obtains the effective ocean observation file, reorders and integrates these effective ocean observation files according to the assimilation area, the assimilation time window, and the variable name to be assimilated, generates the temporary ocean observation data file, and sends the temporary ocean observation data file to the four-dimensional variational sub-module; 8.2 The four-dimensional variational sub-module receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the preprocessing subsystem, receives the temporary ocean observation data file from the ocean data preprocessing sub-module, performs incremental four-dimensional variational calculation on the temporary ocean observation data file, generates the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem; Step 9: The coupled model calculation subsystem receives the assimilated atmospheric analysis field, the assimilated atmospheric boundary field, the assimilated ocean analysis field, and the assimilated ocean boundary field from the data assimilation subsystem, receives the configuration file from the data collection subsystem, receives the ocean model grid file, the wave model grid file, and the wave boundary field from the preprocessing subsystem, initializes the subsystem parameters through the coupling module and starts the atmospheric model module, the ocean model module, and the wave model module, integrates the state equations of the atmosphere, the ocean, and the waves respectively, exchanges variables of the three model modules through the coupling module, obtains the forecast fields of the atmosphere, the ocean, and the waves and sends them to the post-processing subsystem; The method is as follows: 9.1 The coupling module receives the configuration file from the data collection subsystem, and initializes the working directory of the coupling mode calculation subsystem, the allocation of the number of parallel nodes, the start date t of the integration of the atmospheric mode module, the ocean mode module, and the wave mode module s and the end date t e of the integration, the integration time step Ta of the atmospheric mode module, the integration time step To of the ocean mode module, the integration time step Tw of the wave mode module, the atmospheric coupling exchange step, the ocean coupling exchange step, the wave coupling exchange step, the output path, and the output frequency, and sends a start signal to the atmospheric mode module, the ocean mode module, and the wave mode module; After the atmospheric model module, ocean model module, and wave model module receive the start signals respectively, they perform start-up integration calculations, output of coupled variables, and output of corresponding forecast fields in parallel. The method is that the atmospheric model module executes step 9.2.1, the ocean model module executes step 9.2.2, and the wave integration module executes step 9.2.3 simultaneously, and the three model modules perform integration calculations in parallel: 9.2.1 The atmospheric model module receives the start signal from the coupling module, receives the assimilated atmospheric analysis field and assimilated atmospheric boundary field from the atmospheric data assimilation module of the data assimilation subsystem, uses the assimilated atmospheric analysis field as the initial value of the state equation system in the atmospheric model module, uses the assimilated atmospheric boundary field as the boundary condition of the state equation system in the atmospheric model module, starts the non-hydrostatic atmospheric model integration calculation, sends long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind to the coupling module during the integration process, and at the same time receives the sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the atmospheric model module with the received sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data and continues the integration, finally obtains the atmospheric forecast field, sends the atmospheric forecast field to the post-processing subsystem, sends the integration completion signal to the coupling module, and proceeds to 9.3; 9.2.2 The ocean model module receives the start signal from the coupling module, receives the ocean model grid file from the ocean preprocessing module of the preprocessing subsystem, receives the assimilated ocean analysis field and assimilated ocean boundary field from the ocean data assimilation module of the data assimilation subsystem, uses the assimilated ocean analysis field as the initial condition of the ocean state equation system in the ocean model module, uses the assimilated ocean boundary field as the boundary condition of the ocean state equation system in the ocean