Satellite-borne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation
By constructing a three-dimensional variational assimilation method with a dedicated observation operator and error model, GNSS-R data is deeply integrated with the GSI system, solving the accuracy problem of satellite observation under strong wind or cloud and rain conditions. This enables the generation of high-precision, spatially continuous sea surface wind speed fields, which is suitable for typhoon monitoring, climate research, and wind energy resource assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-03-27
AI Technical Summary
The accuracy of existing satellite observations of sea surface wind speed decreases under strong wind or cloud and rain conditions. GNSS-R data is difficult to effectively assimilate and utilize in the existing analysis system, resulting in insufficient accuracy and spatial continuity of wind speed fusion, which limits its application in complex sea conditions.
By employing a three-dimensional variational assimilation-based approach, a dedicated observation operator and error model are constructed. The GNSS-R sea surface wind speed product is deeply integrated with the GSI system. By minimizing the cost function, the optimal mapping between observation and model is achieved, generating a high-precision sea surface wind speed field with high spatial continuity.
It improves the spatial continuity and observation accuracy of sea surface wind speed fields, especially under strong winds and complex sea conditions. It is suitable for typhoon monitoring, climate research and ocean dynamic analysis, and supports numerical forecasting and wind energy resource assessment.
Smart Images
Figure CN121741784A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of marine remote sensing and numerical weather prediction technology, specifically to a spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation. Background Technology
[0002] Sea surface wind speed is a core parameter of air-sea interaction, and its accurate acquisition is crucial for typhoon monitoring, climate research, and ocean dynamic analysis, directly determining the reliability of related research and applications. Satellite remote sensing is currently the mainstream observation method, with scatterometers and radiometers being the most widely used. However, these technologies are easily interfered with under strong winds or cloudy and rainy conditions, leading to a significant decrease in observation accuracy and failing to meet the high-precision requirements of complex sea conditions.
[0003] Spaceborne GNSS-R technology provides a new approach to sea surface wind speed observation. It retrieves wind speed by analyzing the time delay-Doppler power waveform and other characteristics of GNSS sea surface reflected signals. This technology boasts advantages such as wide coverage, high time resolution, and immunity to cloud and rain effects, with a retrieval range exceeding 50 m / s, effectively supplementing traditional observation methods. However, integrating this advantageous data into existing analytical systems remains a pressing issue.
[0004] Three-dimensional variational assimilation (3DVar) technology achieves optimal fusion of multi-source data by minimizing the cost function of the observation and model background fields, and its application value is particularly prominent in wind speed fusion. However, existing research mainly focuses on scatterometer and radiometer data, and the error modeling and observation operator design for GNSS-R observation characteristics are still imperfect, which restricts its application.
[0005] This technical deficiency prevents GNSS-R from fully realizing its advantages, limiting its potential in wind speed fusion and making it difficult for existing fusion methods to overcome the limitations of traditional data. This hinders the improvement of the accuracy and spatial continuity of the wind speed analysis field and makes it difficult to support related research and applications under extreme sea conditions. Summary of the Invention
[0006] The present invention aims to provide a spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation, in order to solve the problems of large observation error, small coverage and spatial discontinuity of existing satellite observations of sea surface wind speed in strong wind sea areas, as well as the difficulty in effectively assimilating and utilizing GNSS-R sea surface wind speed data.
[0007] To achieve the above objectives, the present invention employs the following technical solution: a spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation, comprising the following steps: S1. The variable information of the acquired GNSS-R inverted second-level sea surface wind speed product is converted into a format and a BUFR file is generated as assimilated observation data. S2, acquire preprocessed PREPBUFR format ground and upper-air meteorological observation data, and assist in constraining the background field of the model; S3 processes geographic data and GFS meteorological element fields through WPS, and generates background field files through WRF initialization and model integration. S4 uses the 3DVar method to set the background error covariance matrix and the observation error covariance matrix, designs an observation operator to realize the mapping between observation and model space, and minimizes the cost function to obtain the optimal analysis field; S5 runs the GSI program to complete assimilation and fusion, outputting a gridded sea surface wind speed analysis field for subsequent application scenarios.
