Numerical weather mode multi-source data assimilation method based on GNSS (Global Navigation Satellite System) water vapor chromatography result

Abstract

The invention discloses a numerical weather mode multi-source data assimilation method based on a GNSS (Global Navigation Satellite System) water vapor chromatography result, which relates to the field of weather forecast and numerical weather modes and comprises the following steps: arranging a numerical weather mode in a target area; collecting multi-source data of the target area, wherein the multi-source data comprises GNSS water vapor chromatography data, satellite data, radar data and observation data of a ground meteorological station, a sonde and an aircraft; performing data assimilation on each item of data in the multi-source data, and converting the multi-source data into a unified data format; and permutation and combination are carried out on all parameterization schemes of cloud microphysics, cumulus convection, a boundary layer, a land surface and a long and short wave radiation process in the numerical weather mode, the actual operation effect of each scheme combination is evaluated, and an optimal parameterization scheme is obtained through screening. According to the multi-source data assimilation method, the initial field of the numerical weather mode can be improved, so that the weather forecast precision is improved.

Description

Technical Field

[0001] This invention relates to the field of meteorological forecasting and numerical weather modeling, and more specifically to a method for assimilating multi-source data of numerical weather models based on GNSS water vapor tomography results. Background Technology

[0002] Currently, numerical weather prediction (NWP) models are the core tool for weather forecasting. They quantitatively predict future weather changes by solving atmospheric dynamics and thermodynamic equations (such as continuity equations, kinematic equations, and thermodynamic equations) and combining them with initial fields and boundary conditions. This technology is widely used in disaster early warning (such as typhoons and rainstorms), aviation and maritime navigation, agricultural meteorological services, and climate change research, and is an indispensable objective forecasting method in modern meteorological operations. The accuracy of NWP models largely depends on the initial field. Improving the initial field through numerical assimilation methods is currently the main means of enhancing the accuracy of NWP models. Current data assimilation for NWP models primarily relies on data sources such as satellite remote sensing data, radiosondes, ground meteorological stations, and aircraft observations.

[0003] However, assimilating only single-source data has significant drawbacks: uneven spatiotemporal coverage (e.g., radiosonde data at 12-hour intervals), error accumulation (systematic bias in satellite cloud water vapor retrieval), and poor adaptability to complex weather conditions. Therefore, assimilation of multi-source data is the main future development trend. In the initial field of numerical weather models, the water vapor field has a significant impact on model accuracy. Currently, the assimilation data mainly used to improve the water vapor field include satellite, radiosonde, and radar data, but these have drawbacks such as large cloud area errors, low vertical resolution (satellite), low spatiotemporal resolution (radiosonde), limited vertical coverage, and dependence on precipitation signals (radar).

[0004] Therefore, how to assimilate multi-source data to improve the initial field of numerical weather models and enhance forecast accuracy is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method for multi-source data assimilation of numerical weather models based on GNSS water vapor tomography results. By assimilating multi-source data, including GNSS water vapor tomography results and satellite data, into the numerical weather model, the initial field of the numerical weather model is improved and the forecast accuracy is enhanced.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] This invention discloses a method for assimilating multi-source data from a numerical weather model based on GNSS water vapor tomography results, comprising:

[0008] Deploy numerical weather models within the target area;

[0009] Collect multi-source data of the target area, including GNSS water vapor tomography data, satellite data, radar data, and observation data from ground meteorological stations, radiosondes, and aircraft;

[0010] Data assimilation is performed on each data item in the multi-source data to convert the multi-source data into a unified data format;

[0011] The parameterization schemes for cloud microphysics, cumulus convection, boundary layer, land surface, and longwave and shortwave radiation processes in the numerical weather model are arranged and combined, and the actual operating effect of each combination of schemes is evaluated to select the optimal parameterization scheme.

[0012] Furthermore, the deployment of numerical weather models within the target area includes: environmental preparation, dependency library installation, numerical weather model installation, data preprocessing, and runtime configuration.

[0013] The numerical weather model selected is the WRF mode.

[0014] Furthermore, the data preprocessing includes:

[0015] Converting static geographic data into a WRF-readable binary format involves: modifying the configuration file of the geogrid program in the WRF preprocessing system to specify the data path and parameters; and generating a WRF-usable grid file based on the modified configuration file using the geogrid program in the WRF preprocessing system.