model module, starts the non-hydrostatic ocean model integration calculation, sends the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data to the coupling module during the integration process; at the same time receives the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the ocean model module with the received long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data and continues the integration, finally obtains the ocean forecast field, sends the ocean forecast field to the post-processing subsystem, sends the integration completion signal to the coupling module, and proceeds to 9.3; 9.2.3 After receiving the start signal of the wave model, the module receives the wave model grid file and the wave boundary field file from the wave warm start preprocessing module of the preprocessing subsystem, starts the numerical integration calculation of the wave state, and sends the significant wave height, wavelength, wave direction, and wave stress data to the coupling module during the integration process. At the same time, it receives the 10m wind, sea surface current, water level, water depth, and bottom roughness data sent by the coupling module, replaces the corresponding variables in the wave model module with the received 10m wind, sea surface current, water level, water depth, and bottom roughness data, and continues the integration. Finally, it obtains the wave forecast field and the wave warm start text file, sends the wave forecast field to the postprocessing subsystem, sends the integration completion signal to the coupling module, saves the wave warm start text file as the temporary file hotfile.txt, and proceeds to 9.3; 9.3 When the coupling module receives the integration completion signals sent by the atmosphere, ocean, and wave model modules, it ends the coupling mode calculation process and outputs the log text file log.txt to save the basic information during the system operation; Step 10: The postprocessing subsystem receives the atmosphere forecast field, ocean forecast field, and wave forecast field from the coupling mode calculation subsystem, performs interpolation or diagnostic quantity calculation on the atmosphere forecast field, ocean forecast field, and wave forecast field, and generates regional refined air-sea-wave coupled numerical forecast products. The method is as follows: 10.1 The regional atmosphere model output postprocessing module receives the atmosphere forecast field from the atmosphere model module of the coupling mode calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea level pressure, air temperature at 2m height, relative humidity at 2m height, visibility, U wind component at 10m height, and V wind component at 10m height in the atmosphere forecast field, interpolates the components onto the standard isobaric surfaces, saves them as the first part of the atmosphere forecast field NetCDF file, and extracts the accumulated precipitation and humidity at each vertical layer in the atmosphere forecast field, calculates the accumulated precipitation and radar reflectivity diagnostic analysis quantity at the grid points every 3 hours and saves them as the second part of the atmosphere forecast field NetCDF file. The above two parts of the atmosphere forecast field NetCDF files constitute the atmosphere analysis file; 10.2 The regional ocean model output postprocessing module receives the ocean forecast field from the ocean model module of the coupling mode calculation subsystem, extracts the ocean temperature, flow field, and salinity variables in the ocean forecast field, interpolates them onto the isobaths from 0m to 1000m depth, obtains the sea surface temperature, sea surface flow direction, sea surface flow velocity, ocean temperature, salinity, and flow field data at each layer and saves them as the first part of the ocean forecast field NetCDF file; uses the interpolated ocean temperature and salinity at each layer above to calculate the ocean density and sound speed at each layer, obtains the second part of the ocean forecast field NetCDF file. The above two parts of the ocean forecast field NetCDF files constitute the ocean analysis file; The post - processing module for the regional wave model output receives the wave forecast field from the wave model module of the coupled model calculation subsystem, extracts the significant wave height, wave direction, wave speed, and period in the wave forecast field, saves them as independent NetCDF files respectively, and obtains the wave analysis file; the atmospheric analysis file, the ocean analysis file, and the wave analysis file constitute the high - precision forecast products output by the regional refined air - sea - wave coupled numerical prediction system.

2. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that The data collection subsystem in the regional refined air - sea - wave coupled numerical prediction system constructed in the first step saves the regional atmospheric forecast field in the standard GRIB format; saves the regional ocean forecast field in the standard NetCDF format; saves the regional wave forecast field in the standard NetCDF format; archives the atmospheric observation files in the ASCII or PREPBUFR format, and archives the observation error files in the atmospheric observation files in the text document format; archives the ocean observation files in the NetCDF format, and archives the climatological statistical results of the ocean model in the ocean observation files in chronological order. The regional air - sea - wave grid generation module is connected to the data collection subsystem, the coupled model calculation subsystem, the data assimilation subsystem, the high - precision terrain processing module, the atmospheric pre - processing module, the ocean pre - processing module, and the wave warm - start pre - processing module; the regional air - sea - wave grid generation module reads the forecast area longitude and latitude from the configuration file sent by the data collection subsystem, receives the locally smoothed terrain data from the high - precision terrain processing module, performs vertical interpolation and land - sea marking on the atmospheric terrain data in the locally smoothed terrain data to generate the atmospheric model grid file; performs interpolation and land - sea marking on the ocean terrain data in the locally smoothed terrain data, marks and processes the positions of river channels to generate the ocean model grid file and the wave model grid file; sends the atmospheric model grid file to the atmospheric pre - processing module. Sends the ocean model grid file to the ocean pre - processing module. Sends the wave model grid file to the wave warm - start pre - processing module. The high - precision terrain processing module is connected to the data collection subsystem and the regional air - sea - wave grid generation module, obtains the high - precision terrain data from the data collection subsystem, classifies and archives and locally smooths the high - precision terrain data to obtain the locally smoothed terrain data suitable for the regional refined air - sea - wave coupled numerical prediction system, and sends the locally smoothed terrain data to the regional air - sea - wave grid generation module. The wave warm - start pre - processing module is connected to the data collection subsystem and the regional air - sea - wave grid generation module, obtains the wave model grid file from the regional air - sea - wave grid generation module, receives the regional wave forecast field from the data collection subsystem, interpolates the regional wave forecast field according to the wave model grid file to generate the wave model boundary field, and sends the wave model boundary field and the wave model grid file to the coupled model calculation subsystem. The atmospheric preprocessing module is connected to the data collection subsystem, the regional air-sea-wave grid generation module, and the data assimilation subsystem. It obtains the regional atmospheric forecast field from the data collection subsystem, obtains the atmospheric model grid file from the regional air-sea-wave grid generation module, decodes the regional atmospheric forecast field, interpolates the decoded atmospheric forecast field onto the atmospheric model grid file to obtain the atmospheric initial field and the atmospheric boundary field, and sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem; The ocean preprocessing module is connected to the data collection subsystem and the regional air-sea-wave grid generation module. It obtains the regional ocean forecast field, tidal data, and regional atmospheric forecast field from the data collection subsystem, receives the ocean model grid file from the regional air-sea-wave grid generation module, performs grid vertical interpolation preprocessing on the regional ocean forecast field and tidal data to obtain the ocean initial field and the ocean boundary field of the regional ocean model, and sends the ocean model grid file, the ocean initial field, and the ocean boundary field to the data assimilation subsystem; interpolates the regional atmospheric forecast field onto the ocean model grid file to obtain the atmospheric forcing field of the ocean model, and sends the atmospheric forcing field to the data assimilation subsystem; sends the ocean model grid file to the coupled model calculation subsystem; The data assimilation subsystem is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem. The atmospheric data assimilation module is connected to the data collection subsystem, the atmospheric preprocessing module of the preprocessing subsystem, and the coupled model calculation subsystem; The atmospheric data preprocessing sub-module is connected to the data collection subsystem and the atmospheric assimilation sub-module. It obtains the atmospheric observation file from the data collection subsystem, performs data preprocessing and quality control on the atmospheric observation file to generate an atmospheric observation temporary data file, and sends the atmospheric observation temporary data file to the atmospheric assimilation sub-module; The background error covariance sub-module is connected to the atmospheric assimilation sub-module, uses the background field information in the background error covariance sub-module to calculate the background error covariance, generates a background error covariance matrix, and sends the background error covariance matrix to the atmospheric assimilation sub-module; The atmospheric assimilation sub-module is connected to the data preprocessing sub-module, the background error covariance sub-module, and the preprocessing subsystem. It obtains the atmospheric initial field and the atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the background error covariance matrix from the background error covariance sub-module, obtains the atmospheric observation temporary data file from the atmospheric data preprocessing sub-module, performs three-dimensional variational assimilation on the atmospheric initial field and the atmospheric boundary field to generate the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem; The ocean data assimilation module is connected to the data collection subsystem, the preprocessing subsystem, and the coupled model calculation subsystem; The ocean data preprocessing sub-module is connected to the data collection subsystem and the four-dimensional variational sub-module. It receives the ocean observation file from the data collection subsystem, performs data preprocessing and quality control on the ocean observation file to generate an ocean observation temporary data file and sends it to the four-dimensional variational sub-module; The four-dimensional variational sub-module is connected to the ocean preprocessing module, the ocean data preprocessing sub-module, and the coupled model calculation subsystem of the preprocessing subsystem. It receives the ocean model grid file, the ocean initial field, the boundary field, and the atmospheric forcing field from the ocean preprocessing module, and receives the ocean observation temporary data file from the ocean data preprocessing sub-module. It performs variational assimilation calculations on the ocean observation temporary data file to obtain the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the ocean model module of the coupled model calculation subsystem; The coupling module is connected to the atmospheric model module, the ocean model module, and the wave model module. According to the forecast time read from the configuration file sent by the data collection subsystem, the atmospheric coupling exchange step, the ocean coupling exchange step, and