[0008] The principles and advantages of this scheme are: In existing technologies, traditional scatterometer and radiometer data are directly connected to the assimilation system without considering the unique observation mechanism of GNSS-R data, resulting in the inability to realize the value of the data.
[0009] This solution starts from the physical characteristics of sea surface wind speed in GNSS-R and constructs a dedicated observation operator to ensure that GNSS-R observation information can be accurately converted into quantified data that can be identified and fused by the assimilation system. At the same time, it conducts targeted error modeling based on the observation error distribution pattern of GNSS-R data to avoid the adaptation deviation of traditional general error models to GNSS-R data.
[0010] Building upon this foundation, the GNSS-R sea surface wind speed product, after undergoing characteristic adaptation processing, is deeply integrated with the GSI three-dimensional variational assimilation algorithm. This achieves, for the first time, precise adaptation between GNSS-R data and the GSI three-dimensional variational assimilation system. This design requires a deep understanding of the underlying logic of GNSS-R observation principles and the variational assimilation algorithm, overcoming several technical challenges such as "quantitative conversion of observation features" and "accurate fitting of error distribution." Furthermore, by fully leveraging the advantages of GNSS-R's wide coverage and strong anti-interference capabilities, combined with the multi-source data fusion capabilities of the GSI algorithm, optimal integration of multi-source information is achieved, ultimately outputting a high-precision, highly spatially continuous sea surface wind speed field. This effectively improves the spatial continuity and observation accuracy of the sea surface wind speed field, especially under extreme sea conditions such as strong winds and clouds / rain, making it suitable for strong winds and complex sea conditions, and providing a new technical approach for marine wind field observation and utilization.
[0011] Furthermore, this solution adopts a modular design approach, with a clear methodological structure and strong compatibility. It can be directly coupled with existing meteorological model systems without requiring large-scale reconstruction of existing systems, thus solving the "compatibility barrier" problem in the application of new technologies and possessing broad application potential. Attached Figure Description
[0012] Figure 1This is a flowchart illustrating the spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation of the present invention. Figure 2 This is a map showing the distribution of sea surface wind speed data from the Tianmu-1 constellation GNSS-R at 12:00 (UTC) ±3 hours on September 23, 2025. Figure 3 This is the sea surface wind speed analysis field after the assimilation experiment at 12:00 (UTC) on September 23, 2025. Detailed Implementation
[0013] The following detailed description illustrates the specific implementation method: The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation in this embodiment fuses the data with the numerical model background field under the GSI (Gridpoint Statistical Interpolation) three-dimensional variational assimilation framework by error modeling and observation operator construction targeting the characteristics of GNSS-R data. This generates a high-precision, spatially continuous sea surface wind speed analysis product, thereby improving the accuracy and spatiotemporal consistency of the sea surface wind speed product and supporting marine meteorological analysis and numerical forecasting business applications.
[0014] Specifically, this involves a sea surface wind speed fusion method based on three-dimensional variational assimilation (3DVar) for spaceborne GNSS-R (Global Navigation Satellite System Reflectometry), as shown in the attached figure. Figure 1 As shown, it includes the following steps: S1. The variable information of the acquired GNSS-R inverted second-level sea surface wind speed product is converted into a format and a BUFR file is generated as assimilated observation data.
[0015] In this embodiment, the Level 2 (L2) sea surface wind speed product retrieved from the Tianmu-1 occultation meteorological observation constellation GNSS-R includes two types of data: global sea surface wind speed and cyclonic sea surface wind speed, with a corresponding altitude of 10 meters. Its variable information includes sea surface wind speed data and its quality label, latitude and longitude, time, etc. The sea surface wind speed data and its quality label, latitude and longitude, time, and other variable information are converted into a format to generate a BUFR (Binary Universal Form for the Representation of meteorological data) file that meets the input requirements of the GSI system, which will be used as the satellite observation data for subsequent assimilation.