[0016] Converting meteorological data into a WRF-readable binary format involves: linking the configuration file corresponding to the meteorological data and running the script file; linking the original meteorological data file; running the ungrib program to generate intermediate files; and then running the metgrid program to generate the final binary file.

[0017] Furthermore, the static geographic data includes: terrain elevation, land use type, and soil type data.

[0018] Furthermore, the step of assimilating each data item in the multi-source data to convert the multi-source data into a unified data format specifically includes:

[0019] The GNSS water vapor tomography data is assimilated using an indirect method, specifically by converting the wet refractive index obtained from the tomography into relative humidity, and then using WRFDA to assimilate the GNSS water vapor tomography data in conjunction with existing temperature and air pressure data.

[0020] The satellite data was assimilated using CRTM or RTTOV radiative transfer models;

[0021] The radar data is directly assimilated via WRFDA.

[0022] Furthermore, the specific formula for converting the wet refractive index into relative humidity is as follows:

[0023]

[0024] Among them, P w WR represents the water vapor pressure at the lattice point, T represents the wet refractive index, k1 = 77.6890 K / hpa, k2 = 71.2952 K / hpa;

[0025]

[0026] Among them, P s C represents the saturated water vapor pressure. i Let n represent T. i-1 express;

[0027] RH=P w / P s ;

[0028] Here, RH represents relative humidity.

[0029] Furthermore, the parameterization schemes for cloud microphysics include Ferrier, Lin, WSM3, WSM5, WSM6, and Thompson;

[0030] The parameterization schemes for cumulus convection include KF, BMJ, and GD;

[0031] The parameterization schemes for the boundary layer include YSU and MYNN;

[0032] The parameterization scheme for the land surface includes the Noah land surface scheme;

[0033] The parameterization schemes for long-wave and short-wave radiation include RRTM and the Dudhia physical parameterization scheme.

[0034] Furthermore, the actual operational effectiveness of each scheme combination is evaluated using the mean absolute error (MAE), bias (BIAS), and relative error (RE), as shown in the formula:

[0035]

[0036] BIAS=M i -O i ;

[0037]

[0038] Among them, M i This refers to actual precipitation at the monitoring station; O iThe model calculates station precipitation, where i is the station number and N is the total number of stations.

[0039] Furthermore, it also includes using the MPI distributed computing framework to enable parallel computation of numerical weather model models.

[0040] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a multi-source data assimilation method for numerical weather models based on GNSS water vapor tomography results. This method can improve the initial field of numerical weather models, thereby improving the accuracy of weather forecasts. It has significant effects in areas such as accurate forecasting of meteorological disasters, optimization of aviation and maritime navigation, support for agricultural meteorological services, and verification and improvement of climate models. Specifically:

[0041] Accurate forecasting of meteorological disasters: In extreme weather events such as typhoons and rainstorms, by assimilating GNSS water vapor tomography data and multi-source observation data, the numerical weather model's ability to characterize water vapor fields is improved, the path, intensity, and precipitation forecast of severe weather are optimized, and a reliable basis is provided for disaster prevention and mitigation.

[0042] Aviation and maritime navigation optimization: Utilizing high spatiotemporal resolution water vapor assimilation results, refined weather forecast products are generated to support aviation meteorological services (such as turbulence and icing warnings) and maritime route planning, thereby improving transportation safety and efficiency.

[0043] Agricultural meteorological service support: Through assimilated high-precision humidity and precipitation forecast data, it guides agricultural irrigation, pest and disease control, and crop growth cycle management, thus contributing to the development of smart agriculture.

[0044] Climate model validation and improvement: Applying multi-source data assimilation results to long-term climate simulations to validate the reliability of regional climate models and provide data support for climate change research. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] This invention discloses a method for assimilating multi-source data from a numerical weather model based on GNSS water vapor tomography results, such as... Figure 1 As shown, it includes:

[0049] Deploy numerical weather models within the target area;

[0050] Collect multi-source data of the target area, including GNSS water vapor tomography data, satellite data, radar data, and observation data from ground meteorological stations, radiosondes, and aircraft;

[0051] Data assimilation is performed on each data item in the multi-source data to convert the multi-source data into a unified data format;

[0052] We arranged and combined various parameterization schemes for cloud microphysics, cumulus convection, boundary layer, land surface, and longwave and shortwave radiation processes in numerical weather models, evaluated the actual operational effects of each scheme combination, and selected the optimal parameterization scheme.