the wave coupling exchange step are used to initialize the coupled model calculation subsystem and perform data exchange between the atmospheric model module, the ocean model module, and the wave model module; The coupling module receives the configuration file from the data collection subsystem and performs initial parameter configuration on the coupled model calculation subsystem; When the integration calculation of the atmospheric model module reaches the preset atmospheric coupling exchange step in the configuration file, the coupling module obtains atmospheric variables from the atmospheric model module. The atmospheric variables include long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data. When the integration calculation of the ocean model module reaches the preset ocean coupling exchange step in the configuration file, the coupling module obtains ocean variables from the ocean model module. The ocean variables include sea surface temperature, sea surface current, water level, water depth, and bottom roughness data. When the integration calculation of the wave model module reaches the preset wave coupling exchange step in the configuration file, the coupling module obtains wave variables from the wave model module. The wave variables include significant wave height, wavelength, wave direction, and wave stress data. The coupling module performs data exchange between the atmospheric model module, the ocean model module, and the wave model module respectively; The atmospheric model module is connected to the atmospheric data assimilation module, the post-processing subsystem, and the coupling module of the data assimilation subsystem. It receives the assimilated atmospheric analysis field and the assimilated atmospheric boundary field from the atmospheric data assimilation module of the data assimilation subsystem, and performs numerical integration calculations on the non-hydrostatic atmospheric numerical equations in the atmospheric model module according to the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to obtain atmospheric variables and the atmospheric forecast field. It sends the atmospheric forecast field to the post-processing subsystem, and transfers the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind data in the atmospheric variables to the coupling module. Through the coupling module, it transfers the long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, and precipitation data in the atmospheric variables to the ocean model module, and transfers the 10-meter wind data in the atmospheric variables to the wave model module. It receives the sea surface temperature in the ocean variables and the significant wave height, wavelength, wave direction, and wave stress data in the wave variables from the coupling module and updates the atmospheric variables; The ocean model module is connected to the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem, the ocean data assimilation module of the data assimilation subsystem, the post - processing subsystem, and the coupling module. It receives the ocean model grid file from the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem, and receives the assimilated ocean analysis field and the assimilated ocean boundary field from the ocean data assimilation module of the data assimilation subsystem. According to the assimilated ocean analysis field and the assimilated ocean boundary field, it performs numerical integration calculations on the non - hydrostatic ocean numerical equations in the ocean model module to obtain ocean variables and the ocean forecast field, sends the ocean forecast field to the post - processing subsystem, and transfers the sea surface temperature, sea surface current, water level, water depth, and bottom roughness data in the ocean variables to the coupling module. Through the coupling module, it transfers the sea surface temperature and sea surface current data to the atmospheric model module, and transfers the sea surface current, water level, water depth, and bottom roughness data to the wave model module; It receives the long - wave radiation, short - wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10 - meter wind, significant wave height, wave length, wave direction, and wave stress data in the atmospheric variables through the coupling module to update the ocean variables; The wave model module is connected to the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem, the wave heat - start pre - processing module of the pre - processing subsystem, the coupling module, and the post - processing subsystem. It receives the wave model grid file from the regional ocean - atmosphere - wave grid generation module of the pre - processing subsystem and receives the wave boundary field from the wave heat - start pre - processing module. According to the wave boundary field and the static initial field with a default initial value of zero or the hot - start file hotfile.txt generated by the previous forecast, it performs numerical integration calculations to obtain wave variables and the wave forecast field, sends the wave forecast field to the post - processing subsystem, and transfers the significant wave height, wave length, wave direction, and wave stress data in the wave variables to the coupling module. Through the coupling module, it transfers the significant wave height, wave length, wave direction, and wave stress data to the atmospheric model module and transfers the significant wave height, wave length, wave direction, and wave stress data to the ocean model module; it receives the 10 - meter wind, sea surface current, water level, water depth, and bottom roughness data through the coupling module to update the wave variables; The regional atmospheric model output post-processing module receives the atmospheric forecast field from the atmospheric model module of the coupled model calculation subsystem, extracts the geopotential height, temperature, relative humidity, horizontal wind field, vertical velocity, cloud water content, cloud ice content, sea level pressure, air temperature at 2 m height, relative humidity at 2 m height, visibility, U wind component at 10 m height, and V wind component at 10 m height in the atmospheric forecast field, interpolates the components onto standard isobaric surfaces respectively, saves them as the first part of the atmospheric forecast field NetCDF file, and extracts the accumulated precipitation and humidity at each vertical layer in the atmospheric forecast field, calculates the accumulated precipitation at 3-hourly grid points and the radar reflectivity diagnostic analysis quantity and saves them as the second part of the atmospheric forecast field NetCDF file. The above two parts of the atmospheric forecast field NetCDF file constitute the atmospheric analysis file; the regional ocean model output post-processing module receives the ocean forecast field from the ocean model module of the coupled model calculation subsystem, extracts the ocean temperature, flow field, and salinity variables in the ocean forecast field, interpolates them onto the isobath surfaces from 0 m to 1000 m depth, obtains the sea surface temperature, sea surface current direction, sea surface current velocity, ocean temperature, salinity, and flow field data at each layer and saves them as the first part of the ocean forecast field NetCDF file. Using the ocean temperature and salinity interpolated above, calculates the ocean density and sound speed at each layer to obtain the second part of the ocean forecast field NetCDF file. The above two parts of the ocean forecast field NetCDF file constitute the ocean analysis file; the regional wave model output post-processing module receives the wave forecast field from the wave model module of the coupled model calculation subsystem, extracts the significant wave height, wave direction, and period variables in the wave forecast field, and saves them as independent NetCDF files respectively to obtain the wave analysis file.

3. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 2, characterized in that The dedicated line database is installed on the high-performance computing center server; the coupling module uses the coupler C-Coupler3.

0.

4. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that The method by which the data collection subsystem in the second step obtains the configuration file, atmospheric, oceanic, and wave-related data from the dedicated line database and sends the atmospheric, oceanic, and wave-related data to the preprocessing subsystem and the data assimilation subsystem respectively, and sends the configuration file to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem is as follows: 2.1 Data collection subsystem The data collection subsystem obtains the configuration file from the dedicated line database, reads the longitude and latitude of the forecast area therein, and obtains the global atmospheric forecast field data starting from the current day with a forecast duration of 10 days from the dedicated line database. The global atmospheric forecast field data is output every 3 hours, and each output is saved as a separate data file. The global atmospheric forecast field data is a collection of a group of data files; clips the global atmospheric forecast field data according to the longitude and latitude of the forecast area, combines the air temperature, pressure, humidity, and wind field variable files at the same time step into a single-time-step regional atmospheric forecast field, saves the regional atmospheric forecast field in the standard GRIB format in sequence according to the date and data source, and sends the regional atmospheric forecast field to the atmospheric preprocessing module of the preprocessing subsystem. 2.2 The data collection subsystem obtains the global ocean forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of one day from the dedicated line database; the global ocean forecast field data outputs sea temperature, salinity, sea surface height, and flow field variables each time and saves them as sea temperature, salinity, sea surface height, and flow field variable data files respectively. The global ocean forecast field data is a collection of a set of data files; it is trimmed according to the forecast area range, and the sea temperature, salinity, sea surface height, and flow field variable files at the same time step are merged into a regional ocean forecast field, and are saved as a set of standard NetCDF formats in sequence according to the date and data source, and the regional ocean forecast field is sent to the ocean preprocessing module of the preprocessing subsystem; 2.3 The data collection subsystem obtains the global wave forecast field data with a starting forecast time of the current day, a forecast duration of 10 days, and an output frequency of one day from the dedicated line database; the global wave forecast field data outputs significant wave height, wave period, wave length, and wave direction variables each time and saves them as significant wave height, wave period, wave length, and wave direction variable data files respectively. The global wave forecast field data is a collection of a set of data files; it is trimmed according to the forecast area range, and the significant wave height, wave period, wave length, and wave direction variable files are merged into a regional wave forecast field, and are saved in the standard NetCDF format according to the date and data source, and the regional wave forecast field is sent to the wave warm start preprocessing module of the preprocessing subsystem; 2.4 The data collection subsystem obtains high-precision terrain data and tidal data from the dedicated line database; the high-precision terrain data includes ETOP2 dataset and GEO dataset, which include water depth, rivers, lakes, forests, grasslands, ice and snow surfaces, deserts, land surfaces, and corresponding soil temperature and soil moisture information; the tidal data comes from the TPXO tidal data provided by Oregon State University, including four semi-diurnal tides: M2, S2, N2, K2, four diurnal tides K1, O1, P1, Q1, and Mf, Mm, M4, MS4, MN4 partial tide information, with a resolution of 0.25°×0.25°, and the high-precision terrain data and tidal data are sent to the ocean preprocessing module of the preprocessing subsystem; 2.5 The data collection subsystem obtains real-time atmospheric observation data and real-time ocean observation data from the dedicated line database, processes the real-time atmospheric observation data into binary files, renames them according to the data source and date, and archives them in the NetCDF format to obtain atmospheric observation files; processes the real-time ocean observation data into binary files, renames them according to the data source, and archives them in the NetCDF format to obtain ocean observation files; sends the atmospheric observation files to the atmospheric data assimilation module of the data assimilation subsystem, and sends the ocean observation files to the ocean data assimilation module of the data assimilation subsystem; 2.6 The data collection subsystem obtains the configuration file from the dedicated line database and sends the configuration file to the preprocessing subsystem, the data assimilation subsystem, and the coupled model calculation subsystem.

5. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that 3.5 The method by which the regional air-sea-wave grid generation module in step 3.5 receives the smoothed atmospheric and ocean terrain data from the high-precision terrain processing module and processes the initial grids of the atmospheric model, ocean model, and wave model according to the types of terrain, the land-sea boundary, and the types of river channels in the terrain data is as follows: 3.5.1 Assign values to the land-sea masks in the initial grids of the atmospheric model, ocean model, and wave model, mark ocean points as 1 and land points as 0, and at the same time perform waterway marking processing at the river estuaries, with the waterway positions being 1 and the coastal land being 0; 3.5.2 Interpolate the elevation of the smoothed atmospheric terrain data, different land types, including rivers, lakes, forests, grasslands, ice and snow surfaces, deserts, and land surfaces, and the corresponding soil temperature and soil moisture information to the initial grid of the atmospheric model to generate an atmospheric model grid file; 3.5.3 Interpolate the water depth, sea-land boundary, and river channel information of the smoothed ocean terrain data to the initial grids of the ocean model and wave model respectively to generate an ocean model grid file and a wave model grid file.

6. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, wherein The method by which the atmospheric preprocessing module of the preprocessing subsystem described in step 4 receives the configuration file and the regional atmospheric forecast field from the data collection subsystem, receives the atmospheric model grid file from the regional air-sea-wave grid generation module, decodes and interpolates the preprocessing of the regional atmospheric forecast field, and obtains the atmospheric initial field and the atmospheric boundary field and sends them to the data assimilation subsystem is as follows: 4.1 The atmospheric preprocessing module receives the configuration file, the regional atmospheric forecast field from the data collection subsystem, and the atmospheric model grid file from the regional air-sea-wave grid generation module; 4.2 The atmospheric preprocessing module links the regional atmospheric forecast field to the working directory and renames all the data in the regional atmospheric forecast field in alphabetical order according to the file type and the English alphabet in the form of a soft link; 4.3 The atmospheric preprocessing module makes a variable table based on the variable types and names of wind speed, wind direction, temperature, and geopotential height in the global atmospheric forecast field data, and decodes and interpolates the global atmospheric forecast field through the WRF preprocessing system; the variable table records the GRIB encoding, variable name, variable unit, and variable layer information of the global atmospheric forecast field data; 4.4 The atmospheric preprocessing module reads the longitude and latitude of the forecast area from the configuration file, extracts the required wind speed, wind direction, temperature, and geopotential height variables from the GRIB encoding of the global atmospheric forecast field according to the variable table, and writes them into an intermediate format file to form intermediate format meteorological data; 4.5 The atmospheric preprocessing module uses the interpolation algorithm in the WRF model to horizontally interpolate the intermediate format meteorological data onto the atmospheric model grid file to produce the atmospheric initial field and the atmospheric boundary field; 4.6 The atmospheric preprocessing module sends the atmospheric initial field and the atmospheric boundary field to the data assimilation subsystem.

7. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that The method by which the ocean preprocessing module of the preprocessing subsystem described in the fifth step receives the configuration file, regional ocean forecast field, tidal data, and regional atmospheric forecast field from the data collection subsystem, receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module, performs interpolation preprocessing on the regional ocean forecast field, tidal data, and regional atmospheric forecast field to obtain the atmospheric forcing field, ocean initial field, and ocean boundary field, and sends the ocean model grid file, atmospheric forcing field, ocean initial field, and ocean boundary field to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem is as follows: 5.1 The ocean preprocessing module receives the configuration file, regional ocean forecast field, tidal data, and regional atmospheric forecast field from the data collection subsystem, and receives the ocean model grid file from the regional ocean-atmosphere-wave grid generation module; 5.2 The ocean preprocessing module screens the regional ocean forecast field according to the grid point positions in the grid file, determines the default data at the corresponding grid point positions as outliers, marks them as invalid values, i.e., NAN, obtains the screened regional ocean forecast field, and saves it as a temporary file in NetCDF format; 5.3 The ocean preprocessing module interpolates the screened regional ocean forecast field according to the ocean model grid file to generate the ocean initial field and ocean boundary field; 5.4 The ocean preprocessing module interpolates the regional atmospheric forecast field to generate the atmospheric forcing field; 5.5 The ocean preprocessing module sends the atmospheric forcing field, ocean model initial field, ocean boundary field, and ocean model grid file to the data assimilation subsystem, and sends the ocean model grid file to the coupled model calculation subsystem.

8. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that The method by which the wave warm start preprocessing module of the preprocessing subsystem described in the sixth step obtains the regional wave forecast field from the data collection subsystem, receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module, interpolates the regional wave forecast field to obtain the wave boundary field, and sends the wave boundary field to the coupled model calculation subsystem is as follows: 6.1 The wave warm start preprocessing module receives the regional wave forecast field from the data collection subsystem and receives the wave model grid file from the regional ocean-atmosphere-wave grid generation module; 6.2 The wave preprocessing module interpolates the regional wave forecast field according to the wave model grid file to generate the wave boundary field; 6.3 The wave warm start preprocessing module sends the wave boundary field and the wave model grid file to the coupled model calculation subsystem.

9. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that The method by which the atmospheric data assimilation module of the data assimilation subsystem described in the seventh step receives the atmospheric initial field and atmospheric boundary field from the atmospheric preprocessing module of the preprocessing subsystem, receives the atmospheric observation file and configuration file from the data collection subsystem, preprocesses the atmospheric observation file to obtain the atmospheric observation temporary data file, calculates the background error covariance of the simulation area to obtain the background error covariance matrix, and finally performs regional atmospheric three-dimensional variational assimilation to obtain the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the coupled model calculation subsystem is as follows: 7.1 The atmospheric data preprocessing sub-module receives the atmospheric observation file and the configuration file from the data collection subsystem, converts various observation data files in the atmospheric observation file into ASCII or PREPBUFR formats recognizable by the atmospheric assimilation sub-module; saves the observation error file in the atmospheric observation file as a text document and sends it to the atmospheric assimilation sub-module; 7.2 The atmospheric data preprocessing sub-module, according to the assimilation area and assimilation time window in the configuration file, eliminates the observations in the atmospheric observation file that are not within the assimilation area and assimilation time window, reorders the variable data in the atmospheric observation file according to time, calculates the pressure or height from the observed variables using the hydrostatic hypothesis, checks the vertical consistency and super-adiabatic conditions of the multi-level observed variables, sets the observation error according to the observation error file in the atmospheric observation data, and obtains the atmospheric observation temporary data file and sends it to the atmospheric assimilation sub-module; 7.3 The background error covariance sub-module randomly generates a set of background field perturbations, standardizes each perturbation to ensure its mean is zero, generates standardized perturbations, and then calculates the background error covariance matrix using the standardized perturbations. The formula is: where the number of perturbations N is an integer between 20 and 100, is the randomly generated standardized perturbation, T represents the matrix transpose operation, and the background error covariance matrix B is sent to the atmospheric assimilation sub-module; 7.4 The atmospheric assimilation sub-module receives the atmospheric initial field and the atmospheric boundary field from the preprocessing subsystem, receives the atmospheric observation temporary data file from the atmospheric data preprocessing sub-module, receives the background error covariance matrix from the background error covariance sub-module, performs three-dimensional variational assimilation, obtains the assimilated atmospheric analysis field and the assimilated atmospheric boundary field, and sends the assimilated atmospheric analysis field and the assimilated atmospheric boundary field to the atmospheric model module of the coupled model calculation subsystem.

10. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that 8.2 The method by which the four-dimensional variational sub-module described in step 8 receives the atmospheric forcing field, the ocean initial field, the ocean boundary field, and the ocean model grid file from the preprocessing subsystem, receives the ocean observation temporary data file from the ocean data preprocessing sub-module, performs incremental four-dimensional variational calculation on the ocean observation temporary data file, generates the assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem is as follows: 8.2.1 The four-dimensional variational sub-module performs grid sparsification processing on the ocean model grid file to obtain a low-resolution ocean grid file; 8.2.2 The four-dimensional variational sub-module interpolates the atmospheric forcing field, the ocean initial field, and the ocean boundary field to the low-resolution ocean grid file to obtain the low-resolution atmospheric forcing field, ocean initial field, and ocean boundary field; 8.2.3 The four-dimensional variational sub-module extracts the observation positions and times from the ocean observation file, and performs four-dimensional variational assimilation calculation at the observation time for the low-resolution atmospheric forcing field, ocean initial field, ocean boundary field, and ocean observation temporary data file at low resolution to obtain the low-resolution four-dimensional variational analysis increment, and stores the low-resolution four-dimensional variational analysis increment as a NetCDF format temporary file; 8.2.4 The four-dimensional variational sub-module uses the low-resolution four-dimensional variational analysis increment and the ocean model grid file, and through linear interpolation, obtains the high-resolution four-dimensional variational analysis increment and stores it as a NetCDF format temporary file; 8.2.5 The four-dimensional variational sub-module adds the high-resolution four-dimensional variational analysis increment to the high-resolution background field in the ocean observation file to obtain the high-resolution assimilated ocean analysis field and the assimilated ocean boundary field, and sends the assimilated ocean analysis field and the assimilated ocean boundary field to the coupled model calculation subsystem; the high-resolution background field is the climatological statistical result of the ocean model and comes from the ocean observation file sent by the data collection subsystem.

11. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, wherein 9.2.1 The method for the atmospheric model module to receive the start signal from the coupling module, receive the assimilated atmospheric analysis field and the assimilated atmospheric boundary field from the atmospheric data assimilation module of the data assimilation subsystem, use the assimilated atmospheric analysis field as the initial value of the state equation system in the atmospheric model module, use the assimilated atmospheric boundary field as the boundary condition of the state equation system in the atmospheric model module, and start the non-hydrostatic atmospheric model integration calculation to obtain the atmospheric forecast field is as follows: 9.2.1.1 Initialize the atmospheric model module. At 00:00 on the start date t of the integration, the initial integration step n = 0. When Ta is the integration step length of the atmospheric model module, n(Ta) is the current total integration time 0 of the atmospheric model module at the initial moment; s When the time is 00:00 at the start date t of the integration, the initial integration step n = 0. When Ta is the integration step length of the atmospheric model module, n(Ta) is the current total integration time 0 of the atmospheric model module at the initial moment; 9.2.1.2 Let n = n + 1, and the atmospheric model module performs non-hydrostatic atmospheric model integration calculations to obtain an atmospheric forecast field, which is a NetCDF format file; as the number of integration steps n increases, the time of the integration result of the atmospheric model module continuously moves backward from 00:00 at time t s by Ta for each integration step, and the integration results in the atmospheric forecast field continuously increase; 9.2.1.3 Let n(Ta) be the total integration time of the atmospheric model module. If n(Ta) = t e -t s , then the integration of the atmospheric model module ends. The atmospheric forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and then it ends. If n(Ta) < t e -t s , go to step 9.2.1.4; 9.2.1.4 If n(Ta) < t e -t s and n(Ta) is an integer multiple of the atmospheric coupling exchange step length Tca, the atmospheric model module will send long-wave radiation, short-wave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, and 10-meter wind to the coupling module; at the same time, the atmospheric model module receives the sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the atmospheric model module with the received sea surface temperature, significant wave height, wavelength, wave direction, and wave stress data, and proceeds to step 9.2.1.2 for the non-hydrostatic atmospheric model integration of the n+1 step; if n(Ta) < t e -t s and n(Ta) is not an integer multiple of Tca, no data output and exchange are performed, and the process directly proceeds to step 9.2.1.2 for the integration calculation of the n+1 time step.

12. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, wherein 9.2.2 The method for the ocean model module to receive the start signal from the coupling module, receive the ocean model grid file from the ocean preprocessing module of the preprocessing subsystem, receive the assimilated ocean analysis field and the assimilated ocean boundary field from the ocean data assimilation module of the data assimilation subsystem, use the assimilated ocean analysis field as the initial condition of the ocean state equation system in the ocean model module, use the assimilated ocean boundary field as the boundary condition of the ocean state equation system in the ocean model module, and start the non-hydrostatic ocean model integration calculation to obtain the ocean forecast field is as follows: 9.2.2.1 Initialize the ocean model module, and the starting date of integration is t s At 00:00, the initial integration step n = 0. When To is the integration step size of the ocean model module, n(To) is the current total integration time of the ocean model module at the initial moment, which is 0; 9.2.2.2 Let n = n + 1, and the ocean model module performs non-hydrostatic ocean model integration calculations to obtain an ocean forecast field, which is a NetCDF format file; as the number of integration steps n increases, the time of the integration result of the ocean model module advances continuously from 00:00 at time t s by the time of each integration step To, and the integration results in the ocean forecast field continue to increase; 9.2.2.3 Let n(To) = the total time for the ocean model module integration calculation. If n(To) = t e -t s , then the integration of the ocean model module ends. The ocean forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module, and it ends; if n(To) < t e -t s , go to step 9.2.2.4; 9.2.2.4 If n(To) < t e -t s and n(To) is an integer multiple of the ocean coupling exchange step length Tco, the ocean model module sends sea surface temperature, sea surface current, water level, water depth, and bottom roughness data to the coupling module; meanwhile, the ocean model module receives longwave radiation, shortwave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data sent by the coupling module, replaces the corresponding variables in the ocean model module with the received longwave radiation, shortwave radiation, latent heat, sensible heat, wind stress, evaporation, precipitation, 10-meter wind, significant wave height, wavelength, wave direction, and wave stress data, and proceeds to step 9.2.2.2 for the non-hydrostatic ocean model integration of the n+1 step; if n(To) < t e -t s and n(To) is not an integer multiple of Tco, no data output and exchange are performed, and the process directly proceeds to step 9.2.2.2 for the integration calculation of the n+1 time step.

13. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that 9.2.3 The method for the wave model to receive the start signal, receive the wave model grid file and the wave boundary field file from the wave warm start preprocessing module of the preprocessing subsystem, and start the wave state numerical integration calculation to obtain the wave forecast field and the wave warm start text file is as follows: 9.2.3.1 Initialize the ocean wave model module, the start date of integration is t s at 00:00 UTC. The initial integration step n = 0. When Tw is the integration step of the ocean wave model module, n(Tw) is the current total integration time of the ocean wave model module at the initial moment, which is 0. Determine whether there is an ocean wave initial field file hotfile.txt in the current working directory. If hotfile.txt exists, the ocean wave model module reads hotfile.txt as the initial condition of the ocean wave spectrum equation in the ocean wave model module at n = 0. hotfile.txt is a temporary file generated after the previous integration. If the previous integration does not exist, there is no hotfile.txt, and go to step 9.2.3.2; if hotfile.txt does not exist, all the initial values of the wave height, wavelength, wave period, wave direction parameter, and wave energy spectrum in the ocean wave spectrum equation of the ocean wave model module are initialized to 0, and go to step 9.2.3.2; 9.2.3.2 Let \(n = n + 1\), and the wave model module performs numerical integration calculation of the wave state to obtain a wave forecast field, which is a NetCDF format file; as the number of integration steps \(n\) increases, the time of the integration result of the wave model module continuously moves backward from 00:00 at time \(t\) s by the time \(T_w\) of each integration step, and the integration results in the wave forecast field continuously increase; 9.2.3.3 Let n(Tw) be the total integration time of the ocean wave model module. If n(Tw) = t e -t s , then the integration of the ocean wave model module ends. The ocean wave forecast field is sent to the post-processing subsystem, and the integration completion signal is sent to the coupling module. At the same time, an ocean wave warm start text file is generated, saved as the hotfile.txt temporary file, and temporarily stored in the current working directory as the initial state field in the ocean wave model module when the next coupled forecast is started. The content of the hotfile.txt file is the ocean wave state information, including wave height, wavelength, wave period, wave direction parameters, wave energy spectrum; boundary information, including the sea surface wind field; grid information, including the spatial resolution and layout information of the grid, and it ends. If n(Tw) < t e -t s , go to step 9.2.3.4; 9.2.3.4 If n(Tw) < t e -t s and n(Tw) is an integer multiple of the ocean wave coupling exchange step length Tcw, the ocean wave model module sends the significant wave height, wavelength, wave direction, and wave stress data to the coupling module; meanwhile, the ocean model module receives the 10-meter wind, sea surface current, water level, water depth, and bottom roughness data sent by the coupling module, replaces the corresponding variables in the ocean wave model module with the received 10-meter wind, sea surface current, water level, water depth, and bottom roughness data, and proceeds to step 9.2.3.2 for the (n + 1)-step ocean wave model integration; if n(Tw) < t e -t s and n(Tw) is not an integer multiple of Tcw, no data output and exchange are performed, and the process directly proceeds to step 9.2.3.2 for the integration calculation at the (n + 1)-th time step.

14. The regional refined air-sea-wave coupled numerical prediction method based on the non-hydrostatic model according to claim 1, characterized in that 9.3 The content of the work log text file log.txt includes the number of parallel cores of the three models, the initial state of the atmospheric model module, the initial state of the ocean model module, the initial state of the wave model module, the integration time, the name of the coupling variable, and the integration end signal.

Images