[0016] S2, acquire preprocessed PREPBUFR format ground and upper-air meteorological observation data, and assist in constraining the background field of the model.
[0017] In this embodiment, surface and upper-air meteorological observation data preprocessed by the Global Data Assimilation System (GDAS) of the National Center for Environmental Prediction (NCEP) are acquired. The data format is PREPBUFR (Preprocessed BUFR), and the observation data consists of multi-source conventional observation data, including but not limited to conventional observation data from ground stations, ships, buoys, radiosondes, and aircraft. This data is used to assist in constraining the background field of the model and enhance the multi-source consistency of the assimilation system.
[0018] S3 processes geographic data and GFS meteorological element fields through WPS, and generates background field files through WRF initialization and model integration.
[0019] In this embodiment, the preparation of the pattern background field file mainly includes the following sub-steps: S3.1 uses the geogrid.exe subroutine of the WRF Preprocessing System (WPS) to define the assimilation region and interpolate static geographic fields.
[0020] S3.2, obtain the multi-layer meteorological element field provided by the Global Forecast System (GFS), and use the WPS ungrib.exe subroutine to reproject the data and extract the required meteorological parameters.
[0021] The spatial resolution of the multi-layer meteorological element field is 0.25°, with a time interval of 6 hours. The multi-layer meteorological element field includes temperature, humidity, wind speed, wind direction, and air pressure.
[0022] S3.3 uses the metgrid.exe subroutine of WPS to perform spatiotemporal interpolation on surface parameters and meteorological data, interpolating meteorological parameters into the assimilation region.
[0023] S3.4 Run the WRF initialization program (real.exe) to perform vertical interpolation and physical consistency checks, and generate the initial field and boundary conditions.
[0024] S3.5 Execute the WRF mode main program (wrf.exe), complete the numerical integration within the given time window, and output the wrfout file as the background field input data for the GSI assimilation system.
[0025] S4 employs the 3DVar method, sets the background error covariance matrix and the observation error covariance matrix, designs an observation operator to realize the mapping between observation and model space, and minimizes the cost function to obtain the optimal analysis field.
[0026] In this embodiment, within the GSI assimilation framework, the assimilation mode is set to three-dimensional variational assimilation (3DVar), and the background error covariance matrix B and the observation error covariance matrix R are defined. B describes the spatial correlation and error propagation characteristics of sea surface wind speed; R is determined by statistical estimation of GNSS-R wind speed inversion error and observation noise.
[0027] In the GSI system, a GNSS-R wind speed observation operator H is introduced to realize the nonlinear mapping relationship between the observation space and the model space. The mapping relationship between the observation and model spaces is then expressed as follows: ; In the formula, For observation vectors; The optimal solution is yet to be found. This represents the observation error. Operator H performs spatial and temporal interpolation matching between observation points and model grid points to obtain the corresponding simulated observation values.
[0028] The objective of three-dimensional variational assimilation is to minimize the cost function. Therefore, the optimal analysis field is obtained by minimizing the cost function between the observation and the model background field. The minimized cost function is expressed as follows: ; In the formula, The optimal solution is yet to be found. For background scene; These are the observed values simulated using the analytical variable x; The background error covariance matrix; Let be the observation error covariance matrix. This is achieved by minimizing... To obtain the optimal solution analysis field This achieves optimal fusion of the background field and the observation field.
[0029] S5 runs the GSI program to complete assimilation and fusion, outputting a gridded sea surface wind speed analysis field for subsequent application scenarios.
[0030] This embodiment mainly includes the following sub-steps: S5.1, run the GSI executable program gsi.x to perform assimilation calculations, complete the three-dimensional variational fusion calculation of GNSS-R observation data, conventional observation data and model background field, and obtain the optimal sea surface wind speed analysis field.