[0053] In one specific embodiment, deploying a numerical weather model within a target area includes: environmental preparation, dependency library installation, numerical weather model installation, data preprocessing, and runtime configuration.

[0054] Select WRF mode for numerical weather patterns.

[0055] Specifically, the WRF model, as a mesoscale numerical weather prediction model, mainly consists of four parts: the preprocessing system WPS, the data assimilation system WRFDA, the main module ARW / NMM, and the postprocessing system. Through its comprehensive preprocessing and assimilation systems, it can assimilate multi-source data, including meteorological data, geographic data, satellite data, and radar data.

[0056] In one specific embodiment, data preprocessing includes:

[0057] Converting static geographic data into a WRF-readable binary format involves: modifying the configuration file of the geogrid program in the WRF preprocessing system to specify the data path and parameters; and generating a WRF-usable grid file based on the modified configuration file using the geogrid program in the WRF preprocessing system.

[0058] Converting meteorological data into a WRF-readable binary format involves: linking the configuration file corresponding to the meteorological data and running the script file; linking the original meteorological data file; running the ungrib program to generate intermediate files; and then running the metgrid program to generate the final binary file.

[0059] Specifically, in the data preprocessing, static geographic data such as terrain height, land use type, and soil type are converted into a WRF-readable binary format (such as NC format). The configuration file WPS / geogrid / GEOGRID.TBL is modified to specify the data path and parameters. The geogrid program in WPS is then used to generate a WRF-usable geo_em.d0X.nc file.

[0060] When processing meteorological data, link the Vtable corresponding to the meteorological data, run link_grib.csh to link the meteorological data GRIB file, run ungrib.exe to generate intermediate files, and run metgrid.exe to generate the met_em.d0X.YYYY-MM-DD_HH.nc file. Through the above steps, the preprocessing work of WRF mode is completed.

[0061] In one specific embodiment, static geographic data includes: terrain elevation, land use type, and soil type data.

[0062] In one specific embodiment, data assimilation is performed on each data item in the multi-source data to convert the multi-source data into a unified data format, specifically including:

[0063] The indirect method was used to assimilate GNSS water vapor tomography data. Specifically, the wet refractive index obtained from the tomography was converted into relative humidity, and GNSS water vapor tomography data was assimilated using WRFDA in combination with existing temperature and air pressure data.

[0064] Assimilate satellite data using CRTM or RTTOV radiative transfer models;

[0065] Radar data is directly assimilated via WRFDA.

[0066] Specifically, satellite radiation data is assimilated using the CRTM or RTTOV radiative transfer model. Satellite radiation data is typically stored in BUFR format. RTTOV or CRTM is used to convert the model background field (temperature, humidity, air pressure) into simulated brightness temperature data. Download the CRTM or RTTOV coefficient file (e.g., crtm_coeffs_2.3.0) and link it to the WRFDA directory. Configure the settings in namelist.input as follows:

[0067] ob_format = 2, ! BUFR format;

[0068] rtm_option=2,! Use CRTM;

[0069] Radar reflectivity and velocity can be directly assimilated using WRFDA, converting radar base data (such as NEXRA D) into a WRFDA-recognizable ob.radar file. This requires converting polar coordinates to geographic coordinates and enabling radar observation variables in namelist.input, specifically:

[0070] use_radar_rv = true;

[0071] use_radar_rf = true;

[0072] All data is assimilated in either LITTLE_R (ASCII text) or PREPBUFR (binary) format.

[0073] In a specific embodiment, the formula for converting wet refractive index to relative humidity is as follows:

[0074]

[0075] Among them, P w WR represents the water vapor pressure at the lattice point, T represents the wet refractive index, k1 = 77.6890 K / hpa, k2 = 71.2952 K / hpa;

[0076]

[0077] Among them, P s C represents the saturated water vapor pressure. i denoted by the empirical coefficient, given by Wexler in 1976, and n representing the upper limit exponent of the summation, which determines the highest degree of the polynomial terms in the formula (in the formula, n is generally taken as 4 to 8; when n = 4, C0 = -0.60436117 × 10⁻⁶). 4 C1 = 0.1893292601 × 10 2 C2 = -0.28244925 × 10 -1 C3 = 0.17250331 × 10 -4 C4 = 0.2858487 × 10), C n+1 Represents the empirical coefficient of the logarithmic term, which is a specific constant multiplied by the natural logarithm term lnT; i represents the summation index, which varies from 0 to n during the summation process; T i-1 Represents T raised to the power of i-1;

[0078] RH = Pw / P s ;

[0079] Here, RH represents relative humidity.

[0080] In one specific embodiment, the parameterization schemes for cloud microphysics include Ferrier, Lin, WSM3, WSM5, WSM6, and Thompson;

[0081] The parameterization schemes for cumulus convection include KF, BMJ, and GD;

[0082] Parameterization schemes for boundary layers include YSU and MYNN;

[0083] Parametric schemes for land surface include the Noah land surface scheme;

[0084] Parameterization schemes for long-wave and short-wave radiation include RRTM and the Dudhia physical parameterization scheme.

[0085] In a specific embodiment, the mean absolute error (MAE), bias (BIAS), and relative error (RE) are used to evaluate the actual operating performance of each combination of schemes, using the following formula:

[0086]

[0087] BIAS=M i -O i ;

[0088]

[0089] Among them, M i This refers to actual precipitation at the monitoring station; O i The model calculates station precipitation, where i is the station number and N is the total number of stations.

[0090] Specifically, based on existing models, we selected and combined schemes with good simulation results and wide applicability. The cloud microphysics schemes selected were (Ferrier, Lin, WSM3, WSM5, WSM6, Thompson); the cumulus convection schemes were (KF, BMJ, GD); the boundary layer schemes were (YSU, MYNN); the land surface scheme was the Noah land surface scheme; and the longwave and shortwave radiation schemes used were RRTM and Dudhia physical parameterization schemes, respectively. These schemes were combined and arranged, and the mean absolute error (MAE), bias (BIAS), and relative error (RE) were used to evaluate the scheme performance. The smaller the MAE and RE values, the higher the model accuracy.

[0091] The three indicators are normalized using the following formula:

[0092] norm value= (original value - min(all values ​​of this indicator)) / (max(all values ​​of this indicator) - min(all values ​​of this indicator));

[0093] When processing, BIAS uses the absolute value |BIAS|, and assigns weights w to the three indicators. mae =0.5, w bias =0.3, w re =0.2;

[0094] Overall score = (norm mae *w mae )+(norm bias *w bias )+(norm re *w re );

[0095] The scheme with the lowest overall score is selected as the parameterization scheme with the best effect.

[0096] In one specific embodiment, the method also includes utilizing the MPI distributed computing framework to enable parallel computation of the numerical weather model, thereby improving computational efficiency. Specifically, the MPI library is installed, the WRF source code directory is navigated to the distributed memory parallelism option (dmpar), and the MPI wrapper compiler (mpif90 or mpicc) is specified. The MPI process performs finite difference calculations (such as advection terms, pressure gradient terms, Coriolis force terms, etc.) on the non-hydrostatic compressible Euler equations and at each time step and on the grid points within the subdomain. This part of the calculation is completely independent for each subdomain and is the main source of acceleration for parallel computation.

[0097] In one specific embodiment, the method also includes result verification and application, comparing the forecast accuracy before and after assimilation, and outputting forecast results for application in various scenarios. For example:

[0098] Precise meteorological disaster forecasting scenario: In the typhoon-prone areas of southeastern coastal China, a dense network of GNSS receiving stations is deployed to acquire water vapor tomography data; simultaneously, satellite radiation data (such as GPM and Himawari-8) and radar reflectivity data are integrated. The WRFDA module is used to assimilate the multi-source data into the initial field of the WRF model. Comparing the typhoon track forecast errors before and after assimilation, the results show that the 24-hour track forecast error is reduced by 15% after assimilation.

[0099] Aviation and Maritime Navigation Optimization Scenario: In an international aviation hub area, the numerical weather model WRF was deployed to assimilate GNSS water vapor tomography data, aircraft meteorological relay data (AMDAR), and ground meteorological station observation data to generate 1km resolution short-term nowcast products. Validation results show that the assimilated model improves the prediction accuracy of low-level wind shear and cumulus clouds by 20%, significantly optimizing flight scheduling decisions.