[0031] S5.2 outputs a gridded sea surface wind speed distribution field, which can be directly used as the initial condition for subsequent WRF model integration, or used in applications such as numerical weather prediction, marine meteorological reanalysis, and wind energy resource assessment.
[0032] The following detailed description uses specific embodiments.
[0033] A certain sea area was selected as the assimilation experiment area, which is approximately 110°–135° east longitude and 10°–35° north latitude, and the time period is from 00:00 to 18:00 on September 23, 2025 (UTC).
[0034] (a) Obtain the GNSS-R inverted sea surface wind speed product for September 23, 2025, from the Tianmu-1 occultation meteorological observation constellation, as shown in the attached image. Figure 2 The sea surface wind speed data distribution map shown includes wind speed, latitude and longitude, time, and quality labeling information. The data underwent preprocessing, including removing observation points with substandard quality, filtering data within the assimilation area, and dividing it into data files for each of the last 6 hours. This data was then converted to the BUFR format supported by the GSI system. The resulting GNSS-R wind speed files are input into the GSI assimilation system as satellite observation data.
[0035] (b) Obtain routine observation data from the NCEP GDAS system in PREPBUFR format, including observations from ground stations, ships, buoys, radiosondes, and aircraft. This data undergoes uniform quality control and can be directly used for assimilation to enhance data constraint capabilities.
[0036] (c) Define the assimilation zone using the WRF preprocessing system (WPS); interpolate geographic data such as topography, land use and soil type into the assimilation zone; download meteorological data fields (temperature, humidity, wind speed, wind direction, air pressure, etc.) provided by the Global Forecast System (GFS) and interpolate them into the assimilation zone.
[0037] The initial field and boundary conditions are generated by performing vertical interpolation and consistency checks using the WRF initialization program (real.exe).
[0038] Run the WRF mode (wrf.exe) to complete the integration and output the background field wrfout file as the input for GSI assimilation.
[0039] (d) The system is configured using the three-dimensional variational assimilation (3DVar) method within the GSI framework.
[0040] This includes outputting the background field in WRF mode; setting the observation data to GNSS-R and conventional observation data; setting the assimilation window to 6 hours; analyzing variables such as U and V wind components and other meteorological elements; and setting the background error covariance and observation error covariance.
[0041] Then, a GNSS-R wind speed observation operator is constructed to achieve spatial interpolation matching between observation points and model grid points. The optimal analysis field is obtained by minimizing the cost function according to equation (2).
[0042] (e) Input the prepared data into the GSI system and perform three-dimensional variational assimilation calculations to obtain the sea surface wind speed analysis field for the region of 110°–135°E and 10°–35°N. During the assimilation calculation process, the GSI system will gradually adjust the values of the analysis field according to the defined cost function and relevant parameter settings through an iterative optimization algorithm, so as to minimize the cost function while satisfying the constraints of the background field and the observation field, and finally obtain the optimal sea surface wind speed analysis field, as shown in the appendix. Figure 3 The sea surface wind speed analysis field after the assimilation experiment is shown.
[0043] The obtained sea surface wind speed analysis field was applied, and the results are as follows: Application 1: Used for numerical weather prediction, the generated sea surface wind speed analysis field is input into the WRF model as initial conditions to simulate typhoon path prediction.
[0044] Application 2: Used for marine meteorological reanalysis. The sea surface wind speed analysis field was applied to the marine meteorological reanalysis of the region between 110° and 135° east longitude and 10° and 35° north latitude. The generated reanalysis dataset is superior to the dataset generated by traditional methods in terms of spatial resolution and temporal continuity, providing higher quality data support for climate research.
[0045] Application 3: Wind energy resource assessment. This sea surface wind speed analysis field can also be used for wind energy resource assessment in the region, providing a scientific basis for wind energy development and utilization, and helping to optimize the site selection and layout of wind energy projects.
[0046] In this embodiment, addressing the common shortcoming of existing technologies that lack a dedicated GNSS-R observation mechanism, a customized observation operator is first designed based on the reflection signal characteristics of GNSS-R sea surface wind speed products. Simultaneously, targeted error modeling is conducted by combining the observation error distribution patterns of GNSS-R data, improving the adaptability of the variational assimilation system in sea surface wind speed fusion. Then, fully utilizing the observation coverage advantage of spaceborne GNSS-R sea surface wind speed products, and combining the GSI three-dimensional variational assimilation algorithm, the inherent core mechanism of "minimizing the cost function between observation and model background fields" effectively improves the spatial continuity and observation accuracy of the sea surface wind speed field, generating a high-precision, spatially continuous sea surface wind speed analysis product.
[0047] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.
Claims
1. A spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation, characterized in that, Includes the following steps: S1. The variable information of the acquired GNSS-R inverted second-level sea surface wind speed product is converted into a format and a BUFR file is generated as assimilated observation data. S2, acquire preprocessed PREPBUFR format ground and upper-air meteorological observation data, and assist in constraining the background field of the model; S3 processes geographic data and GFS meteorological element fields through WPS, and generates background field files through WRF initialization and model integration. S4 uses the 3DVar method to set the background error covariance matrix and the observation error covariance matrix, designs an observation operator to realize the mapping between observation and model space, and minimizes the cost function to obtain the optimal analysis field; S5 runs the GSI program to complete assimilation and fusion, outputting a gridded sea surface wind speed analysis field for subsequent application scenarios.
2. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that: In S4, the GNSS-R wind speed observation operator H is introduced into the GSI system, and the mapping relationship between the observation and model space is expressed as follows: ； In the formula, Let x be the observation vector; x is the optimal solution to be found. This represents the observation error.
3. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 2, characterized in that: In S4, the optimal analysis field is obtained by minimizing the cost function between the observations and the model background field. The minimized cost function is expressed as: ； In the formula, x is the optimal solution to be found; For background scene; For observation vectors; For observation operators; These are the observed values simulated using the analytical variable x; The background error covariance matrix; Let be the observation error covariance matrix.
4. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that: In S1, the sea surface wind speed products acquired include two types of data: global sea surface wind speed and cyclone sea surface wind speed, with a corresponding altitude of 10 meters.
5. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that, S3 also includes the following sub-steps: S3.1 uses the geogrid.exe subroutine of the WRF preprocessing system to define the assimilation region and interpolate static geographic fields; S3.2, obtain the multi-layer meteorological element field provided by the Global Forecast System (GFS), and use the WPS ungrib.exe subroutine to reproject the data and extract the required meteorological parameters; S3.3 uses the metgrid.exe subroutine of WPS to perform spatiotemporal interpolation on surface parameters and meteorological data, interpolating meteorological parameters into the assimilation region; S3.4, Run the WRF initialization program, perform vertical interpolation and physical consistency checks, and generate the initial field and boundary conditions; S3.5 Executes the WRF mode main program, completes numerical integration within a given time window, and outputs the wrfout file as background field input data for the GSI assimilation system.
6. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that, S5 also includes the following sub-steps: S5.1, run the GSI executable program gsi.x to complete the three-dimensional variational fusion calculation of GNSS-R observation data, conventional observation data and model background field, and obtain the optimal sea surface wind speed analysis field; S5.2 outputs a gridded sea surface wind speed distribution field.
7. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 5, characterized in that: In S3.2, the spatial resolution of the multi-layer meteorological element field is 0.25°, and the time interval is 6 hours.
8. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 5, characterized in that: The multi-layered meteorological field includes temperature, humidity, wind speed, wind direction, and air pressure.
9. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that: In S1, the variable information includes sea surface wind speed data and its mass label, latitude and longitude, and time.
10. The spaceborne GNSS-R sea surface wind speed fusion method based on three-dimensional variational assimilation according to claim 1, characterized in that: In S2, the observation data consists of multi-source conventional observation data, including but not limited to ground stations, ships, buoys, radiosondes, and aircraft.