[0100] Agricultural meteorological service support scenario: For a plain agricultural area, a regional irrigation index model is constructed using assimilated high-precision humidity field data combined with soil moisture observations. In practical applications, the model-guided irrigation plan improves water resource utilization by 12% while reducing the risk of soil salinization caused by over-irrigation.

[0101] Climate model validation and improvement scenario: In a sparsely observed meteorological area on a plateau, a multi-source data assimilation experiment was conducted, combining GNSS water vapor tomography data, Fengyun-4 satellite radiation data, and radiosonde data. Results showed that the assimilated model significantly enhanced its precipitation forecasting capability for plateau low-vortex systems, with a 24-hour precipitation forecast TS score improvement of 0.25.

[0102] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0103] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for assimilating multi-source data from a numerical weather model based on GNSS water vapor tomography results, characterized in that, include: Deploy numerical weather models within the target area; Collect multi-source data of the target area, including GNSS water vapor tomography data, satellite data, radar data, and observation data from ground meteorological stations, radiosondes, and aircraft; Data assimilation is performed on each data item in the multi-source data to convert the multi-source data into a unified data format; The parameterization schemes for cloud microphysics, cumulus convection, boundary layer, land surface, and longwave and shortwave radiation processes in the numerical weather model are arranged and combined, and the actual operating effect of each combination of schemes is evaluated to select the optimal parameterization scheme.

2. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results as described in claim 1, characterized in that, The deployment of numerical weather models within the target area includes: environmental preparation, dependency library installation, numerical weather model installation, data preprocessing, and runtime configuration. The numerical weather model selected is the WRF mode.

3. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 2, characterized in that, The data preprocessing includes: Converting static geographic data into a WRF-readable binary format involves: modifying the configuration file of the geogrid program in the WRF preprocessing system to specify the data path and parameters; and generating a WRF-usable grid file based on the modified configuration file using the geogrid program in the WRF preprocessing system. Converting meteorological data into a WRF-readable binary format involves: linking the configuration file corresponding to the meteorological data and running the script file; linking the original meteorological data file; running the ungrib program to generate intermediate files; and then running the metgrid program to generate the final binary file.

4. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 3, characterized in that, The static geographic data includes: terrain elevation, land use type, and soil type data.

5. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 1, characterized in that, The step of assimilating each data item in the multi-source data and converting the multi-source data into a unified data format specifically includes: The GNSS water vapor tomography data is assimilated using an indirect method, specifically by converting the wet refractive index obtained from the tomography into relative humidity, and then using WRFDA to assimilate the GNSS water vapor tomography data in conjunction with existing temperature and pressure data. The satellite data was assimilated using CRTM or RTTOV radiative transfer models; The radar data is directly assimilated via WRFDA.

6. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 5, characterized in that, The specific formula for converting the wet refractive index into relative humidity is as follows: Among them, P w WR represents the water vapor pressure at the lattice point, T represents the wet refractive index, k1 = 77.6890 K / hpa, k2 = 71.2952 K / hpa; Among them, P s C represents the saturated water vapor pressure. i Let n represent T. i-1 express; RH=P w / P s ; Here, RH represents relative humidity.

7. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 1, characterized in that, The parameterization schemes for cloud microphysics include Ferrier, Lin, WSM3, WSM5, WSM6, and Thompson; The parameterization schemes for cumulus convection include KF, BMJ, and GD; The parameterization schemes for the boundary layer include YSU and MYNN; The parameterization scheme for the land surface includes the Noah land surface scheme; The parameterization schemes for long-wave and short-wave radiation include RRTM and the Dudhia physical parameterization scheme.

8. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 1, characterized in that, The actual operational effectiveness of each scheme combination is evaluated using the mean absolute error (MAE), bias (BIAS), and relative error (RE). The formula is as follows: BIAS = M i -THE i ; Among them, M i This refers to actual precipitation at the monitoring station; O i The model calculates station precipitation, where i is the station number and N is the total number of stations.

9. The method for assimilation of multi-source data from a numerical weather model based on GNSS water vapor tomography results according to claim 1, characterized in that, It also includes using the MPI distributed computing framework to enable parallel computation of numerical weather model models.

Images